IEEE CS 2022 Report

Hasan Alkhatib, Paolo Faraboschi, Eitan Frachtenberg, Hironori Kasahara, Danny Lange, Phil Laplante, Arif Merchant, Dejan Milojicic, and Karsten Schwan with contributions by: Mohammed AlQuraishi, Angela Burgess, David Forsyth, Hiroyasu Iwata, Rick McGeer, and John Walz Preface

In 2013-14, nine technical leaders wrote a report, entitled complete and exhaustive, it is inevitable that some technolo- IEEE CS 2022, surveying 23 innovative technologies that gies have been omitted, such as Bitcoin, future transportation, could change the industry by the year 2022. The report and the general notion of what technology contributes to the covers security cross-cutting issues, open intellectual prop- mankind. Our position, as well as the premise that this docu- erty movement, sustainability, massively online open courses, ment brings, is that technology is the enabler. What humanity quantum computing, device and nanotechnology, device takes out of it really depends on human society. and nanotechnology, 3D integrated circuits, multicore, pho- The IEEE CS 2022 report was presented at the Computer So- tonics, universal memory, networking and interconnectivity, ciety of India Congress, at the Information Processing Society software-defined networks, high-performance computing, of Japan (IPSJ) Congress, at the IEEE CS Board of Governors, cloud computing, the Internet of Things, natural user interfac- at the IEEE CS Industrial Advisory Board, and at Belgrade es, 3D printing, big data and analytics, machine learning and Chapter. We received positive feedback and excellent ideas intelligent systems, computer vision and pattern recognition, for improvement. This is a living document, because the tech- life sciences, computational biology and bioinformatics, and nology continuously changes. We intend to use this document robotics for medical care. for the IEEE CS strategic planning that takes place every These technologies, tied into a scenario that we call seam- three years. We hope that the IEEE CS will be able to come up less intelligence, present a view of the future. For each of the with similar reports regularly in the future. 23 technologies, there is a description of the state of the art, I thank Hasan Alkhatib, Paolo Faraboschi, Eitan Frachtenberg, challenges, where we think the technology will go, and its Hironori Kasahara, Danny Lange, Phil Laplante, Arif Merchant, disruption. To confirm the report’s prediction, we surveyed and Karsten Schwan for making this journey to 2022 together. IEEE members about technology drivers and disruptors. We Without their vision, technical knowledge, and creativity, this also tried to predict what kind of society the world would document would not be possible. require with these 23 technologies. Finally, we analyzed the IEEE digital library to better understand the degree to which these technologies are covered today and by which Societies, so that we can make better ties. This document is intended for computer science professionals, students, and professors, as well as laymen interested Dejan S Milojicic, in technology and technology use. While we tried to be IEEE Computer Society President 2014, February 2014

2 CONTENTS

3 FIGURES

Figure 1. Landscape of 23 technologies...... 7 Figure 18. Coverage of some of the top disruptors Figure 31. The breakdown of networking and in IEEE Libraries by individual societies...... 118 interconnectivity by sponsoring societies...... 149 Figure 2. IT ecosystem from supply to demand...... 21 Figure 19. The breakdown of 22 technologies by Figure 32. The breakdown of software-defined Figure 3. Two integration scenarios: a 2.5D component using periodical articles...... 137 networks by sponsoring societies...... 150 a silicon interposer and a full 3D stack using TSVs...... 34 Figure 20. The breakdown of 22 technologies by Figure 33. The breakdown of HPC by sponsoring Figure 4. An integration scenario combining 2.5D integration sponsoring societies...... 138 societies...... 151 of multiple 3D-stacked components...... 35 Figure 21. The breakdown of security cross-cutting Figure 34. The breakdown of cloud computing by Figure 5. Illustration of the severity of the DRAM issues by sponsoring societies...... 139 sponsoring societies...... 152 capacitor “trench.”...... 38 Figure 22. The breakdown of open intellectual property Figure 35. The breakdown of IoT by sponsoring Figure 6. Simplified phase-change memory (PCM) cell by sponsoring societies...... 140 societies...... 153 and spin-transfer torque (STT) cell...... 39 Figure 23. The breakdown of sustainability by Figure 36. The breakdown of natural user interfaces Figure 7. Simplified memristor (ReRAM) cell...... 40 sponsoring societies...... 141 by sponsoring societies...... 154 Figure 8. Multicore. many-cores landscape...... 44 Figure 24. The breakdown of MOOC by sponsoring Figure 37. The breakdown of 3D printing by Figure 10. Rule-of-thumb of using photonics vs. electronics societies...... 142 sponsoring societies...... 155 based on distance and required bandwidth...... 48 Figure 25. The breakdown of quantum computing Figure 38. The breakdown of big data analytics Figure 11. Roadmap of industrial photonics by sponsoring societies...... 143 by sponsoring societies...... 156 technologies...... 50 Figure 26. The breakdown of device and nano-technology Figure 39. The breakdown of machine learning and Figure 12. A representative switch ASIC pipeline...... 58 by sponsoring societies...... 144 intelligent systems by sponsoring societies...... 157 Figure 13. High-performance computing...... 64 Figure 27. The breakdown of 3D integrated circuits Figure 40. The breakdown of computer vision and Figure 14. Comparing classes of HPC and their by sponsoring societies...... 145 pattern analysis by sponsoring societies...... 158 feasibility to deliver in the cloud...... 65 Figure 28. The breakdown of universal memory Figure 41. The breakdown of life sciences by Figure 15. Comparison of major drivers...... 113 by sponsoring societies...... 146 sponsoring societies...... 159 Figure 16. Comparison of major disruptors...... 114 Figure 29. The breakdown of muticore by Figure 42. The breakdown of computational biology sponsoring societies...... 147 by sponsoring societies...... 160 Figure 17. Coverage of some of the top drivers in IEEE Libraries by individual societies...... 117 Figure 30. The breakdown of photonics by Figure 43. The breakdown of robotics in medicine sponsoring societies...... 148 by sponsoring societies...... 161 TABLES

Table 1. IEEE SWOT...... 122 Table 3. Search keywords summary...... 135 Table 2. Breakdown of who and how Table 4. Google and IEEE Xplore search will be benefiting from IEEE CS...... 123 results combined...... 162

4 CONTENTS 1. INTRODUCTION...... 5 1.1 Goals...... 6 1. Introduction 1.2 Target Audience...... 6 1.3 Process...... 6 1.4 Technologies Landscape...... 7 1.5 Document Organization...... 7 Predicting the future is hard and risky. Predicting to better understand where technologies are going. 2. SEAMLESS INTELLIGENCE the future in the computer industry is even harder The book Innovators Dilemma and its sequels best CONTENTSSCENARIO...... 8 and riskier due to dramatic changes in technology describe the process of innovation and disruption. 3. 23 TECHNOLOGIES IN 2022...... 14 and limitless challenges to innovation. Only a small Nine technical leaders of the IEEE Computer 4. DRIVERS AND DISRUPTORS...... 112 fraction of innovations truly disrupt the state of the Society joined forces to write a technical report, 5. TECHNOLOGY COVERAGE IN art. Some are not practical or cost-effective, some entitled IEEE CS 2022, symbolically surveying IEEE XPLORE AND BY are ahead of their time, and some simply do not 23 potential technologies that could change the IEEE SOCIETIES...... 115 have a market. There are numerous examples of landscape of computer science and industry by the 6. IEEE COMPUTER SOCIETY superior technologies that were never adopted be- IN 2022...... 120 year 2022. In particular, this report focuses on 3D cause others arrived on time or fared better in the printing, big data and analytics, open intellectual 7. SUMMARY AND NEXT STEPS.....124 market. Therefore this document is only an attempt property movement, massive- 8. AUTHORS...... 126 ly online open courses, security APPENDIX I. 23 Technologies Coverage cross-cutting issues, universal in IEEE Publications...... 134 memory, 3D integrated circuits, photonics, cloud computing, computational biology and bioinformatics, device and nanotechnology, sustainability, high-performance computing, the Internet of Things, life sciences, machine learning and intelligent systems, natural user Only a small interfaces, networking and inter- fraction of connectivity, quantum computing, software-defined networks, multi- truly innovations core, and robotics for medical care. disrupt the state of the art.

5 CONTENTS 1. INTRODUCTION...... 5 1.1 Goals contributes to mankind, a question frequently 1.1 Goals...... 6 As authors, we had the following goals in mind asked by those who have seen this material to 1.2 Target Audience...... 6 when we started writing the document: date. Our premise, echoed in this document, is that technology is the enabler. What humanity takes out 1.3 Process...... 6 • Predict the future technologies that will disrupt of it really depends on human society. 1.4 Technologies Landscape...... 7 the state of the art. 1.5 Document Organization...... 7 • Help researchers understand the future impact 1.3 Process 2. SEAMLESS INTELLIGENCE of various technologies. CONTENTSSCENARIO...... 8 • Help laymen—a general audience—understand The core team of nine technologists met twice by 3. 23 TECHNOLOGIES IN 2022...... 14 where technology is evolving and the implica- phone in preparation for a face-to-face meeting in Seattle, collocated with an IEEE Board of Gover- 4. DRIVERS AND DISRUPTORS...... 112 tions for human society. nors gathering. We brainstormed about possible 5. TECHNOLOGY COVERAGE IN • Help the IEEE Computer Society understand IEEE XPLORE AND BY how it should be organized for this future. technologies and came up with a list that has since IEEE SOCIETIES...... 115 been trimmed. Each team member chose two to 6. IEEE COMPUTER SOCIETY 1.2 Target Audience three technology areas to describe, and two mem- IN 2022...... 120 This document was intended for computer science bers wrote the scenario. 7. SUMMARY AND NEXT STEPS.....124 professionals, students, and professors, as well as We describe each of the 23 technologies by follow- 8. AUTHORS...... 126 laymen interested in technology and technology ing a common approach—summary of the state of APPENDIX I. 23 Technologies Coverage use. It is equally targeted to the members of the the art, challenges, where we think the technology in IEEE Publications...... 134 Computer Society and similar Societies around the will go, and its disruption—and tie them into a sce- world, as we dare to predict what kind of future nario that we call seamless intelligence. Together, professional society will be best suited to take they present a similar view of the future. these technologies to the next level through its We held another face-to-face meeting in the IEEE publications, conferences, communities, standards, Computer Society’s Washington, DC, office to courses, and artifacts in support of our profession brainstorm the future of the IEEE Computer Soci- and humanity. ety. Ultimately, we are attempting to predict what While we tried to be complete and exhaustive, it kind of future Society will be needed for our pro- Only a small is inevitable that some technologies and aspects fession, for the professionals who will be learning, have been omitted. Examples include electron- fraction of practicing, and putting into use the technologies ic money, such as Bitcoin, and various forms of we present here. innovations truly transportation, such as autonomous vehicles. Also Most of our other interaction was by email. In a few disrupt the state missing is the general notion of what technology of the art. cases, we reverted to technologists outside of our

6 IEEE CS 2022 , Report DRAFT Introduction 1/26/2014 5:18 PM

Life Sciences (21) Computational Biology Medical Robotics (23) Market Category and Bioinformatics (22)

Computer Vision and Pattern Recognition (20) CONTENTS Machine Learning and Intelligent Systems (19) Big Data and Analytics (18) team who helped us write 1. INTRODUCTION...... 5 Technologies Natural User Interfaces (16) 3D Printing (17) on the topics of life scienc- 1.1 Goals...... 6 High-Performance Computing (13) Cloud Computing (14) Internet of Things (15) 1.2 Target Audience...... 6 es, bioinformatics, robotics, and software-defined Networking & Interconnectivity (11) Software-Defined Networks (12) 1.3 Process...... 6 networks. 3D Integrated Circuits (7) Multicore (8) Photonics (9) Universal Memory (10) 1.4 Technologies Landscape...... 7 Quantum Computing (5) Device and Nanotechnology (6) 1.5 Document Organization...... 7 Independently, we surveyed a few thousand IEEE Human Capital Massively Online Open Courses (4) 2. SEAMLESS INTELLIGENCE members on technology CONTENTSSCENARIO...... 8 Open Intellectual Property Movement (2) Sustainability (3) drivers and disruptors, and Policies 3. 23 TECHNOLOGIES IN 2022...... 14 Security Cross-Cutting Issues (1) they confirmed some of our 4. DRIVERS AND DISRUPTORS...... 112 Figure 1. Landscape of 23 technologies. predictions and provided Figure 1 . Landscape 23 of technologies. (The numbers (The numbers after the after technology the represent technology the represent subsection the in subsection Section in Section 3.) 3.) 5. TECHNOLOGY COVERAGE IN another perspective on IEEE XPLORE AND BY IEEE SOCIETIES...... 115 the future of technology 1.5 Document Organizationrevisit it in future attempts to categorize technolo- The rest of this document is organized as follows. Section 2 describes the seamless intelligence scenario 6. IEEE COMPUTER SOCIETY advancements. that ties 23 the technology areas gy areas. together wcases and sho their potential benefits. It also serves as a use IN 2022...... 120 Finally, we were helped by IEEEcase Computer introduction Society for the individual technology areas presented in Section 3. Technology drivers and 7. SUMMARY AND NEXT STEPS.....124 staff for copyediting, pictures, anddisruptors numerous are other presented 1.5 in 4 Section , based Document on the survey Organization we conducted. Strengths, weaknesses, 8. AUTHORS...... 126 details. opportunities, and threats to IEEE are presented 5 in Section . They serve the purpose of better understanding how the The IEEE rest Computer of this Society document should is organized grow h in the future, whic is the as follows. theme of Section APPENDIX I. 23 Technologies Coverage 6. Summary and next Section steps are 2 describes provided 7 in Section . The the authors seamless as well intelligence as other contributors are in IEEE Publications...... 134 1.4 Technologies Landscapepresented in Section 8 . scenario that ties the 23 technology areas togeth- When we originally discussed the 23 technologies, er and showcases their potential benefits. It also we observed them all as equal. However, some of serves as a use case introduction for the individual the feedback we received from the IEEE Comput- technology areas presented in Section 3. Technol- er Society Industrial Advisory Board was that our ogy drivers and disruptors are presented in Section 23 technologies really fit into a larger landscape, 4, based on the survey we conducted. Strengths, comprising policies, human capital, technologies, weaknesses, opportunities, and threats to IEEE are and market categories. The image below is an at- presented in Section 5. They serve the purpose of Only a small tempt to classify the technology areas according to better understanding how the IEEE Computer Soci- offered classification. We could have evolved this ety should grow in the future, which is the theme of fraction of model further and tried to populate it with other Section 6. Summary and next steps are provided in innovations truly elements, but we felt that our bottom-up approach Section 7. The authors as well as other contributors disrupt the state was sufficient for this document’s needs. We may are presented7 in Section 8. of the art.

7 CONTENTS 1. INTRODUCTION...... 5 2. SEAMLESS INTELLIGENCE SCENARIO...... 8 2. Seamless 2.1 Introduction...... 8 2.2 State of the Art...... 8 Intelligence 2.3 Challenges and Opportunities..... 9 2.4 What Will Likely Happen...... 12 CONTENTS2.5 Potential Disruptions...... 12 Scenario 2.6 Summary...... 13 3. 23 TECHNOLOGIES IN 2022...... 14 4. DRIVERS AND DISRUPTORS...... 112 5. TECHNOLOGY COVERAGE IN IEEE XPLORE AND BY 2.1 Introduction phase where intelligence becomes seamless and IEEE SOCIETIES...... 115 Since the inception of digital computing in the mid- ubiquitous to those who can afford and use state- 6. IEEE COMPUTER SOCIETY 1940s, society has witnessed a historic revolution in of-the-art information technology. IN 2022...... 120 the acquisition, processing, and communication of This new reality is the expected result of the 7. SUMMARY AND NEXT STEPS.....124 information. This revolution has transformed every confluence of multiple information and commu- 8. AUTHORS...... 126 aspect of society through increased automation, nication technologies. Computing devices—from APPENDIX I. 23 Technologies Coverage ubiquitous access to information, and pervasive the very small, such as wearable devices and in IEEE Publications...... 134 human networking. chips embedded under the skin, to the computers inside our mobile devices, laptops, desktops, 2.2 State of the Art home servers, TV sets, and refrigerators, to the computing cloud that we reach via the Internet— We continue to witness an increase in the numbers, are interconnected via different communication shapes, and sizes of computing devices, from mi- and networking technologies. Together, they cro-scale to mega-scale, as well as a combinatorial form an intelligent mesh, a computing and com- increase in connectivity, both local and global. As We are in a phase munication ecosystem that augments reality with a result of this pervasive penetration of computing information and intelligence gathered from our of seamless and communication capabilities, human knowl- fingertips, eyes, ears, and other senses, and even and ubiquitous edge, intelligence, and connectivity are increasingly directly interfaced to our brain waves. intelligence. enhanced and augmented by information technology. By 2022, we project that we will be well into a

8 CONTENTS 1. INTRODUCTION...... 5 2.3 Challenges and 2. SEAMLESS INTELLIGENCE Opportunities SCENARIO...... 8 At the heart of this revolution is seamless 2.1 Introduction...... 8 networking, where the transition from one 2.2 State of the Art...... 8 network device to another is transparent and 2.3 Challenges and Opportunities..... 9 uninterrupted. Various wireless networking 2.4 What Will Likely Happen...... 12 technologies—from Near-Field Communi- CONTENTS2.5 Potential Disruptions...... 12 cation (NFC), to Bluetooth, to Wi-Fi, 4G, and 2.6 Summary...... 13 5G—are integrated with high-speed wired networking and the Internet, allowing any- 3. 23 TECHNOLOGIES IN 2022...... 14 where-to-anywhere access. But to achieve will require notification of peers about their mutual 4. DRIVERS AND DISRUPTORS...... 112 seamlessness and realize logical end-to-end presence. But to achieve interoperability, identity 5. TECHNOLOGY COVERAGE IN connectivity, we will need communications to run federation will require standards developed by and IEEE XPLORE AND BY agreed upon among identity providers. In addition, IEEE SOCIETIES...... 115 independently on top of any form of physical networking, regardless of device or location. Through meta-identity information will play a major role, 6. IEEE COMPUTER SOCIETY IN 2022...... 120 virtualized end-to-end connectivity, total integra- capturing a person’s profile and managing preferences while, for example, shopping, eating, and 7. SUMMARY AND NEXT STEPS.....124 tion of all the ecosystem devices that cater to our specific needs can be achieved. This new world traveling (specifically, a hotel could detect a guest’s 8. AUTHORS...... 126 will require sophisticated intelligent coordination preferred type of bed, floor level, or smoking status APPENDIX I. 23 Technologies Coverage software; voice, image, and motion recognition and automatically fulfill a reservation accordingly). in IEEE Publications...... 134 will transform human–computer interfaces into a Cloud services that offer APIs to facilitate appli- seamless interaction between the user and all the cation mash-ups will lead to intelligent software computing devices in that person’s life. that can integrate multiple services together and Another gap between today and 2022 is seamless achieve results that are difficult to imagine today. reliance on federated identity and the use of more We see the current power in mashing up location sophisticated identity technologies. Access will be data with maps as an illustration of what future We are in a phase authorized based on capabilities and access tokens mash-ups might look like. rather than strictly on identity. Private applications of seamless The combination of powerful voice and facial rec- will still require strict identity–for example, discov- ognition, massive identity databases, and powerful and ubiquitous ering from a specific social network that a specific tracking will likely result in a new norm that po- intelligence. friend happens to be at the same café as the user tentially translates into a significant loss of privacy

9 CONTENTS 1. INTRODUCTION...... 5 compared to today. Technology will enable many teaching and learning, augmented by automated 2. SEAMLESS INTELLIGENCE benefits, but controlling its use and preventing mis- and interactive learning outside the classroom as SCENARIO...... 8 use will require collective social action. well as through distance participation. By 2022, we 2.1 Introduction...... 8 On the other hand, pervasive and massive identity expect that the experiments with MOOCs will lead to a refined model in which they become comple- 2.2 State of the Art...... 8 recognition could also result in myriad benefits, mentary to ongoing instruction models. We also 2.3 Challenges and Opportunities..... 9 such as cashless and contactless financial transac- project that the classroom will involve less instruc- 2.4 What Will Likely Happen...... 12 tions, the ability to cross borders without stopping for inspection, and walking into a coffee shop in a tion and more dialogues with the expert professor, CONTENTS2.5 Potential Disruptions...... 12 resulting from the ability to use technology to learn 2.6 Summary...... 13 foreign country and having the barista offer up your favorite coffee because your preferences appeared outside the classroom. Students will enjoy learn- 3. 23 TECHNOLOGIES IN 2022...... 14 on her counter screen as you approached the shop. ing more, requiring less time and gaining deeper 4. DRIVERS AND DISRUPTORS...... 112 comprehension of their subject material. While 5. TECHNOLOGY COVERAGE IN The application of seamless and pervasive intel- MOOCs will become part of the education ecosys- IEEE XPLORE AND BY ligence will penetrate many aspects of our lives, tem, making them effective will be a challenge. The IEEE SOCIETIES...... 115 particularly healthcare. Imagine walking into a future holds even more for the integration of work 6. IEEE COMPUTER SOCIETY hospital and having your entire medical history be and education via augmented reality. As someone IN 2022...... 120 accessible to the attending medical professional is working, for example, she will get customized 7. SUMMARY AND NEXT STEPS.....124 from a centrally managed health vault: you won’t information that progressively trains her. This will 8. AUTHORS...... 126 need to state what medications you are current- revolutionize several sectors, including customer APPENDIX I. 23 Technologies Coverage ly taking or what immunization your child most care and the learning of services and products in IEEE Publications...... 134 recently received. Progress in 3D printing already [Saracco]. lets your dentist automatically shape your crown molding while you wait. Physicians will also be able Progress in robotics will likely transform the way to use less invasive procedures, such as having a mass transit is handled today to fully automated, patient swallow a small camera to track the entire autonomous vehicles. Imagine a driverless taxi, digestive track without needing to perform an in- just large enough to accommodate you and your cision; medication and medical devices could even baggage, dispatched to your hotel to take you to the airport, automatically navigating the best route be customized on the fly. We are in a phase along the way. Naturally, it already knows your of seamless Seamless and pervasive intelligence is impacting departure terminal from a prior seamless informa- and ubiquitous education more disruptively. The traditional model tion exchange. Autonomous vehicles will transform of campus-based education is changing by vir- intelligence. the topology of urban areas, dynamically creating tue of the availability of better methods for both one-way streets and preferential lanes. The traffic

10 CONTENTS 1. INTRODUCTION...... 5 layout will change continuously. This might also 2. SEAMLESS INTELLIGENCE lead to a change in the concept of car ownership, SCENARIO...... 8 transforming vehicles into utilities to use and drop 2.1 Introduction...... 8 [Saracco]. 2.2 State of the Art...... 8 Continuity in computing—from basic sensory 2.3 Challenges and Opportunities..... 9 processing, to simple event and location tracking, 2.4 What Will Likely Happen...... 12 to calendaring and collaboration support, to per- CONTENTS2.5 Potential Disruptions...... 12 sonal applications—will be augmented by powerful 2.6 Summary...... 13 computing in the cloud and massively distributed systems. Big data analysis will take place in the 3. 23 TECHNOLOGIES IN 2022...... 14 background, providing continuous intelligence to 4. DRIVERS AND DISRUPTORS...... 112 executives who run major organizations, enabling enforcing the strong security measures that can 5. TECHNOLOGY COVERAGE IN both the tracking and coordination of major busi- achieve unprecedented levels of safety in the ser- IEEE XPLORE AND BY vice of peace. Smart sensors, surveillance cameras, IEEE SOCIETIES...... 115 ness activities and intelligent choices based on real-life data intelligence. and eavesdropping devices integrated with identity 6. IEEE COMPUTER SOCIETY recognition systems will allow law enforcement to IN 2022...... 120 Developments in cloud computing will transition track and capture or quarantine individuals who 7. SUMMARY AND NEXT STEPS.....124 computing from a physical experience to a virtual might otherwise cause harm to others in society. one available to any user via a simple device oper- 8. AUTHORS...... 126 Conversely, this access to such intrusive technol- ating on ubiquitous networks with seamless con- APPENDIX I. 23 Technologies Coverage ogy can violate individual rights and invade the nectivity. The results of large computations running in IEEE Publications...... 134 privacy of innocent people. The onus is again on on massive cloud infrastructures will be available society to limit the use of seamless connectivity to as affordable services that almost anyone can acceptable norms. access and utilize. However, history has also taught us that more and more processing power becomes On the downside, the gap between developed available at the edges and in the hands of custom- and underdeveloped countries could continue to ers/users. In that regard, the cloud can be seen as increase. The seamlessness enjoyed in developed countries will be missed when a traveler finds it We are in a phase a processing fabric, part of the ambient environment, and a commodity. Cloud gets implemented hard to use a smart card at a merchant or a train of seamless as technicians and economists decide; to the end station in an area that does not have that technol- and ubiquitous user, it is irrelevant [Saracco]. ogy. The rapid evolution of increased automation and the spread of pervasive intelligence in tradi- intelligence. Seamless and ubiquitous intelligence aids in tional uses in everyday activities will accentuate

11 CONTENTS 1. INTRODUCTION...... 5 the differences between the have and have-not na- political activism, security, and safety, or it can be 2. SEAMLESS INTELLIGENCE tions. Nonetheless, underdeveloped countries will used for militarization, to invade privacy, and to SCENARIO...... 8 continue to enjoy access to advances in computing, push the Big Brother phenomenon worldwide, even 2.1 Introduction...... 8 particularly the use of inexpensive yet smart mobile in countries that consider and pride themselves on 2.2 State of the Art...... 8 devices. The trend seems to be toward facilitating being free. In general, any technology has its ups further social networking rather than real enhance- and downs. The man who invented the first ship 2.3 Challenges and Opportunities..... 9 ment in productivity tools. Furthermore, ubiquitous also invented the shipwreck and the castaway! 2.4 What Will Likely Happen...... 12 computational and educational services will grow This is something that we need to understand. CONTENTS2.5 Potential Disruptions...... 12 ever more accessible to any population that meets Even if technology is used in the best possible 2.6 Summary...... 13 the basic connectivity requirements. way, it will still bring along some downsides. It just 3. 23 TECHNOLOGIES IN 2022...... 14 changes the landscape, and along with it, the ups 4. DRIVERS AND DISRUPTORS...... 112 2.4 What Will Likely Happen and downs [Saracco]. 5. TECHNOLOGY COVERAGE IN The future we want versus the future we do not IEEE XPLORE AND BY 2.5 Potential Disruptions IEEE SOCIETIES...... 115 want: information and communication technology is advancing at a pace that is surpassing our The emergence of the mobile smart device sector 6. IEEE COMPUTER SOCIETY IN 2022...... 120 abilities as a society to direct. It is the scale and in the past decade is likely to continue to disrupt 7. SUMMARY AND NEXT STEPS.....124 speed with which this progress is taking place that the traditional model of desktops and laptops. Mo- is creating this challenge. But there are choices bile applications are also expanding the common 8. AUTHORS...... 126 that free nations can Web platform by enabling applications on mobile APPENDIX I. 23 Technologies Coverage make through regu- devices using their operating systems. in IEEE Publications...... 134 lation and investment On the other end of the computing scale, the that can either lead to emergence of cloud computing primarily based a better world or one on commodity server hardware is pushing and that we do not desire. disrupting the traditional server sector, replacing Technology is a it with computational power as a service over the double-edged sword. network. We are in a phase It can be used for Another disruptive trend emerging as a result advancing healthcare, of seamless of the spread of social networking is resulting in education, science, and ubiquitous countries and regions potentially creating their trade, financial ser- own regional Internets with imposed restrictions on intelligence. vices, social and access to global sites and universal services. This

12 CONTENTS 1. INTRODUCTION...... 5 trend can have a negatively disruptive impact on will offer continuous, uninterrupted services that 2. SEAMLESS INTELLIGENCE the global Internet and freedom of individuals to enhance automation, productivity, collaboration, SCENARIO...... 8 access information and services regardless of geo- and access to intelligence and knowledge that will 2.1 Introduction...... 8 graphic locations and political boundaries. There is be available not only at users’ fingertips but acces- 2.2 State of the Art...... 8 also the impact of regulation, such as the different sible to all human senses, spontaneously, through positions taken by the US and EU in the area of emerging human–computer interfaces. 2.3 Challenges and Opportunities..... 9 premium connectivity, which is allowed in the US 2.4 What Will Likely Happen...... 12 The benefit of technology is what we make of it. but not in the EU. CONTENTS2.5 Potential Disruptions...... 12 Societies will face further challenges in directing 2.6 Summary...... 13 Intellectual property wars among major players in and investing in technologies that benefit human- the industry can present barriers to both the speed ity instead of destroying it or intruding on basic 3. 23 TECHNOLOGIES IN 2022...... 14 of progress and use of technology. Consumers of human rights of privacy and freedom of access 4. DRIVERS AND DISRUPTORS...... 112 technology will ultimately be the victims of such to information. We should stop considering tech- 5. TECHNOLOGY COVERAGE IN wars. nology as something standalone. It is more than a IEEE XPLORE AND BY IEEE SOCIETIES...... 115 piece of the quilt of life: it is reshaping it, and being reshaped itself by humanity. A holistic approach is 6. IEEE COMPUTER SOCIETY 2.6 Summary needed. IN 2022...... 120 By 2022, computing devices will range from na- 7. SUMMARY AND NEXT STEPS.....124 no-scale to mega-scale, with advanced networking 2.6.1 References 8. AUTHORS...... 126 enabling access to a world of integrated services. [Saracco] Roberto Saracco, the author of COMSOC APPENDIX I. 23 Technologies Coverage Virtual connectivity will enable the integration of in IEEE Publications...... 134 relevant computing resources to provide users 2020 Report, Personal Communication. with seamless services. The resulting ecosystem

We are in a phase of seamless and ubiquitous intelligence.

13 CONTENTS 1. INTRODUCTION...... 5 2. SEAMLESS INTELLIGENCE SCENARIO...... 8 3. 23 TECHNOLOGIES IN 2022...... 14 3.1 Security Cross-Cutting Issues...... 14 3.2 The Open Intellectual 3. 23 Technologies in 2022 Property Movement...... 17 3.3 Sustainability...... 20 3.4 Massively Online Open Courses...... 25 3.5 Quantum Computing ...... 28 3.6 Device and Nanotechnology...... 31 3.1 Security Cross-Cutting Issues to collect diverse data about citizens, private 3.7 3D Integrated Circuits...... 33 and public corporations, and profit and nonprofit CONTENTS3.8 Universal Memory...... 38 3.9 Multicore...... 43 3.1.1 Introduction entities alike through a variety of channels. This 3.10 Photonics...... 47 Powerful forces are converging that are of great data includes financial transactions, personal 3.11 Networking and Interconnectivity... 52 concern to individuals and private and public and business correspondence, the movements 3.12 Software-Defined Networks...... 55 entities. These powerful forces will cause people, of people and assets, and personal and business 3.13 High-Performance Computing businesses, and groups to pause before releasing relationships. The fourth force is institution/mu- (HPC) ...... 63 nicipalities and crowd-sourced information. This 3.14 Cloud Computing...... 68 certain information to government, merchants, may be the first that will be exploited and will have 3.15 The Internet of Things...... 74 and even other citizens and to consider the conse- 3.16 Natural User Interfaces...... 77 quences of every activity in which they engage. an impact on society. The final force is the grow- 3.17 3D Printing ...... 81 ing ability and determination of malevolent actors 3.18 Big Data and Analytics ...... 84 The first of these forces is the exponential growth in acquiring information about people, business 3.19 Machine Learning and of large data repositories (see big data in Section entities, and objects such as critical infrastructure. Intelligent Systems ...... 89 3.18) of personal and corporate information. The Malevolent actors can include adversarial govern- 3.20 Computer Vision and Pattern second is the enhanced capability to analyze this ment agents, criminals, malcontents, and personal Recognition...... 92 data for various patterns (see data analytics). The 3.21 Life Sciences...... 96 or business enemies. third force is the advancing technological ability 3.22 Computational Biology and The convergence of these forces requires Bioinformatics...... 102 tradeoff decisions to be made about privacy 3.23 Medical Robotics...... 108 versus security. In order to protect individu- 4. DRIVERS AND DISRUPTORS...... 112 als and corporate entities from malevolent 5. TECHNOLOGY COVERAGE IN actors, governments must monitor personal IEEE XPLORE AND BY IEEE SOCIETIES...... 115 and business transactions and examine associations of people with other people 6. IEEE COMPUTER SOCIETY IN 2022...... 120 and with corporations and affinity groups. 7. SUMMARY AND NEXT STEPS.....124 Governments must track movements of 8. AUTHORS...... 126 APPENDIX I. 23 Technologies Coverage 14 in IEEE Publications...... 134 CONTENTS 1. INTRODUCTION...... 5 2. SEAMLESS INTELLIGENCE people and goods, monitor the utilization of private their own analytics for commercial or nefarious SCENARIO...... 8 and public resources, and mine data repositories in purposes. In some cases, such intrusions are lim- 3. 23 TECHNOLOGIES IN 2022...... 14 order to investigate or predict crimes, all in the in- ited by law, but the limitations vary by country, are 3.1 Security Cross-Cutting Issues...... 14 terest of protecting the public. In allowing govern- hard to enforce, and offer little protection should 3.2 The Open Intellectual ments to conduct these activities, however, individ- the intrusion come from a government agency au- Property Movement...... 17 uals and corporations must surrender privacy. How thorized to bypass the limits. Cryptographic meth- 3.3 Sustainability...... 20 much privacy should an individual, a corporation, ods to protect a user’s privacy exist for some uses, 3.4 Massively Online Open Courses...... 25 or an affinity group surrender in order to ensure an such as OpenPGP for email, but are often hard to 3.5 Quantum Computing ...... 28 acceptable level of security from threats? Should configure and use, and are sometimes blocked or 3.6 Device and Nanotechnology...... 31 3.7 3D Integrated Circuits...... 33 these limits be set legislatively? Is it feasible to pro- not well supported. CONTENTS3.8 Universal Memory...... 38 tect one’s privacy without legislative support, and 3.9 Multicore...... 43 what tools are available, or need to be available, to 3.10 Photonics...... 47 make it possible? 3.11 Networking and Interconnectivity... 52 3.12 Software-Defined Networks...... 55 3.1.2 State of the Art 3.13 High-Performance Computing (HPC) ...... 63 Social networking sites such as Facebook and 3.14 Cloud Computing...... 68 Twitter can be monitored and predictive analytics 3.15 The Internet of Things...... 74 used to investigate crimes or predict the potential 3.16 Natural User Interfaces...... 77 for crimes. Machine-to-machine networks (see 3.17 3D Printing ...... 81 Internet of Things in Section 3.15) can be used to 3.18 Big Data and Analytics ...... 84 3.19 Machine Learning and track individual products or subsystems of inter- Intelligent Systems ...... 89 ests using RFID; whole systems and people can 3.20 Computer Vision and Pattern be tracked via GPS, terrestrial imaging, satellite Recognition...... 92 imaging, black boxes, and other low-technology 3.21 Life Sciences...... 96 means. Powerful Internet search engines exist, 3.1.3 Challenges 3.22 Computational Biology and Bioinformatics...... 102 and coupled with the ability to capture, store, and There is a balance between security and privacy. 3.23 Medical Robotics...... 108 analyze large amounts of data from surface mail Citizens, corporations, and other groups accept a 4. DRIVERS AND DISRUPTORS...... 112 and parcels, email traffic, telephone conversations, certain level of intrusion, provided a certain level of financial transactions, consumer purchases, and In- 5. TECHNOLOGY COVERAGE IN security is afforded. Every person, corporation, and IEEE XPLORE AND BY ternet sites visited, government agencies can mine group, however, has a different level of sensitivity IEEE SOCIETIES...... 115 this data and use predictive analytics to spot po- to intrusion and a different notion of acceptable 6. IEEE COMPUTER SOCIETY tential threats before they occur or to investigate security risk. IN 2022...... 120 crimes. Private entities and malevolent actors may There are political challenges of fostering public also gain access to this information and conduct 7. SUMMARY AND NEXT STEPS.....124 trust that transactions and movement are safe 8. AUTHORS...... 126 APPENDIX I. 23 Technologies Coverage 15 in IEEE Publications...... 134 CONTENTS 1. INTRODUCTION...... 5 2. SEAMLESS INTELLIGENCE without being overly intrusive. Commercial orga- against any increases in intrusion if the benefit of SCENARIO...... 8 nizations, such as Internet service providers, have increased security is not demonstrated. 3. 23 TECHNOLOGIES IN 2022...... 14 little incentive to provide and support privacy-en- 3.1 Security Cross-Cutting Issues...... 14 hancement tools, and in some cases, are under 3.1.6 Summary 3.2 The Open Intellectual pressure from regulators to avoid changes that will The growth of large data repositories of personal Property Movement...... 17 block law enforcement agencies from accessing 3.3 Sustainability...... 20 information, where data from many sources may private communications. 3.4 Massively Online Open Courses...... 25 be aggregated, combined with data analytics that 3.5 Quantum Computing ...... 28 enable deduction of surprisingly detailed patterns 3.6 Device and Nanotechnology...... 31 3.1.4 Where We Think It Will Go of information regarding individuals and groups, 3.7 3D Integrated Circuits...... 33 Citizens and corporate entities and groups have has opened a Pandora ’s Box of privacy issues. CONTENTS3.8 Universal Memory...... 38 always accepted a certain level of intrusion in order Privacy intrusions can come from both authorized 3.9 Multicore...... 43 3.10 Photonics...... 47 to ensure some level of security. Technological ad- sources such as law enforcement and corporations 3.11 Networking and Interconnectivity... 52 vances have simply focused more attention on this that have been explicitly granted permission, as 3.12 Software-Defined Networks...... 55 problem. If governments can show that real securi- well as malevolent actors such as identity thieves. 3.13 High-Performance Computing ty is achieved through surrendering a certain level We face a tradeoff among privacy, security, and (HPC) ...... 63 of privacy, then new technological advancements convenience. Changes in laws and improvements 3.14 Cloud Computing...... 68 that can perform accurate predictive and forensic in privacy-enhancement tools and techniques may 3.15 The Internet of Things...... 74 analytics will be embraced in exchange for a cer- be needed to help users find a balance between 3.16 Natural User Interfaces...... 77 3.17 3D Printing ...... 81 tain level of privacy being sacrificed. On the other the degree of intrusion they can tolerate and the 3.18 Big Data and Analytics ...... 84 hand, consumer demand for privacy-enhancing security they desire. 3.19 Machine Learning and tools may lead to changes that make it easier for Intelligent Systems ...... 89 individuals to protect their privacy, perhaps at the 3.1.7 References 3.20 Computer Vision and Pattern cost of some effort and inconvenience. Recognition...... 92 W. Diffie and S.E. Landau, Privacy on the Line: The Poli- 3.21 Life Sciences...... 96 tics of Wiretapping and Encryption, MIT Press, 2007. 3.22 Computational Biology and 3.1.5 Potential Disruptions S. Landau, “Politics, Love, and Death in a World of No Bioinformatics...... 102 Access to vast quantities of personal information Privacy,” IEEE Security & Privacy, vol. 11, no. 3, 2013, pp. 3.23 Medical Robotics...... 108 either in one repository (e.g. the Affordable Health- 11-13. 4. DRIVERS AND DISRUPTORS...... 112 care Database) or through aggregation of multiple I. Goldberg, “Privacy-Enhancing Technologies for the 5. TECHNOLOGY COVERAGE IN databases creates an irresistible target for hackers. Internet III: Ten Years Later,” Digital Privacy: Theory, IEEE XPLORE AND BY If infiltrated, no one can safely depend on their own IEEE SOCIETIES...... 115 Technologies, and Practices, Auerbach, 2007, pp. 3–18. identity being protected or can trust the identity of 6. IEEE COMPUTER SOCIETY IN 2022...... 120 anyone else with whom they engage in a personal or business transaction. The public may rebel 7. SUMMARY AND NEXT STEPS.....124 8. AUTHORS...... 126 APPENDIX I. 23 Technologies Coverage 16 in IEEE Publications...... 134 CONTENTS 1. INTRODUCTION...... 5 2. SEAMLESS INTELLIGENCE as a police force to ensure that the content is not SCENARIO...... 8 maliciously corrupted. The IP is generated for the 3. 23 TECHNOLOGIES IN 2022...... 14 beneficial use of humankind and is often covered by 3.1 Security Cross-Cutting Issues...... 14 the well-known Creative Commons license. 3.2 The Open Intellectual Property Movement...... 17 Open IP can be found in the form of information 3.3 Sustainability...... 20 repositories (e.g., Wikipedia), open source software 3.4 Massively Online Open Courses...... 25 (e.g., Linux), media repositories (e.g., Flickr), open 3.5 Quantum Computing ...... 28 access publishing (e.g., Public Library of Science), 3.6 Device and Nanotechnology...... 31 3.2 The Open Intellectual open systems (e.g., World Wide Web), protocols 3.7 3D Integrated Circuits...... 33 Property Movement (e.g., TCP/IP), programming languages (e.g., Ada), CONTENTS3.8 Universal Memory...... 38 3.9 Multicore...... 43 open hardware standards (e.g., USB), and even 3.10 Photonics...... 47 3.2.1 Introduction hardware designs (e.g., Open Compute Project) 3.11 Networking and Interconnectivity... 52 Open intellectual property (IP), such as that found and 3D models for home printing (e.g., Blendswap). 3.12 Software-Defined Networks...... 55 in open source software, open standards, and open Much of the intellectual underpinnings of the 3.13 High-Performance Computing access publishing (along with crowd-sourcing as (HPC) ...... 63 open IP movement can be found in The Wisdom a means of producing information) is a significant 3.14 Cloud Computing...... 68 of Crowds [Surowiecki]. Surowiecki argues that 3.15 The Internet of Things...... 74 positive byproduct of the ubiquity of the World these crowd-driven movements are less subject 3.16 Natural User Interfaces...... 77 Wide Web. It is rapidly expanding into areas where to political forces and more dependent on ex- 3.17 3D Printing ...... 81 property was traditionally proprietary, such as pert knowledge, are necessarily more well-co- 3.18 Big Data and Analytics ...... 84 hardware design. Continued growth of the open ordinated, and more trust is established than 3.19 Machine Learning and IP movement will continue to generate significant in plan-driven IP development by hierarchical Intelligent Systems ...... 89 benefits to humankind. 3.20 Computer Vision and Pattern teams. Interestingly, even the prerequisite initial Recognition...... 92 But along with these benefits come significant funding for costly projects is now often obtained 3.21 Life Sciences...... 96 challenges and risks, including security and trust, via crowd-funding. This further democratizes and 3.22 Computational Biology and expands the range of available open IP, which Bioinformatics...... 102 motivation for innovators, and diminishment of traditionally was limited on the higher end to a 3.23 Medical Robotics...... 108 individuality. few organizations with appropriate resources 4. DRIVERS AND DISRUPTORS...... 112 and inclinations to share their results. 5. TECHNOLOGY COVERAGE IN 3.2.2 State of the Art IEEE XPLORE AND BY Open IP is information contained in freely accessible All crowd-sourcing applications, such as the afore- IEEE SOCIETIES...... 115 repositories in which volunteers, often in very large mentioned open IP ones, have four basic elements: 6. IEEE COMPUTER SOCIETY numbers, produce and vet the content. Users of this a division of labor, computing and communications IN 2022...... 120 information also provide feedback to the communi- technology, a crowd of human workers, and a labor 7. SUMMARY AND NEXT STEPS.....124 ty, driving innovation, correcting errors, and acting market [Grier]. Social media such as Facebook, 8. AUTHORS...... 126 APPENDIX I. 23 Technologies Coverage 17 in IEEE Publications...... 134 CONTENTS 1. INTRODUCTION...... 5 2. SEAMLESS INTELLIGENCE Twitter, Digg, and others provide a ready platform SCENARIO...... 8 for other kinds of open information applications. 3. 23 TECHNOLOGIES IN 2022...... 14 For example, social media information has been 3.1 Security Cross-Cutting Issues...... 14 used to create contemporaneous trouble-spot 3.2 The Open Intellectual maps and help relief agencies share information in Property Movement...... 17 response to disasters (Gao et al.). Large numbers 3.3 Sustainability...... 20 3.4 Massively Online Open Courses...... 25 of volunteer participants using community sites 3.5 Quantum Computing ...... 28 (e.g., PatientsLikeMe) have collaborated to share 3.6 Device and Nanotechnology...... 31 personal information in the creation of the medical 3.7 3D Integrated Circuits...... 33 data used for disease research and epidemiology CONTENTS3.8 Universal Memory...... 38 (Reidl and Reidl). And crowd-sourcing has been information creation, what incentives are there 3.9 Multicore...... 43 used to maintain contemporaneously accurate for individuals to contribute? Not every human is 3.10 Photonics...... 47 maps and to translate large quantities of text from altruistically motivated, and it is unlikely that some- 3.11 Networking and Interconnectivity... 52 one language to another. There are also various 3.12 Software-Defined Networks...... 55 one can earn a living through the micropayments 3.13 High-Performance Computing entertainment applications of crowd-source-like offered by some crowd-based initiatives. When (HPC) ...... 63 communities, such as in massively multiplayer information creators forego a copyright, there is a 3.14 Cloud Computing...... 68 games and in the creation of artistic and educa- blurring of public-private relationships, and some 3.15 The Internet of Things...... 74 tional works. Similarly, there have been educational measure of individuality is lost. 3.16 Natural User Interfaces...... 77 benefits from the proliferation of crowd-sourced The distributed and often unchecked nature of 3.17 3D Printing ...... 81 classes online. 3.18 Big Data and Analytics ...... 84 the crowd-sourced worker can also lead to mis- 3.19 Machine Learning and takes, cheating, and poor-quality work. While Intelligent Systems ...... 89 3.2.3 Challenges crowd-sourcing has built-in mechanisms for 3.20 Computer Vision and Pattern Recognition...... 92 Safety, truth, and accuracy: Is the information work-checking and fault-tolerance, these mecha- 3.21 Life Sciences...... 96 contained in open information repositories (e.g., nisms are imperfect (Grier). Wikipedia) true? Is the open source software 3.22 Computational Biology and Ostensibly beneficial open intellectual movements Bioinformatics...... 102 downloaded for use in a critical application safe to could actually be ruses designed to trick human 3.23 Medical Robotics...... 108 use, or does it contain a critical defect or a security workers for some nefarious purposes. There are 4. DRIVERS AND DISRUPTORS...... 112 flaw? Eric Raymond, one of the fathers of the open instances of such ruses being perpetrated already, source software movement, contends that “with 5. TECHNOLOGY COVERAGE IN for example, the notorious “Captcha Busting Tro- IEEE XPLORE AND BY enough eyes, all bugs are shallow,” but this obser- jan,” in which a game was used to trick users into IEEE SOCIETIES...... 115 vation isn’t always correct. Crowds can be fooled, solving Captcha puzzles that were actually intend- 6. IEEE COMPUTER SOCIETY and collective intelligence can be wrong [Cox]. IN 2022...... 120 ed to thwart automated email account generation. If open information creation displaces commercial 7. SUMMARY AND NEXT STEPS.....124 Such an approach could be used to, say, recruit an 8. AUTHORS...... 126 APPENDIX I. 23 Technologies Coverage 18 in IEEE Publications...... 134 CONTENTS 1. INTRODUCTION...... 5 2. SEAMLESS INTELLIGENCE army of volunteers who think they are working on Open IP could dramatically accelerate innovation, SCENARIO...... 8 some important mathematical problem into using information dissemination, and quality of life im- 3. 23 TECHNOLOGIES IN 2022...... 14 brute-force techniques to crack passwords on a provements (particularly in disadvantaged nations). 3.1 Security Cross-Cutting Issues...... 14 secure site. On the hardware side, open designs could acceler- 3.2 The Open Intellectual ate technological developments and lower prices Property Movement...... 17 Another large arena challenged by the open IP for devices from the hobbyist’s toys to high-end 3.3 Sustainability...... 20 trend is legal. There is the obvious conflict with servers. 3.4 Massively Online Open Courses...... 25 laws to protect IP from sharing and use, such as 3.5 Quantum Computing ...... 28 trademark and patent laws. Already, several nota- The open movement could also greatly change the 3.6 Device and Nanotechnology...... 31 ble examples exist where open IP was challenged way society views IP ownership as it shifts from 3.7 3D Integrated Circuits...... 33 in court (e.g., the SCO versus IBM lawsuit over private to public. CONTENTS3.8 Universal Memory...... 38 Linux). As open data encompasses more and more 3.9 Multicore...... 43 A “scandal” involving flawed information in open IP 3.10 Photonics...... 47 fields, such as the ability to freely print 3D models (either through mistake or deliberate malfeasance) 3.11 Networking and Interconnectivity... 52 at home, legal challenges will range from liability could cause a major disaster that calls into ques- 3.12 Software-Defined Networks...... 55 over manufactured parts (including weapons) to tion the entire open information movement. The le- 3.13 High-Performance Computing ownership of their design. Current copyright laws gal system has yet to adapt to the rapidly changing (HPC) ...... 63 can limit the rights over a 3D model but are ill 3.14 Cloud Computing...... 68 reality of open IP, and we may risk bottlenecking equipped to address the rights over the resulting 3.15 The Internet of Things...... 74 the promised progress in litigation and paperwork. 3.16 Natural User Interfaces...... 77 physical output. 3.17 3D Printing ...... 81 3.2.6 Summary 3.18 Big Data and Analytics ...... 84 3.2.4 Where We Think It Will Go The open IP movement has moved beyond an ex- 3.19 Machine Learning and Open IP generation will be very successful in Intelligent Systems ...... 89 perimental phase and will be a permanent fixture in 3.20 Computer Vision and Pattern certain niches, for example, encyclopedias, open society. How impactful this movement will be may Recognition...... 92 standards, and open programming language. It will largely depend on government actions or inactions 3.21 Life Sciences...... 96 be only partially successful in certain niches (e.g., regarding the treatment of this property; it might 3.22 Computational Biology and open access publishing, open source software). Bioinformatics...... 102 also be subject to the chaotic events of fate. Open IP movements may fail in other domains. 3.23 Medical Robotics...... 108 4. DRIVERS AND DISRUPTORS...... 112 3.2.7 References 3.2.5 Potential Disruptions 5. TECHNOLOGY COVERAGE IN L.P. Cox, “Truth in Crowdsourcing,” IEEE Security & IEEE XPLORE AND BY For certain market segments, it might be impos- Privacy, vol. 9, no. 5, 2011, pp. 74-76. IEEE SOCIETIES...... 115 sible for the free market to compete with open G. Huiji, G. Barbier, and R. Goolsby, “Harnessing the 6. IEEE COMPUTER SOCIETY information counterparts, for example, in academic Crowdsourcing Power of Social Media for Disaster IN 2022...... 120 publishing, and we may see the end of the tradi- Relief,” IEEE Intelligent Systems, vol. 26, no. 3, 2011, pp. 7. SUMMARY AND NEXT STEPS.....124 tional paid-for scholarly journal. 10-14 8. AUTHORS...... 126 APPENDIX I. 23 Technologies Coverage 19 in IEEE Publications...... 134 CONTENTS 1. INTRODUCTION...... 5 2. SEAMLESS INTELLIGENCE D.A. Grier, “Not for All Markets,” Computer, vol. 44, no. 5, Water is used both to cool datacenters and to pro- SCENARIO...... 8 2011, pp. 6-8. duce equipment. In certain areas, it is a scarce re- 3. 23 TECHNOLOGIES IN 2022...... 14 J. Riedl and E. Riedl, “Crowdsourcing Medical Re- source and must be handled with a lot of care—for 3.1 Security Cross-Cutting Issues...... 14 search,” Computer, vol. 46, no. 1, 2013, pp. 89-92. example, in the Middle East and India. In some cas- 3.2 The Open Intellectual es, water can be contaminated during the process Property Movement...... 17 J. Surowiecki, The Wisdom of Crowds, Random House and require treatment. Carbon is produced when 3.3 Sustainability...... 20 Digital, 2005. 3.4 Massively Online Open Courses...... 25 burning fuel and needs to be removed by plants. 3.5 Quantum Computing ...... 28 Materials (steel aluminum, etc.) used during the 3.6 Device and Nanotechnology...... 31 production of various pieces of equipment must be 3.7 3D Integrated Circuits...... 33 recycled or otherwise contribute to ever-increasing CONTENTS3.8 Universal Memory...... 38 garbage dumps. Global warming is a result of heat- 3.9 Multicore...... 43 ing and cooling datacenters, in addition to other 3.10 Photonics...... 47 factors, and can have detrimental consequences 3.11 Networking and Interconnectivity... 52 to the Earth, especially as temperatures and water 3.12 Software-Defined Networks...... 55 3.13 High-Performance Computing levels rise. (HPC) ...... 63 There are three aspects of sustainability in computer 3.14 Cloud Computing...... 68 systems: economical, the financial impact of energy 3.15 The Internet of Things...... 74 spent running CPUs, memory, networking, storage, 3.16 Natural User Interfaces...... 77 3.3 Sustainability 3.17 3D Printing ...... 81 etc.; environmental, the impact on the environment, 3.18 Big Data and Analytics ...... 84 such as how much CO is spent or how much water 3.3.1 Introduction 2 3.19 Machine Learning and is used in running datacenters; and social, summariz- Intelligent Systems ...... 89 Sustainability in computer science is defined as a ing the impact on the area where computer systems 3.20 Computer Vision and Pattern means of maintaining/preserving resources in IT Recognition...... 92 are executing, for example, the GDP of the region, service delivery to users. It is a confluence of sup- stability of the region, any temporary influences, such 3.21 Life Sciences...... 96 ply and demand, where the IT ecosystem plays an 3.22 Computational Biology and as earthquakes, tsunamis, etc. Bioinformatics...... 102 important role (see Figure 2) [12]. 3.23 Medical Robotics...... 108 Multiple Earth resources are the focus of sustain- 3.3.2 State of the Art 4. DRIVERS AND DISRUPTORS...... 112 ability. Electricity (gas, coal, etc.), for example, is Today, sustainability-aware technologists prefer to 5. TECHNOLOGY COVERAGE IN critical in many datacenters, not only because it consider cradle-to-cradle design, that is, resource IEEE XPLORE AND BY contributes to operational costs, but also because IEEE SOCIETIES...... 115 consumption from a product’s inception to its it impacts overall sustainability. The more power retirement. This includes all resources used to ship 6. IEEE COMPUTER SOCIETY used from renewable energy sources, the more IN 2022...... 120 the product, its usage throughout its lifetime, and sustainable operations will be. finally the recycling of it. 7. SUMMARY AND NEXT STEPS.....124 8. AUTHORS...... 126 APPENDIX I. 23 Technologies Coverage 20 in IEEE Publications...... 134 CONTENTS IEEE CS 2022 , Report DRAFT 23 Technologies in 2022 1/26/2014 5:18 PM 1. INTRODUCTION...... 5 2. SEAMLESS INTELLIGENCE Standards have increas- SCENARIO...... 8 ing importance in abiding 3. 23 TECHNOLOGIES IN 2022...... 14 by “green” energy usage 3.1 Security Cross-Cutting Issues...... 14 guidelines, disposing of ma- 3.2 The Open Intellectual terials, and recycling equip- Property Movement...... 17 ment that is obsolete. 3.3 Sustainability...... 20 3.4 Massively Online Open Courses...... 25 Technology can help in 3.5 Quantum Computing ...... 28 this regard by turning off 3.6 Device and Nanotechnology...... 31 infrastructure when it is not 3.7 3D Integrated Circuits...... 33 used or energy proportion- CONTENTS3.8 Universal Memory...... 38 3.9 Multicore...... 43 ate, optimizing the load by 3.10 Photonics...... 47 moving it around the data- 3.11 Networking and Interconnectivity... 52 center to minimize energy 3.12 Software-Defined Networks...... 55 consumption. 3.13 High-Performance Computing (HPC) ...... 63 Sustainability can be 3.14 Cloud Computing...... 68 achieved at all levels of the 3.15 The Internet of Things...... 74 system, from savings in ma- Figure 2 . IT ecosystem from supply to demand. 3.16 Natural User Interfaces...... 77 terials, such as NVM mem- Figure 2. IT ecosystem from supply to demand. 3.17 3D Printing ...... 81 ories or photonics, which3.3.3 Challenges 3.18 Big Data and Analytics ...... 84 number of users connected to the Internet (expect- have much lower powerWith consumption the increasing compared population ed to be growth near 3 (3 billion billion by 2025 [15] more ted consumers predic ), energy by [14] 2030 con) - and number of 3.19 Machine Learning and to DRAMs and electronic interconnects, to making Intelligent Systems ...... 89 users connected to the Internet sumption (expected will also rise. to North be America [15] near 3 billion by 2025 ), and energy Western consumption will 3.20 Computer Vision and Pattern tradeoffs in hardware also architecture, rise. including North America using Europe and are Western saturated Europe in terms are saturated in terms of of Internet Internet users users and mobile Recognition...... 92 dark silicon, which enablesphones, only but parts of we systems can to expect and mobile growth phones, in but continental we can expect China growth and India, in South Asia, Africa, and South America. 3.21 Life Sciences...... 96 be turned on, to using The intelligent amount migration of of data virtu - produced continental is China larger and than India, South the Moore’s Asia, Africa, law and equivalent in processing, and the Internet 3.22 Computational Biology and al machines to enable consolidation and powering Bioinformatics...... 102 of Things will introduce additional South data America. produced The amount-‐ computer to-‐computer of data and produced device-‐to -‐iscomputer. down parts of unused datacenters, all the way 3.23 Medical Robotics...... 108 larger than the Moore’s law equivalent in process- At the same time, scaling of technology, particularly for CPUs, has all but stopped, and new ways of up to giving applications and services intelligent ing, and the Internet of Things will introduce addi- 4. DRIVERS AND DISRUPTORS...... 112 using parallelism have been adopted. In the past, power t was almost free and no on most people’s placement and design policies to enable optimal tional data produced computer-to-computer and 5. TECHNOLOGY COVERAGE IN utilization. minds. With increasing power consumption requirements, datacenters are now built near power plants IEEE XPLORE AND BY device-to-computer. or where ambient cooling reduces their cooling costs. Yet datacenter cooling still affects global warming. IEEE SOCIETIES...... 115 The increasing power consumption At of the same hardware time, scaling components of technology, has led particular to - more power capping. Computer 6. IEEE COMPUTER SOCIETY 3.3.3 Challenges manufacturers have started ly to for CPUs, think has upfront all but stopped, about recycling and new ways the of materials used to produce IT devices. IN 2022...... 120 With the increasing population growth (3 billion Computers are frequently using assembled parallelism in have areas been where adopted. they In will the past, be energy sold or at hubs where the 7. SUMMARY AND NEXT STEPS.....124 more consumers predicted by 2030 [14]) and required to deliver equipment to customers is optimized. 8. AUTHORS...... 126 APPENDIX I. 23 Technologies Coverage However, the same production processes can be optimized in many other industries. Transporting 21 in IEEE Publications...... 134 computing products is the same as any other good. The opportunity to use technology in these areas are vast, such as deploying intelligent and sustainable data sensors, educating professionals, building and deploying sustainable resources into ecosystems to oversee [13] processes, etc. .

20 CONTENTS 1. INTRODUCTION...... 5 2. SEAMLESS INTELLIGENCE power was almost free and not on most people’s Of most interest is power capping and power-driv- SCENARIO...... 8 minds. With increasing power consumption re- en management, the ability to think in terms of how 3. 23 TECHNOLOGIES IN 2022...... 14 quirements, datacenters are now built near power much energy a program will consume, not just how 3.1 Security Cross-Cutting Issues...... 14 plants or where ambient cooling reduces their long it will execute. For this to happen, more de- 3.2 The Open Intellectual cooling costs. Yet datacenter cooling still affects tailed instrumentation is required, to help software Property Movement...... 17 global warming. The increasing power consumption systems make policy decisions. 3.3 Sustainability...... 20 of hardware components has led to more power 3.4 Massively Online Open Courses...... 25 In addition, the increased configurability found in 3.5 Quantum Computing ...... 28 capping. Computer manufacturers have started to hardware, such as turning off parts of computers 3.6 Device and Nanotechnology...... 31 think upfront about recycling the materials used to and using dark silicon, will enable better optimi- 3.7 3D Integrated Circuits...... 33 produce IT devices. Computers are frequently as- zation in specific applications. For example, the CONTENTS3.8 Universal Memory...... 38 sembled in areas where they will be sold or at hubs 3.9 Multicore...... 43 existence of different CPUs (powerful or less pow- where the energy required to deliver equipment to erful), GPGPUs, accelerators, FPGAs, etc., can be 3.10 Photonics...... 47 customers is optimized. 3.11 Networking and Interconnectivity... 52 optimized for different applications, enabling better 3.12 Software-Defined Networks...... 55 However, the same production processes can be power utilization for different applications at differ- 3.13 High-Performance Computing optimized in many other industries. Transporting ent times and overall aggregate sustainability. (HPC) ...... 63 computing products is the same as any other good. 3.14 Cloud Computing...... 68 The opportunity to use technology in these areas 3.15 The Internet of Things...... 74 3.3.5 Potential Disruptions are vast, such as deploying intelligent and sustain- 3.16 Natural User Interfaces...... 77 There are several opportunities for disruptive 3.17 3D Printing ...... 81 able data sensors, educating professionals, building improvement to sustainability. Frequently, it is 3.18 Big Data and Analytics ...... 84 and deploying sustainable resources into ecosys- the new or improved use of existing technologies 3.19 Machine Learning and tems to oversee processes, etc. [13]. Intelligent Systems ...... 89 that becomes disruptive. End-to-end resource 3.20 Computer Vision and Pattern management in manufacturing is one of the ob- Recognition...... 92 3.3.4 Where We Think It Will Go vious examples. While it has been approached in 3.21 Life Sciences...... 96 First and foremost, sustainability awareness is computer equipment manufacturing, it has yet to 3.22 Computational Biology and required at all levels. Technology can substantially be widely adopted in other areas of manufactur- Bioinformatics...... 102 help in many areas of productivity, making processing and industries. There is a huge potential for 3.23 Medical Robotics...... 108 es much more sustainable. Once sustainability can conscious and sustainable approach to resource 4. DRIVERS AND DISRUPTORS...... 112 be measured, it can be controlled. management. 5. TECHNOLOGY COVERAGE IN IEEE XPLORE AND BY The second issue is the regulations and incentives One specific example of resource management is IEEE SOCIETIES...... 115 governments can introduce to both prevent com- in sustainable or smart cities. The use of the In- 6. IEEE COMPUTER SOCIETY panies and individuals from malpractice in terms of ternet of Things further improves the benefits of IN 2022...... 120 sustainability and to encourage them to improve smart cities, by enabling innovation at many levels. 7. SUMMARY AND NEXT STEPS.....124 sustainability of their business. New environmental approaches to cooling with 8. AUTHORS...... 126 APPENDIX I. 23 Technologies Coverage 22 in IEEE Publications...... 134 CONTENTS 1. INTRODUCTION...... 5 2. SEAMLESS INTELLIGENCE zero-energy datacenters combine solar energy Another potential disruption is new generations of SCENARIO...... 8 with careful datacenter management, for example. chips based on graphene and on metalferroids. In 3. 23 TECHNOLOGIES IN 2022...... 14 Electronic cars are another obvious disruptive both cases, there might be a three-order magni- 3.1 Security Cross-Cutting Issues...... 14 tude reduction in power consumption [16]. 3.2 The Open Intellectual technology whose benefits are in terms of re- Property Movement...... 17 duced pollution and eliminating non-renewable 3.3 Sustainability...... 20 energy sources. Remaining issues include the lack 3.3.6 Summary 3.4 Massively Online Open Courses...... 25 of acceleration (even though Tesla’s line of cars There are many ways how technology can help 3.5 Quantum Computing ...... 28 addresses this), refueling time, battery capacity, improve sustainability. Big data analytics (see 3.6 Device and Nanotechnology...... 31 and lifetime. Section 3.18) and Internet of Things (see Section 3.7 3D Integrated Circuits...... 33 3.15) will further enable and automate sustain- CONTENTS3.8 Universal Memory...... 38 Light-emitting diodes (LEDs) are another exam- able processes. Satellites sending images of air 3.9 Multicore...... 43 ple of an existing technology that can disrupt the pollution could enable quick detection and early 3.10 Photonics...... 47 future in terms of sustainability. Today’s LEDs are prevention. Governance, standards, and increased 3.11 Networking and Interconnectivity... 52 used in automotive lighting (traffic lights, cars, 3.12 Software-Defined Networks...... 55 awareness will also help from the oversight and planes, etc.) and are not widely deployed for gen- 3.13 High-Performance Computing process perspective. Holistic approaches, such as eral-purpose lightning. However, they have the (HPC) ...... 63 cradle-to-cradle, will be increasing required. For 3.14 Cloud Computing...... 68 advantage of being incandescent light sources in example, it is estimated that there will be over 6 3.15 The Internet of Things...... 74 terms of lower energy consumption, small sizes, billion phones by 2017—only a holistic approach 3.16 Natural User Interfaces...... 77 robustness, quick switching times, long lifetimes, will be able to address this electronic waste. At the 3.17 3D Printing ...... 81 etc. Once the cost is reduced and voltage/currency same time, cloud computing will help with reducing 3.18 Big Data and Analytics ...... 84 control improved, they could have a sustainable 3.19 Machine Learning and and optimizing power consumption through con- advantage of fluorescent lighting. Intelligent Systems ...... 89 solidating resources, and social media platforms, 3.20 Computer Vision and Pattern Consumer energy storage, new types of batteries Recognition...... 92 such as Twitter, can quickly increase public aware- (silicon anode; lithium iron phosphate), and renew- 3.21 Life Sciences...... 96 ness in case of violations. able energies, including increased solar energy 3.22 Computational Biology and Sustainability has become an important factor in Bioinformatics...... 102 use and biofuels, can be disruptive technologies industry and public awareness. It has been sub- 3.23 Medical Robotics...... 108 impacting many industries. Improved consumer stantially improved, but the growing needs are and home energy management have many savings 4. DRIVERS AND DISRUPTORS...... 112 increasing a gap with the available reserves of opportunities. Appliance lifecycle assessment tools 5. TECHNOLOGY COVERAGE IN water, energy, materials, and greenhouse gases. IEEE XPLORE AND BY could predict when it is more sustainable to replace Therefore, humanity needs to continue and even IEEE SOCIETIES...... 115 them; smart appliances can react and adjust to the increase sustainability to protect our future. 6. IEEE COMPUTER SOCIETY grid disturbances and price changing to optimize IN 2022...... 120 consumption and cost; and a similar impact can be 7. SUMMARY AND NEXT STEPS.....124 achieved by facilities energy management. 8. AUTHORS...... 126 APPENDIX I. 23 Technologies Coverage 23 in IEEE Publications...... 134 CONTENTS 1. INTRODUCTION...... 5 2. SEAMLESS INTELLIGENCE 3.3.7 References [11] W. Adams, “The Future of Sustainability: Re-think- SCENARIO...... 8 [1] X. Fan et al., “Power Provisioning for a Ware- ing Environment and Development in the Twen- 3. 23 TECHNOLOGIES IN 2022...... 14 house-sized Computer,” ISCA, ACM, 2007. ty-first Century,” Report of the IUCN Renowned 3.1 Security Cross-Cutting Issues...... 14 [2] P. Mahadevan et al., “A Power Benchmarking Thinkers Meeting, 2006. 3.2 The Open Intellectual Framework for Networking devices,” IFIP Network- [12] C. Patel, “Sustainable Ecosystems: Enabled by Property Movement...... 17 ing Conf., 2009. Supply and Demand Management,” M.K. Aguilera 3.3 Sustainability...... 20 [3] C. Belady et al., “Green Grid Data Center Power et al., eds., Proc. ICDCN 2011, LNCS 6522, 2011, pp. 3.4 Massively Online Open Courses...... 25 Efficiency Metrics: PUE AND DCIE,” White Paper 6, 12-28. st 3.5 Quantum Computing ...... 28 The Green Grid. 2008. [13] NSF Cyberinfrastructure for 21 Century Science 3.6 Device and Nanotechnology...... 31 [4] R. Sharma et al., “Water Efficiency Management in and Engineering (CIF21), Advanced Computing 3.7 3D Integrated Circuits...... 33 Data Centers: Metrics and Methodology,” ISSST, Infrastructure, Vision and Strategic Plan. 2012. CONTENTS3.8 Universal Memory...... 38 2009. [14] McKinsey Global Institute, McKinsey Sustainability 3.9 Multicore...... 43 [5] R. Silliman, “Vendor Survey Analysis: Benchmark- & Resource Productivity Practice, Resource Revo- 3.10 Photonics...... 47 ing Hardware Support Operations, N. Am., 2009,” lution: Meeting the World’s Energy, Materials, Food, 3.11 Networking and Interconnectivity... 52 and Water Needs, 2011. 3.12 Software-Defined Networks...... 55 ID Number G00165983. [6] C.D. Patel et al., “Energy Flow in the Information [15] McKinsey Global Institute, “Disruptive Technolo- 3.13 High-Performance Computing gies: Advances that Will Transform Life, Business, (HPC) ...... 63 Technology Stack,” Proc. IMECE, 2006. and the Global Economy,” 2013. 3.14 Cloud Computing...... 68 [7] “The Ecoinvent v2 Database,” PRé Consultants; [16] Saracco, Personal Communication. 3.15 The Internet of Things...... 74 http://www.pre.nl/ecoinvent/default.htm. 3.16 Natural User Interfaces...... 77 [8] T.J. Breen et al., “From Chip to Cooling Tower 3.17 3D Printing ...... 81 Data Center Modeling: Part I, Influence of Server 3.18 Big Data and Analytics ...... 84 Inlet Temperature and Temperature Rise across 3.19 Machine Learning and Cabinet,” Proc. IEEE Intersociety Conf. Thermal Intelligent Systems ...... 89 and Thermomechanical Phenomena in Electronic 3.20 Computer Vision and Pattern Systems (ITHERM), 2008. Recognition...... 92 [9] M. Manos and C. Belady, “What’s Your PUE Strat- 3.21 Life Sciences...... 96 egy?,” 2008; http://blogs.msdn.com/the_ pow- 3.22 Computational Biology and Bioinformatics...... 102 er_of_software/archive/2008/07/07/part-3-what- s-your-pue-strategy.aspx. 3.23 Medical Robotics...... 108 [10] World Bank World Development Indictors (WDI) 4. DRIVERS AND DISRUPTORS...... 112 database, http://data.worldbank.org/indicator. 5. TECHNOLOGY COVERAGE IN IEEE XPLORE AND BY IEEE SOCIETIES...... 115 6. IEEE COMPUTER SOCIETY IN 2022...... 120 7. SUMMARY AND NEXT STEPS.....124 8. AUTHORS...... 126 APPENDIX I. 23 Technologies Coverage 24 in IEEE Publications...... 134 CONTENTS 1. INTRODUCTION...... 5 2. SEAMLESS INTELLIGENCE assessment achieved through automated exams. SCENARIO...... 8 A notable example is Circuits & Electronics, one of 3. 23 TECHNOLOGIES IN 2022...... 14 the first MOOCs offered through EdX. The second 3.1 Security Cross-Cutting Issues...... 14 “connective” learning model has less structure 3.2 The Open Intellectual and content. The learning presumably occurs via Property Movement...... 17 3.4 Massively Online Open Courses1 crowd-sourced interactions through blogs, thread- 3.3 Sustainability...... 20 3.4 Massively Online Open Courses...... 25 ed discussion boards, and email. In either model, 3.5 Quantum Computing ...... 28 3.4.1 Introduction graduate assistants might moderate the interac- 3.6 Device and Nanotechnology...... 31 Often drawing tens of thousands of students to tions and answer questions, but instructor-initiated 3.7 3D Integrated Circuits...... 33 a single section, massively open online courses interaction is rare—if not nonexistent. CONTENTS3.8 Universal Memory...... 38 (MOOCs) offer free, high-quality, university course 3.9 Multicore...... 43 While online or remote delivery of college course content to anyone with Internet access. Requir- content has been available for many decades, 3.10 Photonics...... 47 ing only a computer and Internet access to enroll, 3.11 Networking and Interconnectivity... 52 MOOCs differ in terms of scale and no-cost. Mas- MOOCs can be used for continuing education 3.12 Software-Defined Networks...... 55 sive enrollments allow world-class faculty and cur- 3.13 High-Performance Computing courses and credit-bearing undergraduate courses, ricula to be accessible to anyone. MOOCs can be (HPC) ...... 63 leading to degree programs and even graduation taken anywhere that has Internet access, including 3.14 Cloud Computing...... 68 education. sparsely populated areas, and those locations 3.15 The Internet of Things...... 74 3.16 Natural User Interfaces...... 77 The prospect of achieving huge economies of where it would be impractical to build a physical 3.17 3D Printing ...... 81 scale is alluring to deans and college presidents. university. A MOOC will probably be completed 3.18 Big Data and Analytics ...... 84 World-renowned scholars can reach immense by someone in Antarctica or on the International 3.19 Machine Learning and audiences. High-quality courses can be delivered Space Station soon. Intelligent Systems ...... 89 to heretofore underserved and remote populations, 3.20 Computer Vision and Pattern There are several major players in the MOOC Recognition...... 92 particularly in disadvantaged countries, having space, including Coursera, a consortium of 33 3.21 Life Sciences...... 96 enormous societal impact. These “universities with- colleges and universities; EdX, created by Harvard 3.22 Computational Biology and out walls” have the potential to transform higher and MIT; Kahn Academy, backed by Google and Bill Bioinformatics...... 102 education. But there are significant unresolved Gates; and Udacity. Currently, most MOOCs are 3.23 Medical Robotics...... 108 issues relating to educational quality and financial taken as non-credit bearing, though several univer- 4. DRIVERS AND DISRUPTORS...... 112 sustainability. sities have recently begun awarding credit for com- 5. TECHNOLOGY COVERAGE IN pleting certain MOOCs, passing additional tests, IEEE XPLORE AND BY 3.4.2 State of the Art IEEE SOCIETIES...... 115 and providing certain authenticating artifacts. A MOOC has two basic models. The first involves 6. IEEE COMPUTER SOCIETY MOOC courses can theoretically scale up with- IN 2022...... 120 Web-based and emailed course content, with out limit, from more than 100,000 students today 7. SUMMARY AND NEXT STEPS.....124 to millions in a single course. To date, millions of 1 Some of this article is adapted from Laplante 2013 with permission. 8. AUTHORS...... 126 APPENDIX I. 23 Technologies Coverage 25 in IEEE Publications...... 134 CONTENTS 1. INTRODUCTION...... 5 2. SEAMLESS INTELLIGENCE course enrollments in MOOCs have been recorded, Critics of MOOCs highlight the lack of instruc- SCENARIO...... 8 but it is unclear how many students have actually tor-student and student-student interaction. While 3. 23 TECHNOLOGIES IN 2022...... 14 completed these courses and how many credit it is possible for some students to interact through 3.1 Security Cross-Cutting Issues...... 14 hours have been earned worldwide. group assignments, threaded discussion boards, 3.2 The Open Intellectual and direct email, instructor-to-individual-student Property Movement...... 17 contact is limited to a select few students. In the 3.3 Sustainability...... 20 3.4.3 Challenges United States, the Department of Education re- 3.4 Massively Online Open Courses...... 25 Typical completion rates for MOOCs are less than 3.5 Quantum Computing ...... 28 8 percent of enrolled students, which may include quires courses to have “significant instructor-ini- 3.6 Device and Nanotechnology...... 31 the curious as well as committed and ill-prepared tiated contact” in order for that course to be ap- 3.7 3D Integrated Circuits...... 33 students. These completion rates are an order proved for financial aid credit. CONTENTS3.8 Universal Memory...... 38 3.9 Multicore...... 43 of magnitude lower than in a traditional college Whether the MOOC is hosted by a not-for-profit 3.10 Photonics...... 47 course. entity or a for-profit business, the finances have 3.11 Networking and Interconnectivity... 52 Assessment is another challenge. In order to allow to make sense. It takes significant investment to 3.12 Software-Defined Networks...... 55 for scale, MOOCs typically use multiple-choice, build and maintain the MOOC platform, fill course 3.13 High-Performance Computing matching, simple fill-in-the-blank, and other forms content and pay support staff, teaching assistants, (HPC) ...... 63 and professors (if they are not working pro bono). 3.14 Cloud Computing...... 68 of testing in which scoring can be automated. A pure philanthropic model would see the financial 3.15 The Internet of Things...... 74 Some MOOCs require deliverables that must be burden met entirely through grants, donations, and 3.16 Natural User Interfaces...... 77 assessed manually by instructors or teaching assis- 3.17 3D Printing ...... 81 tants, but these artifacts significantly limit course earnings on some foundation. Some small finan- 3.18 Big Data and Analytics ...... 84 size. cial successes have been reported, but no one 3.19 Machine Learning and has figured out how to make the finances work for Intelligent Systems ...... 89 Authentication of students is problematic, though MOOCs once they scale up and for the long run. 3.20 Computer Vision and Pattern this same problem exists for any online course. Recognition...... 92 There are solutions available, such as using cer- 3.21 Life Sciences...... 96 3.4.4 Where We Think It Will Go tified testing centers or biometric authentication. 3.22 Computational Biology and The value proposition is so compelling that MOOCs Bioinformatics...... 102 But these solutions can be expensive and logisti- will draw thousands of participating colleges and 3.23 Medical Robotics...... 108 cally challenging and will limit the MOOC scale-up universities, thousands of investors, and millions 4. DRIVERS AND DISRUPTORS...... 112 factor. Since most MOOCs use fully automated test grading, it is possible that an oracle will one of students from around the world, but in a limited 5. TECHNOLOGY COVERAGE IN way. Current MOOC offerings are targeted to the IEEE XPLORE AND BY day fool a MOOC test engine. We feel there is 50 undergraduate market, but there will probably be IEEE SOCIETIES...... 115 percent chance someone will write a program that a limited number of professional-, graduate-, and 6. IEEE COMPUTER SOCIETY will pass enough MOOC courses to have obtained IN 2022...... 120 a degree by 2022, arguably passing the Turing test even doctoral-level MOOCs. Even today, however, there are signs of reluctance and disappointment 7. SUMMARY AND NEXT STEPS.....124 for artificial intelligence. 8. AUTHORS...... 126 APPENDIX I. 23 Technologies Coverage 26 in IEEE Publications...... 134 CONTENTS 1. INTRODUCTION...... 5 2. SEAMLESS INTELLIGENCE on behalf of students, instructors, and universities. 3.4.6 Summary SCENARIO...... 8 We believe that most universities will either directly MOOCs have the potential to transform the higher 3. 23 TECHNOLOGIES IN 2022...... 14 participate in MOOCs for a select few credit and educational landscape, but it is too soon to tell how 3.1 Security Cross-Cutting Issues...... 14 significant this impact will be. MOOCs will likely 3.2 The Open Intellectual non-credit courses or grant certain allowances Property Movement...... 17 to those who complete MOOCs, for example, by play a future role predominately in continuing edu- 3.3 Sustainability...... 20 waiving a prerequisite if an appropriate MOOC has cation, course prerequisites, and, on a limited basis, 3.4 Massively Online Open Courses...... 25 been successfully completed. credit-bearing courses. It is unlikely, but possible, 3.5 Quantum Computing ...... 28 that complete credit-bearing courses from accred- 3.6 Device and Nanotechnology...... 31 3.4.5 Potential Disruptions ited universities will be available through MOOCs 3.7 3D Integrated Circuits...... 33 before 2022. CONTENTS3.8 Universal Memory...... 38 With no tuition required, the convenience of on- 3.9 Multicore...... 43 line learning, and access to world-class faculty, 3.10 Photonics...... 47 MOOCs have the potential to draw vast numbers of 3.4.7 References 3.11 Networking and Interconnectivity... 52 students away from traditional bricks-and-mortar V.G. Cerf, “Running AMOOC,” IEEE Internet Computing, 3.12 Software-Defined Networks...... 55 universities. A significant migration of students to vol. 17, no. 3, 2013, p. 88. 3.13 High-Performance Computing (HPC) ...... 63 MOOCs would threaten the viability of some tradi- P.A. Laplante, “Courses for the Masses?,” IT Professional, 3.14 Cloud Computing...... 68 tional colleges and universities, but we believe that vol. 15, no. 2, 2013, pp. 57-59. 3.15 The Internet of Things...... 74 there is a less than 10 percent likelihood that this 3.16 Natural User Interfaces...... 77 disruption will occur. 3.17 3D Printing ...... 81 3.18 Big Data and Analytics ...... 84 MOOCs also threaten to change the role of faculty, 3.19 Machine Learning and student, and teaching assistants and the nature Intelligent Systems ...... 89 of the university. For example, one quality metric 3.20 Computer Vision and Pattern for traditional universities is the average number Recognition...... 92 of students per class, with a lower ratio consid- 3.21 Life Sciences...... 96 3.22 Computational Biology and ered desirable. Automated course delivery and Bioinformatics...... 102 grading allows for immense upscaling of course 3.23 Medical Robotics...... 108 enrollments. Does the growth of MOOCs mean 4. DRIVERS AND DISRUPTORS...... 112 we will need fewer professors but more teaching 5. TECHNOLOGY COVERAGE IN assistants? We believe that there may be pressures IEEE XPLORE AND BY on traditional universities to scale course sizes by IEEE SOCIETIES...... 115 adopting partial MOOC attributes (e.g., more au- 6. IEEE COMPUTER SOCIETY tomated grading) but still preserving a high level of IN 2022...... 120 instructor-student interaction. 7. SUMMARY AND NEXT STEPS.....124 8. AUTHORS...... 126 APPENDIX I. 23 Technologies Coverage 27 in IEEE Publications...... 134 CONTENTS 1. INTRODUCTION...... 5 2. SEAMLESS INTELLIGENCE operations on data and offers—in theory—a de- SCENARIO...... 8 cisive speed advantage over computers based 3. 23 TECHNOLOGIES IN 2022...... 14 on current technology. The promised speed ad- 3.1 Security Cross-Cutting Issues...... 14 vantage is so momentous that many researchers 3.2 The Open Intellectual believe that no conceivable amount of progress Property Movement...... 17 in classical computer science will ever be able to 3.3 Sustainability...... 20 3.4 Massively Online Open Courses...... 25 bridge the gap between the power of QC and clas- 3.5 Quantum Computing ...... 28 sical computation. 3.6 Device and Nanotechnology...... 31 Shor’s algorithm, [1] published in 1994, proved on 3.7 3D Integrated Circuits...... 33 3.5 Quantum Computing a theoretical level that QC could efficiently factor CONTENTS3.8 Universal Memory...... 38 natural numbers. The problem of finding efficient 3.9 Multicore...... 43 3.5.1 Introduction 3.10 Photonics...... 47 algorithms for factoring in classical computing Moore’s law is still going strong and has been for 3.11 Networking and Interconnectivity... 52 remains an open challenge. In fact, the very lack of 3.12 Software-Defined Networks...... 55 several decades since Gordon Moore made his such an efficient factoring algorithm is the founda- 3.13 High-Performance Computing forecast in 1965. Continuing the pattern of Moore’s tion for the security of public-key cryptosystems. (HPC) ...... 63 law, we can expect that the limit for current litho- Shor’s discovery that QC can break the vast major- 3.14 Cloud Computing...... 68 graphic manufacturing processes will be reached ity of cryptographic protocols in use today followed 3.15 The Internet of Things...... 74 within the next few decades. While we will not by a multitude of subsequent theoretical break- 3.16 Natural User Interfaces...... 77 speculate in the exact timing of its demise, it is a 3.17 3D Printing ...... 81 throughs in QC research have generated signifi- 3.18 Big Data and Analytics ...... 84 fact that current approaches to the fabrication of cant public interest and kicked off a quest to build 3.19 Machine Learning and computer chips are starting to run up against the a practical QC device or a quantum computer. Intelligent Systems ...... 89 fundamental difficulties related to the extremely 3.20 Computer Vision and Pattern small scale of circuitry. As quantum effects are 3.5.2 State of the Art Recognition...... 92 known to interfere in the proper functioning of 3.21 Life Sciences...... 96 electronic circuits as they decrease in size, we may QC is still in its infancy. Until now, experiments 3.22 Computational Biology and reach the limit sooner. With time running out for have been carried out in which QC has only been Bioinformatics...... 102 applied to a limited number of quantum bits, so- 3.23 Medical Robotics...... 108 Moore’s law, it may be opportune to explore a para- digm shift from Newtonian or classic computing to called qubits, the quantum equivalent to bits in 4. DRIVERS AND DISRUPTORS...... 112 classical computing. Numerous companies, ac- 5. TECHNOLOGY COVERAGE IN alternative processing methods such as quantum computing (QC). ademic institutions, and national governments IEEE XPLORE AND BY support QC research to develop devices for both IEEE SOCIETIES...... 115 QC is based on the idea of using quantum me- civilian and military purposes. 6. IEEE COMPUTER SOCIETY chanical phenomena to execute our computations IN 2022...... 120 instead of classical Newtonian physics. QC uses Practical and experimental QC is rapidly gaining 7. SUMMARY AND NEXT STEPS.....124 quantum properties to represent data and perform momentum. From the beginning in 2001, when 8. AUTHORS...... 126 APPENDIX I. 23 Technologies Coverage 28 in IEEE Publications...... 134 CONTENTS 1. INTRODUCTION...... 5 2. SEAMLESS INTELLIGENCE Shor’s algorithm was first demonstrated by a group level with the most entropy. Fortunately, this par- SCENARIO...... 8 at IBM using a quantum computer with 7 qubits, to ticular property matches the core problem in most 3. 23 TECHNOLOGIES IN 2022...... 14 D-Wave Systems’ 2007 announcement of the first machine learning algorithms (minima detection 3.1 Security Cross-Cutting Issues...... 14 fully functional QC device supporting 16 qubits, in multidimensional space) and makes them ideal 3.2 The Open Intellectual research efforts have started sprouting in many candidates for QC using D-Wave’s device. Property Movement...... 17 labs. More recently, D-Wave announced that a 3.3 Sustainability...... 20 3.4 Massively Online Open Courses...... 25 512-qubits device would be installed at the new 3.5.3 Challenges 3.5 Quantum Computing ...... 28 Quantum Artificial Intelligence Lab, a collaboration What is the true potential of QC? Are there ba- 3.6 Device and Nanotechnology...... 31 among NASA, Google, and USRA [2]. These orga- sic limits to our ability to control and manipulate 3.7 3D Integrated Circuits...... 33 nizations are investing in practical applications of quantum systems (qubits) that will prevent us CONTENTS3.8 Universal Memory...... 38 QC because they believe it may help solve some of 3.9 Multicore...... 43 moving QC from theory to practice and deploy real their most challenging computer science problems, solutions on practical QC devices? A significant 3.10 Photonics...... 47 particularly in machine learning. 3.11 Networking and Interconnectivity... 52 amount of research and development still needs to 3.12 Software-Defined Networks...... 55 The fact that D-Wave’s QC device is shrouded in happen in this field to answer these crucial ques- 3.13 High-Performance Computing a veil of commercial secrecy has raised questions tions. D-Wave’s QC device appears to be just one (HPC) ...... 63 about whether it has actually managed to build a of many ways that practical QC can materialize. 3.14 Cloud Computing...... 68 viable QC device. While not all are convinced, a We need to understand what it takes to create 3.15 The Internet of Things...... 74 3.16 Natural User Interfaces...... 77 number of research papers exploring D-Wave’s de- general QC devices if at all possible. 3.17 3D Printing ...... 81 vice are lending some credence to the claims made QC is fundamentally changing our approach to 3.18 Big Data and Analytics ...... 84 by the manufacturer. computing and algorithm development. Deeply un- 3.19 Machine Learning and Intelligent Systems ...... 89 D-Wave’s QC device is optimized to find answers derstanding the counterintuitive aspects of QC will 3.20 Computer Vision and Pattern to problems that classical computers can only be essential to fully exploit the potential possibili- Recognition...... 92 solve by exhaustively trying every possible solution, ties it provides. With such a fundamentally differ- 3.21 Life Sciences...... 96 the so-called class of NP-hard problems. This QC ent approach to both computation and computer 3.22 Computational Biology and device utilizes one of nature’s own “algorithms,” architecture, few of today’s computer scientists are Bioinformatics...... 102 quantum annealing, which in a sense is hard-wired well-equipped to take on this challenge. 3.23 Medical Robotics...... 108 into the device’s physical design. When datasets 4. DRIVERS AND DISRUPTORS...... 112 are transferred to the device, they are converted 3.5.4 Where We Think It Will Go 5. TECHNOLOGY COVERAGE IN and represented as qubits. After that, the qubit IEEE XPLORE AND BY Even the most immediate future of QC will be hard IEEE SOCIETIES...... 115 configuration goes through a series of quantum to predict, but we believe that early indications mechanical transitions—quantum annealing—and 6. IEEE COMPUTER SOCIETY point in the direction of the integration of QC with- IN 2022...... 120 a result emerges. The laws of nature dictate that in large classic computing infrastructures where systems want to sink to the lowest possible energy 7. SUMMARY AND NEXT STEPS.....124 it will serve in specialized data-processing roles 8. AUTHORS...... 126 APPENDIX I. 23 Technologies Coverage 29 in IEEE Publications...... 134 CONTENTS 1. INTRODUCTION...... 5 2. SEAMLESS INTELLIGENCE similar to what we saw in the early days of graphi- SCENARIO...... 8 cal processing unit (GPU) deployments dedicated 3. 23 TECHNOLOGIES IN 2022...... 14 to number crunching. Today, GPUs have become 3.1 Security Cross-Cutting Issues...... 14 an integral part of datacenter servers and have 3.2 The Open Intellectual taken over many tasks previously reserved for the Property Movement...... 17 CPU. Perhaps QC will take the same path. 3.3 Sustainability...... 20 3.4 Massively Online Open Courses...... 25 There are other quantum effects worth consider- 3.5 Quantum Computing ...... 28 ing, in particular, wave guide spin technology that 3.6 Device and Nanotechnology...... 31 promises dramatic increases in transistor density 3.7 3D Integrated Circuits...... 33 and a three-order decrease in power consumption. CONTENTS3.8 Universal Memory...... 38 3.9 Multicore...... 43 This can extend Moore’s law into the next decade. 3.10 Photonics...... 47 In addition, graphene is a possible replacement for 3.11 Networking and Interconnectivity... 52 silicon [3]. 3.12 Software-Defined Networks...... 55 3.13 High-Performance Computing 3.5.5 Potential Disruptions (HPC) ...... 63 3.14 Cloud Computing...... 68 Practical quantum computers will be able to solve a mass market. Even these spectacular technologies 3.15 The Internet of Things...... 74 class of problems much more efficiently and quick- seem too rudimentary to exploit the full potential of 3.16 Natural User Interfaces...... 77 ly than classical computer systems. Whether it is quantum mechanics. With the emergence of QC, it 3.17 3D Printing ...... 81 Shor’s factorization algorithm or quantum search appears plausible that we are about to experience 3.18 Big Data and Analytics ...... 84 algorithms, they will execute much faster than any a new wave of innovations that will tear down many 3.19 Machine Learning and existing computational barriers. Intelligent Systems ...... 89 current algorithm can on a classical computing 3.20 Computer Vision and Pattern system. Research and development in QC by nature is much Recognition...... 92 The true impact of QC and the path it will take is broader in scope and further reaching than earlier 3.21 Life Sciences...... 96 technological innovations such as the transistor. 3.22 Computational Biology and not yet known. The potential is staggering since Bioinformatics...... 102 this computing approach at its most fundamental Yet the transistor’s amazing impact proved hard to 3.23 Medical Robotics...... 108 level is only constrained by the laws of physics. predict. We believe that despite that, QC is at such 4. DRIVERS AND DISRUPTORS...... 112 During the Industrial Revolution, technological an early application stage, it possesses a novelty and a potential that suggests the likelihood of an even 5. TECHNOLOGY COVERAGE IN progress was driven and constrained by our under- IEEE XPLORE AND BY standing of thermodynamics and Newtonian me- greater impact than the transistor has had. IEEE SOCIETIES...... 115 chanics—fast forward to the 20th century, when our 6. IEEE COMPUTER SOCIETY deeper understanding of physics shattered these 3.5.6 Summary IN 2022...... 120 constraints, bringing innovations such as lasers, Our understanding of QC is currently undergoing 7. SUMMARY AND NEXT STEPS.....124 transistors, chips, and computing devices to the radical changes as it moves from being an esoteric 8. AUTHORS...... 126 APPENDIX I. 23 Technologies Coverage 30 in IEEE Publications...... 134 CONTENTS 1. INTRODUCTION...... 5 2. SEAMLESS INTELLIGENCE branch of physics and information theory and SCENARIO...... 8 enters into the realm of practical applications. As 3. 23 TECHNOLOGIES IN 2022...... 14 commercial QC comes within reach, new break- 3.1 Security Cross-Cutting Issues...... 14 throughs are occurring at an accelerating pace. 3.2 The Open Intellectual There is now evidence that QC can revolutionize Property Movement...... 17 crucial areas from chemistry that could have a 3.3 Sustainability...... 20 3.4 Massively Online Open Courses...... 25 dramatic impact on drug design to data processing 3.5 Quantum Computing ...... 28 with its ability to efficiently analyze vast amounts 3.6 Device and Nanotechnology...... 31 of data. 3.7 3D Integrated Circuits...... 33 CONTENTS 3.6 Device and Nanotechnology 3.8 Universal Memory...... 38 3.5.7 References 3.9 Multicore...... 43 3.10 Photonics...... 47 [1] M. Nielsen, and I.L. Chuang, Quantum Computa- 3.6.1 State of the Art tion and Quantum Information, Cambridge Universi- 3.11 Networking and Interconnectivity... 52 MEMS and micromachines made from silicon are 3.12 Software-Defined Networks...... 55 ty Press, 2010. evolving into the nanotechnology field, where 3.13 High-Performance Computing (HPC) ...... 63 [2] D-Wave, “D-Wave Two Quantum Computer you might “imagine your life being saved by a 3.14 Cloud Computing...... 68 Selected for New Quantum Artificial Intelligence custom-designed medical machine made from 3.15 The Internet of Things...... 74 Initiative, System to be Installed at NASA’s Ames particles 50,000 times as small as a single strand 3.16 Natural User Interfaces...... 77 Research Center, and Operational in Q3,” 2013; of your hair” [GT-nanotech]. More generally, nan- 3.17 3D Printing ...... 81 http://www.dwavesys.com/en/pressreleases. otechnology is about manipulating systems at the 3.18 Big Data and Analytics ...... 84 html#dwaveus_Google_NASA. level of atoms, molecules, and larger structures. 3.19 Machine Learning and Popular depictions and current technology are Intelligent Systems ...... 89 [3] Saracco, Personal Communication. 3.20 Computer Vision and Pattern showing the capability of rearranging atoms on a Recognition...... 92 silicon substrate to spell a word or of moving them 3.21 Life Sciences...... 96 around to show a sketch or cartoon. 3.22 Computational Biology and Bioinformatics...... 102 A wide range of science and engineering fields 3.23 Medical Robotics...... 108 pursue nanotechnology, including biology and 4. DRIVERS AND DISRUPTORS...... 112 medicine, physics, chemistry, materials science, 5. TECHNOLOGY COVERAGE IN and other engineering disciplines. Nanotechnology IEEE XPLORE AND BY is appearing in products like sunscreens and make- IEEE SOCIETIES...... 115 up, in automobile tires, and in vaccines. There are 6. IEEE COMPUTER SOCIETY already cameras that can be swallowed (at least in IN 2022...... 120 the lab) and/or digested. 7. SUMMARY AND NEXT STEPS.....124 8. AUTHORS...... 126 APPENDIX I. 23 Technologies Coverage 31 in IEEE Publications...... 134 CONTENTS 1. INTRODUCTION...... 5 2. SEAMLESS INTELLIGENCE 3.6.2 Challenges and Opportunities potential unknown interactions with the human SCENARIO...... 8 Medical applications of nanotechnology appear body and in terms of external influences able to 3. 23 TECHNOLOGIES IN 2022...... 14 to hold the most immediate promise for future use them for damaging rather than repairing hu- 3.1 Security Cross-Cutting Issues...... 14 computing environments: think of millions of ex- man bodies. The popular press has already taken 3.2 The Open Intellectual up this issue, worrying about nanoparticles enter- Property Movement...... 17 tremely tiny sensors and actuators pervading some ing the food chain or nanoparticles in sunscreen 3.3 Sustainability...... 20 environment, like a human body under study, to 3.4 Massively Online Open Courses...... 25 understand and then fix it. Science fiction stories interacting with the human body. There are also 3.5 Quantum Computing ...... 28 frequently raise the specter of self-healing bodies, entirely different issues, as when nanoparticles are 3.6 Device and Nanotechnology...... 31 where nanotech quickly heals a wound or rebuilds used in advanced materials that raise entirely new 3.7 3D Integrated Circuits...... 33 entire structures such as bones or organs. challenges, such as the meaning of “metal fatigue” CONTENTS3.8 Universal Memory...... 38 when metals are reinforced with nanoparticles. 3.9 Multicore...... 43 The state of the art remains far from the ac- Can we still predict fatigue for such materials? 3.10 Photonics...... 47 tive nanotech envisioned in sci-fi stories, but by 3.11 Networking and Interconnectivity... 52 2020, we will likely see an increased use of nano- 3.6.3 What Will Likely Happen 3.12 Software-Defined Networks...... 55 tech-based devices in controlled settings. In medi- 3.13 High-Performance Computing cine, swallowing little pills containing cameras may The use of nanodevices and nanoparticles has (HPC) ...... 63 shown great promise (and profit, e.g., in sunscreen 3.14 Cloud Computing...... 68 well be routine parts of office visits, with digestive or makeup) in many fields. It remains unclear, 3.15 The Internet of Things...... 74 processes removing them after some time, but 3.16 Natural User Interfaces...... 77 will we have injections of nanotech into our blood- however, to what extent and in what fields the dire 3.17 3D Printing ...... 81 stream, to, say, better map the heart and the blood visions painted in some of the sci-fi literature of 3.18 Big Data and Analytics ...... 84 vessels connected to it or to trace blood vessels in “smart” nanomachines running amok will be real- 3.19 Machine Learning and the brain? Perhaps, but given the long lead times ized. Certainly, current computing technologies still Intelligent Systems ...... 89 operate at length scales much larger than those 3.20 Computer Vision and Pattern for safe medical technologies, we will not see nan- Recognition...... 92 otech that “cleans up” those blood vessels, remov- of the nanoparticles used in current applications. 3.21 Life Sciences...... 96 ing debris, or that fights invasive organisms. On the Thus, it is really MEMS devices that have the ca- 3.22 Computational Biology and other hand, it will be possible to manufacture such pability of becoming increasingly smart and so- Bioinformatics...... 102 nanotech devices, creating a vision of millions of phisticated in their actions. Studies have begun on 3.23 Medical Robotics...... 108 devices concerned with a single human body and nanoparticles’ possible effects on humans and the 4. DRIVERS AND DISRUPTORS...... 112 their use in many natural and man-made settings. ecosystem, but there will be a need for longitudinal 5. TECHNOLOGY COVERAGE IN studies, going beyond the short-term investiga- IEEE XPLORE AND BY There are many challenges in realizing the nano- tions already being carried out. There has not yet IEEE SOCIETIES...... 115 tech visions articulated above. There are ethical been widespread popular opposition, in contrast 6. IEEE COMPUTER SOCIETY and privacy issues concerning “constant monitor- with the artificially induced changes in DNA that IN 2022...... 120 ing” by millions of tiny devices. There are dangers give rise to new plants banned in many countries. 7. SUMMARY AND NEXT STEPS.....124 from long-lived nanoparticles, both in terms of Whether there will be applications or usage models 8. AUTHORS...... 126 APPENDIX I. 23 Technologies Coverage 32 in IEEE Publications...... 134 CONTENTS 1. INTRODUCTION...... 5 2. SEAMLESS INTELLIGENCE of nanoparticles that give rise to such opposition SCENARIO...... 8 and ensuing legal or governmental actions remains 3. 23 TECHNOLOGIES IN 2022...... 14 unclear. Whether MEMS devices will find common 3.1 Security Cross-Cutting Issues...... 14 application in medical and other areas by 2022 also 3.2 The Open Intellectual remains unclear, in part because of the lengthy Property Movement...... 17 processes involved in launching new medical 3.3 Sustainability...... 20 3.4 Massively Online Open Courses...... 25 technologies. 3.5 Quantum Computing ...... 28 Nanotech is frequently described in terms of “very 3.6 Device and Nanotechnology...... 31 small.” But another important aspect is that nan- 3.7 3D Integrated Circuits...... 33 otech is a different way of creating materials, from CONTENTS3.8 Universal Memory...... 38 bottom up, as Mother Nature does, which is quite 3.9 Multicore...... 43 3.7 3D Integrated Circuits 3.10 Photonics...... 47 different from creating something top down. When 3.11 Networking and Interconnectivity... 52 something is manufactured in a top-down way, the 3.7.1 Introduction 3.12 Software-Defined Networks...... 55 original physical characteristics are not altered, but The desire to overcome the memory bottleneck 3.13 High-Performance Computing in manufacturing bottom up, one can design spe- (HPC) ...... 63 caused by pin issues in planar circuits, along with cific material characteristics [Saracco]. 3.14 Cloud Computing...... 68 the skyrocketing foundry costs of leading-edge 3.15 The Internet of Things...... 74 process designs, have fueled the development of 3.16 Natural User Interfaces...... 77 3.6.4 Summary stacked 2.5D and 3D chips over the last few years 3.17 3D Printing ...... 81 It is clear that MEMS devices, nanoparticles, and [Kni12]. 3.18 Big Data and Analytics ...... 84 their use in a broad set of applications are here to 3.19 Machine Learning and While the trend toward aggressive single-chip, Intelligent Systems ...... 89 stay, as the opportunities arising from their use are simply too numerous to ignore. It will be interesting, SoC-level integration continues, some forces are 3.20 Computer Vision and Pattern also pulling in the opposite direction. A monolithic Recognition...... 92 however, to watch their evolution. 3.21 Life Sciences...... 96 SoC is constrained to a single silicon process, can- 3.22 Computational Biology and not cope with mixed signal components, and has Bioinformatics...... 102 3.6.5 References its volume economics reconciled with the nonlinear 3.23 Medical Robotics...... 108 [Saracco] Saracco, R., Personal Communication. growth of NRE with complexity. The lack of cost-ef- 4. DRIVERS AND DISRUPTORS...... 112 fective lithographic solutions is slowing down raw 5. TECHNOLOGY COVERAGE IN process scaling, and rapidly increasing volumes are IEEE XPLORE AND BY required to absorb the design costs for each new IEEE SOCIETIES...... 115 process node, limiting the number of products that 6. IEEE COMPUTER SOCIETY can be manufactured. Finally, several critical per- IN 2022...... 120 formance factors are shifting away from single-die 7. SUMMARY AND NEXT STEPS.....124 CMOS scaling to system-level considerations, such 8. AUTHORS...... 126 APPENDIX I. 23 Technologies Coverage 33 in IEEE Publications...... 134 CONTENTS 1. INTRODUCTION...... 5 2. SEAMLESS INTELLIGENCE SCENARIO...... 8 3. 23 TECHNOLOGIES IN 2022...... 14 3.1 Security Cross-Cutting Issues...... 14 3.2 The Open Intellectual Property Movement...... 17 3.3 Sustainability...... 20 3.4 Massively Online Open Courses...... 25 3.5 Quantum Computing ...... 28 3.6 Device and Nanotechnology...... 31 3.7 3D Integrated Circuits...... 33 Figure 3. Two integration scenarios: a 2.5D component using a silicon interposer (left) and a full 3D stack using TSVs (right). CONTENTS3.8 Universal Memory...... 38 3.9 Multicore...... 43 as breaking the “memory wall” and providing more 3.7.2 State of the Art 3.10 Photonics...... 47 efficient paths to I/O. 3.11 Networking and Interconnectivity... 52 In simple terms, a 3D-IC (also commonly called 3.12 Software-Defined Networks...... 55 Several factors are pushing stacking technologies system-in-package, or SiP), can be seen as 3.13 High-Performance Computing to the mainstream, especially the desire to increase the modern incarnation of a multichip module (HPC) ...... 63 density (for volume-conscious products), to both (MCM). The two dominant flavors are PiP (pack- 3.14 Cloud Computing...... 68 age-in-package, mounted side by side) and PoP 3.15 The Internet of Things...... 74 decrease cost and power and increase perfor- 3.16 Natural User Interfaces...... 77 mance [Ark12]. By using several smaller dies, stack- (package-on-package, mounted on top of one 3.17 3D Printing ...... 81 ing can enhance the single-die yield versus building another). A variety of substrates are currently used 3.18 Big Data and Analytics ...... 84 a single large SoC. It can also avoid the capital cost in the industry [Woy13], ranging from laminates 3.19 Machine Learning and recovery of the large NRE of a complex SoC de- (similar to FR4 boards, supporting 5 to 25 layers) to Intelligent Systems ...... 89 sign. And, of course, it reduces the bill-of-material ceramics (capable of hundreds of layers) to glass or 3.20 Computer Vision and Pattern metal covered with a layer of dielectric (around 5 Recognition...... 92 (BOM) through the integration of multiple ICs in the 3.21 Life Sciences...... 96 same component. On the power dimension, the use layers) to semiconductors. 3.22 Computational Biology and of local low-power connections reduces the need Relative to a single-die SoC, even a simple 2D SiP Bioinformatics...... 102 of a large number of external power-hungry inter- can provide several advantages, such as the possi- 3.23 Medical Robotics...... 108 connects and PHYs, especially for memory. Using bility to mix signals, optimize the best technology 4. DRIVERS AND DISRUPTORS...... 112 separate dies also enables adoption of separate process for each die, couple the IPs of different 5. TECHNOLOGY COVERAGE IN silicon processes that can be power-optimized for vendors, and offer greater flexibility with derivative IEEE XPLORE AND BY a specific function. Finally, performance improves designs using a different component mix. IEEE SOCIETIES...... 115 due to the increased interconnect speed related 6. IEEE COMPUTER SOCIETY to the wire lengths (short-wide are faster than The introduction of silicon interposers is what the IN 2022...... 120 long-narrow wires), and the power recovery in in- industry refers to as 2.5D integration (Figure 3). 7. SUMMARY AND NEXT STEPS.....124 terconnects helps offset the “dark silicon” problem. It can support very fine tracks, enable active or 8. AUTHORS...... 126 APPENDIX I. 23 Technologies Coverage 34 in IEEE Publications...... 134 CONTENTS 1. INTRODUCTION...... 5 2. SEAMLESS INTELLIGENCE passive configurations, and have mechanical prop- SCENARIO...... 8 erties (such as the coefficient of thermal expan- 3. 23 TECHNOLOGIES IN 2022...... 14 sion) that match the individual silicon slices. The 3.1 Security Cross-Cutting Issues...... 14 2.5D packaging technologies provide tremendous 3.2 The Open Intellectual increase in capacity and performance. While flip- Property Movement...... 17 chip bumps are around 100 µm, the micro bumps 3.3 Sustainability...... 20 3.4 Massively Online Open Courses...... 25 that connect 2.5D dies onto a Si interposer can be 3.5 Quantum Computing ...... 28 shrunk to about 10 µm. The interposer itself can be 200- to 700-µm thick and contain metal layer Figure 4. An integration scenario combining 2.5D integration 3.6 Device and Nanotechnology...... 31 of multiple 3D-stacked components. 3.7 3D Integrated Circuits...... 33 tracks (created using a standard Si process) and CONTENTS3.8 Universal Memory...... 38 thru-silicon-vias (TSVs) that enable efficient con- full heterogeneous 3D (i.e., not just single-vendor 3.9 Multicore...... 43 nections between the upper layer and the package memory chips) becomes mainstream. As of 2013, 3.10 Photonics...... 47 bumps. For example, the Xilinx Virtex-7 2000T volume production of 2.5D and 3D SoCs has pri- 3.11 Networking and Interconnectivity... 52 device supports about 10,000 silicon-speed con- marily used a “turnkey” approach, in which a single 3.12 Software-Defined Networks...... 55 3.13 High-Performance Computing nections between adjacent slices. vertically integrated company provides both the (HPC) ...... 63 The most complex form of SiP co-packaging is full front end (design of the individual ICs) and back 3.14 Cloud Computing...... 68 3D integration, which can also be combined with end (testing and assembly) of the final part. The 3.15 The Internet of Things...... 74 alternative “hybrid” approach, where the foundries 3.16 Natural User Interfaces...... 77 2.5D integration to create very elaborate configurations (Figure 4). Unlike 2.5D, 3D integration di- deal with the front end and the packager with the 3.17 3D Printing ...... 81 back end, is constrained to niche specialty parts to 3.18 Big Data and Analytics ...... 84 rectly connects multiple Si slices with TSVs etched date, mostly because of the ecosystem complexity 3.19 Machine Learning and in the dies themselves. This provides superior Intelligent Systems ...... 89 integration (no need for micro bumps) and higher of dealing with multivendor solutions. 3.20 Computer Vision and Pattern Recognition...... 92 interconnect density with TSVs (about 5 µm). For 3D-IC to evolve beyond vertical developments, 3.21 Life Sciences...... 96 all players in the ecosystem must find a way to 3.22 Computational Biology and 3.73 Challenges work in a cost-effective manner. This implies Bioinformatics...... 102 Major challenges remain in a 3D hybrid ecosystem, providing fast turnaround times, defining a clear 3.23 Medical Robotics...... 108 such as testing, dealing with the business aspect of separation of responsibilities, and defining manu- 4. DRIVERS AND DISRUPTORS...... 112 compounding multi-die yields, and above all, man- facturing and supply-chain roles. 5. TECHNOLOGY COVERAGE IN agement of the supply chain. Several players, such Dissipating the heat building up within the 3D-IC IEEE XPLORE AND BY IEEE SOCIETIES...... 115 as Xilinx, Altera, Cisco, Huawei, and IBM, openly is a major technical challenge, especially because discuss their 2.5D and 3D roadmaps. However, these multiple high-speed components are placed 6. IEEE COMPUTER SOCIETY IN 2022...... 120 it will take some time before 3D-ICs can reach in such a small physical proximity. New heat ex- mass production and at least until 2015 before 7. SUMMARY AND NEXT STEPS.....124 traction technologies are required, especially to 8. AUTHORS...... 126 APPENDIX I. 23 Technologies Coverage 35 in IEEE Publications...... 134 CONTENTS 1. INTRODUCTION...... 5 2. SEAMLESS INTELLIGENCE manage multilayer thermal hotspots and deal dollars to create but no longer works. At the foun- SCENARIO...... 8 with the intermediate layers far from the package dation of the problem is the yield-compounding 3. 23 TECHNOLOGIES IN 2022...... 14 boundaries and heat sinks. issue of multi-die packages, where a large stack 3.1 Security Cross-Cutting Issues...... 14 TSVs are large compared to other silicon structures can be “killed” by a single bad die. At 99 percent 3.2 The Open Intellectual yield (of a simple die) and 5 layers, the compound Property Movement...... 17 (50 to 100 gates): placing them has significant yield goes down to 95 percent, which may still be 3.3 Sustainability...... 20 impact on chip-floor planning. Manufacturability acceptable. Stacking 8 larger dies, like CPU or 3.4 Massively Online Open Courses...... 25 requirements for landing pads and keep-out zones 3.5 Quantum Computing ...... 28 result in placement obstacles. Because TSVs oc- memory, with a 95 percent yield results in a 66 per- 3.6 Device and Nanotechnology...... 31 cupy the metal layers, they also result in additional cent yield, so 5 percent bad becomes 33 percent 3.7 3D Integrated Circuits...... 33 routing obstacles. bad before any assembly loss. Even worse, if the CONTENTS3.8 Universal Memory...... 38 individual layers have different value, a bad $1 die 3.9 Multicore...... 43 Stacking could create effects that were never (e.g., a DRAM layer) can make a $1,000 stack (e.g., 3.10 Photonics...... 47 considered before, and signal integrity challenges a complex CPU) into a keychain. 3.11 Networking and Interconnectivity... 52 emerge when dealing with die-to-die intercon- 3.12 Software-Defined Networks...... 55 nects, shrinking wires and RC delays, unpredict- 3.13 High-Performance Computing 3.7.4 Where We Think It Will Go able electro-migration, and so on. With the added (HPC) ...... 63 Ultra-mobile and mobile products are already the 3.14 Cloud Computing...... 68 complication of multipatterning, stress effects, and first adopters of 3D-IC technology, but across the 3.15 The Internet of Things...... 74 process variations, new design flows will become entire spectrum of IT products, we are increasingly 3.16 Natural User Interfaces...... 77 imperative to address some of these issues. 3.17 3D Printing ...... 81 observing a disruptive transition from printed cir- 3.18 Big Data and Analytics ...... 84 Separate testing of the independent layers is es- cuit boards to 3D-ICs, and packaging will increas- 3.19 Machine Learning and sential to keep yield issues under control. With full ingly play a pivotal role in being able to provide Intelligent Systems ...... 89 3D structures, all but the first and last die are hid- value and differentiation. 3.20 Computer Vision and Pattern den, leaving no way to contact the stacked die for Recognition...... 92 Despite the technology and business challenges, testing. Contact of test probes to thinner “naked” 3.21 Life Sciences...... 96 we expect that over the next five years, 2.5D and dies increases the probability of mechanical stress 3.22 Computational Biology and 3D will become increasingly commonplace. Beyond Bioinformatics...... 102 and fractures. For 2.5D integration, some of these mobile products, other cost- and energy-sensi- 3.23 Medical Robotics...... 108 issues are smaller, and constraining the TSVs to a tive areas include the hyperscale server market, silicon passive interposer eliminates the mechani- 4. DRIVERS AND DISRUPTORS...... 112 networking and storage products, and a variety of cal stress problem for active transistors. 5. TECHNOLOGY COVERAGE IN embedded applications, such as sensors. IEEE XPLORE AND BY Finally, the fundamental challenge of multivendor IEEE SOCIETIES...... 115 Additional emerging technologies claim even bet- 3D-ICs is not technology or cost per transistor: 6. IEEE COMPUTER SOCIETY ter properties than TSV-based 3D. For example, the it’s who takes responsibility when something goes IN 2022...... 120 so-called “monolithic 3D” wafer-scale integration wrong in a chip that costs several hundred million 7. SUMMARY AND NEXT STEPS.....124 uses “patterned vias,” about 50-nm wide, which 8. AUTHORS...... 126 APPENDIX I. 23 Technologies Coverage 36 in IEEE Publications...... 134 CONTENTS 1. INTRODUCTION...... 5 2. SEAMLESS INTELLIGENCE SCENARIO...... 8 3. 23 TECHNOLOGIES IN 2022...... 14 translates into more than 10,000x higher vertical Combining SoC integration and co-packaging 3.1 Security Cross-Cutting Issues...... 14 connections than TSV. It also uses a 100-nm thick will help the continued scaling of power-efficient 3.2 The Open Intellectual silicon layer and yields a total reduction of 3x in system performance, while enabling each die to be Property Movement...... 17 Si area and 12x in chip footprint (a standard wafer made in an optimized process node and enabling 3.3 Sustainability...... 20 with 8 to 9 metal layers could be 1-µm thick). We the design re-use of individual dies across multiple 3.4 Massively Online Open Courses...... 25 expect these (or other) technologies to mature by products. Co-packaging logic and memory dies 3.5 Quantum Computing ...... 28 3.6 Device and Nanotechnology...... 31 2022. can break the memory wall by using short-length, 3.7 3D Integrated Circuits...... 33 low-capacitance, and wide interconnects. Because CONTENTS3.8 Universal Memory...... 38 3.7.5 Disruptions of the nonlinear relationship of complexity and 3.9 Multicore...... 43 As with any major technology shift, 3D-ICs will NRE, the cost of a very complex chip could also be 3.10 Photonics...... 47 reduced by co-packaging two smaller dies with half 3.11 Networking and Interconnectivity... 52 have a significant disruptive effect on the entire breadth of IT products, from mobile devices the functionality (for example, building a 16-core 3.12 Software-Defined Networks...... 55 CPU out of a pair of 8-core dies). Different technol- 3.13 High-Performance Computing to enterprise servers. This technology will pose (HPC) ...... 63 significant new threats to established players by ogies are at play here, progressing from SoC (same 3.14 Cloud Computing...... 68 fundamentally changing the supply-chain flows of die) to SiP (multiple dies on interposer, or 2.5D) to 3.15 The Internet of Things...... 74 important components, such as DRAM and CPUs. full die stacking (multiple dies with TSVs, or 3D). 3.16 Natural User Interfaces...... 77 It will also create new business opportunities for 3.17 3D Printing ...... 81 3.7.7 References 3.18 Big Data and Analytics ...... 84 the industry related to managing the co-packaging 3.19 Machine Learning and of IP blocks from the 3D ecosystem. [Ark12] S. Arkalgud, “2.5D/3D Scaling Walls (presenta- Intelligent Systems ...... 89 tion),” IMAPS Device Packaging Conf., 2013; http://www. Because of the business challenges of multivendor 3.20 Computer Vision and Pattern invensas.com/Company/Documents/Invensas_IMAPS- Recognition...... 92 3D-ICs, we also expect a significant push toward DevicePkg2013_Keynote25D3DScalingWalls.pdf. 3.21 Life Sciences...... 96 vertically integrated products, where new or es- [Wy13] C. Woychik, “Emerging 3D and TSV Packaging 3.22 Computational Biology and tablished players will act as catalysts to integrate Technology (presentation),” SMTA Int’l, 2013; http:// Bioinformatics...... 102 complex 3D-ICs by leveraging a large portfolio of 3.23 Medical Robotics...... 108 www.invensas.com/Company/Documents/Invensas_ IP blocks (or dies) that will appear in the next few SMTA2013_Emerging3DandTSVPackaging.pdf. 4. DRIVERS AND DISRUPTORS...... 112 years. 5. TECHNOLOGY COVERAGE IN [Kni12] J.U. Knickerbocker et al., “2.5D and 3D Tech- IEEE XPLORE AND BY nology Challenges and Test Vehicle Demonstrations,” IEEE SOCIETIES...... 115 3.7.6 Summary 62nd IEEE Electronic Components and Technology Conf. 6. IEEE COMPUTER SOCIETY The expectation in the semiconductor industry is (ECTC), 2012, pp. 1068-1076. IN 2022...... 120 that multi-die co-packaging will be a steady and 7. SUMMARY AND NEXT STEPS.....124 rapidly growing trend to address these concerns. 8. AUTHORS...... 126 APPENDIX I. 23 Technologies Coverage 37 in IEEE Publications...... 134 CONTENTS 1. INTRODUCTION...... 5 2. SEAMLESS INTELLIGENCE SCENARIO...... 8 3. 23 TECHNOLOGIES IN 2022...... 14 form of nonvolatile “universal” memory (NVM) will 3.1 Security Cross-Cutting Issues...... 14 replace DRAM [Xie10]. While it is difficult to predict 3.2 The Open Intellectual exactly when and how, we believe that this transi- Property Movement...... 17 tion is inevitable, and there are already signs of it 3.3 Sustainability...... 20 happening today. This new “universal memory” will 3.4 Massively Online Open Courses...... 25 combine the fast random access characteristics 3.5 Quantum Computing ...... 28 3.8 Universal Memory 3.6 Device and Nanotechnology...... 31 of DRAM and the nonvolatility properties of Flash. 3.7 3D Integrated Circuits...... 33 3.8.1 Introduction As such, it will have the potential to replace a large CONTENTS3.8 Universal Memory...... 38 The next five to seven years will cause very signif- fraction of the memory and storage hierarchy, 3.9 Multicore...... 43 icant shifts to the IT infrastructure, and we believe and consequently cause a tectonic shift in archi- 3.10 Photonics...... 47 that memory and processor architectures are two tectures and the corresponding software to take 3.11 Networking and Interconnectivity... 52 advantage of it [Ran11]. 3.12 Software-Defined Networks...... 55 areas that will change profoundly. We focus here 3.13 High-Performance Computing on memory. (HPC) ...... 63 Because of the charge retention issues and man- 3.14 Cloud Computing...... 68 3.15 The Internet of Things...... 74 ufacturability challenges dictated by the laws of 3.16 Natural User Interfaces...... 77 physics (Figure 5), and despite manufacturers’ 3.17 3D Printing ...... 81 heroic efforts to continue scaling, DRAM’s end is in 3.18 Big Data and Analytics ...... 84 sight [Mut13]. DRAM has had a remarkable lifespan 3.19 Machine Learning and of over 40 years, starting in the late 1960s when Intelligent Systems ...... 89 it was invented and then manufactured in 1970 by 3.20 Computer Vision and Pattern Recognition...... 92 Intel. It has scaled in capacity by a factor of over 3.21 Life Sciences...... 96 8 million, from 1 kbits on a die in 1970 to 8 Gbits 3.22 Computational Biology and today. From this perspective, DRAM’s capacity Bioinformatics...... 102 has been one of the more consistent incarnations 3.23 Medical Robotics...... 108 of Moore’s law and has become one of the foun- 4. DRIVERS AND DISRUPTORS...... 112 dational commodities of the entire IT industry. 5. TECHNOLOGY COVERAGE IN Notwithstanding DRAM’s incredible success, the IEEE XPLORE AND BY number of memory manufacturers has been steadi- Figure 5. Illustration of the severity of the DRAM capacitor IEEE SOCIETIES...... 115 ly decreasing: 20 in 1985, 11 in 1995, 8 in 2007, and “trench.” On the left is a schematic representation of the as- 6. IEEE COMPUTER SOCIETY only 3 today. pect ratio of a DRAM cell in 3x nm process, showing a “trench” IN 2022...... 120 aspect ratio (depth/aperture width) of over 25x. On the right is 7. SUMMARY AND NEXT STEPS.....124 Between now and 2022, we expect that a new the actual silicon cross section of two DRAM cells. 8. AUTHORS...... 126 APPENDIX I. 23 Technologies Coverage 38 in IEEE Publications...... 134 CONTENTS 1. INTRODUCTION...... 5 2. SEAMLESS INTELLIGENCE SCENARIO...... 8 3. 23 TECHNOLOGIES IN 2022...... 14 3.1 Security Cross-Cutting Issues...... 14 3.2 The Open Intellectual Property Movement...... 17 3.3 Sustainability...... 20 3.4 Massively Online Open Courses...... 25 3.5 Quantum Computing ...... 28 3.6 Device and Nanotechnology...... 31 3.7 3D Integrated Circuits...... 33 CONTENTS3.8 Universal Memory...... 38 Figure 6. Simplified phase-change memory (PCM) cell (left) and spin-transfer torque (STT) cell (right). 3.9 Multicore...... 43 3.10 Photonics...... 47 3.8.2 State of the Art covering here), and have in common that the state 3.11 Networking and Interconnectivity... 52 is defined by differences in the cell resistance (un- 3.12 Software-Defined Networks...... 55 The two most visible, expected, and desired met- like DRAM, which stores charge). 3.13 High-Performance Computing rics for memory are capacity (bits per device) and (HPC) ...... 63 cost per bit. To keep advancing these metrics, Of the NVM technologies, at least one (memristor) 3.14 Cloud Computing...... 68 DRAM manufacturers are resorting to 2.5D/3D appears to offer a substantial greater bit density 3.15 The Internet of Things...... 74 packaging and stacking techniques such as the than DRAM [Rib12]. This comes from two factors: 3.16 Natural User Interfaces...... 77 Hybrid Memory Cube (HMC) from Micron [Paw11]. 3.17 3D Printing ...... 81 they are “crossbar” memories that do not need an 3.18 Big Data and Analytics ...... 84 The additional manufacturing steps and yield loss isolation device per cell (leading to greater planar 3.19 Machine Learning and for this type of memory device increases the price density), and they can be layered (on the same Intelligent Systems ...... 89 per bit, but may be able to keep the capacity scal- piece of silicon) as multiple planes for increased 3.20 Computer Vision and Pattern ing growth for a few more years. “effective” planar density. This is in addition to the Recognition...... 92 3.21 Life Sciences...... 96 In parallel, the memory industry has been actively stacking of dies within a 3D package. 3.22 Computational Biology and developing possible replacement technologies for Bioinformatics...... 102 DRAM, all of which are flavors of NVM. A survey 3.83 Challenges 3.23 Medical Robotics...... 108 of literature, patents, and manufacturers’ disclo- DRAM holds state through electric charge, but 4. DRIVERS AND DISRUPTORS...... 112 sures indicates three technologies as the lead- the shrinking of silicon structures has reduced the 5. TECHNOLOGY COVERAGE IN ing contenders: STT-RAM (spin-transfer torque physical size of the cell capacitors to a point that IEEE XPLORE AND BY RAM)[Mos05], PCM (phase-change memory) makes it challenging to retain the charge. This IEEE SOCIETIES...... 115 [Rao08,Lee09], and Memristor [Stru08]. They all is made even worse by the increasingly thinner 6. IEEE COMPUTER SOCIETY have significant investments by the major mem- IN 2022...... 120 insulation layer of deep submicron semiconductor ory manufacturers, show a different balance of processes. The industry is already starting to see 7. SUMMARY AND NEXT STEPS.....124 advantages and disadvantages (which we are not 8. AUTHORS...... 126 APPENDIX I. 23 Technologies Coverage 39 in IEEE Publications...... 134 CONTENTS 1. INTRODUCTION...... 5 2. SEAMLESS INTELLIGENCE scaling-related issues with SCENARIO...... 8 DRAM, which is causing 3. 23 TECHNOLOGIES IN 2022...... 14 quality issues; there ap- 3.1 Security Cross-Cutting Issues...... 14 pears to be no remedy. 3.2 The Open Intellectual Property Movement...... 17 A second big DRAM issue is 3.3 Sustainability...... 20 the ability to manufacture 3.4 Massively Online Open Courses...... 25 cells at increasingly smaller 3.5 Quantum Computing ...... 28 semiconductor nodes. The 3.6 Device and Nanotechnology...... 31 aspect ratio (depth versus 3.7 3D Integrated Circuits...... 33 opening) of the “trench” CONTENTS3.8 Universal Memory...... 38 3.9 Multicore...... 43 used to construct the 3.10 Photonics...... 47 DRAM cell already has an Figure 7. Simplified memristor (ReRAM) cell. 3.11 Networking and Interconnectivity... 52 aspect ratio of about 25x. 3.12 Software-Defined Networks...... 55 As the semiconductor node 3.8.4 Where We Think It Will Go 3.13 High-Performance Computing gets finer, the aperture (surface opening) reduces We expect at least one NVM technology to reach (HPC) ...... 63 in size, making cell manufacturing increasingly dif- 3.14 Cloud Computing...... 68 maturity and volume manufacturing capabilities ficult. Because the trench volume cannot be made 3.15 The Internet of Things...... 74 within the next three to five years. The first place 3.16 Natural User Interfaces...... 77 arbitrarily small (it determines capacitance), DRAM where NVM will materialize is most likely going to 3.17 3D Printing ...... 81 manufacturability will be harder and harder. be today’s primary storage, top-tier layer, where 3.18 Big Data and Analytics ...... 84 The top NVM challenge is not related to technol- NVM’s superior properties will gradually replace 3.19 Machine Learning and ogy, but business. Whereas DRAM has had an NAND and NOR Flash. This will happen across the Intelligent Systems ...... 89 board, from mobile client devices to high-end en- 3.20 Computer Vision and Pattern industry-wide consistency and commonality over Recognition...... 92 time, we expect a much wider set of generational terprise storage products. The SNIA NVM Program- 3.21 Life Sciences...... 96 and manufacturer differences in NVM. As a con- ming Technical Working Group is already actively 3.22 Computational Biology and sequence, no individual NVM technology will likely working on developing a new NVM Programming Bioinformatics...... 102 have the longevity of DRAM. This is especially true Model so that hardware and software vendors can 3.23 Medical Robotics...... 108 as the technologies climb the maturity curve, which align their effort behind a well-defined standard 4. DRIVERS AND DISRUPTORS...... 112 means that there may be a two-stage market for that presents a simple and consistent method of 5. TECHNOLOGY COVERAGE IN the DRAM replacement: a given media may get exposing persistent memory to applications. IEEE XPLORE AND BY IEEE SOCIETIES...... 115 to market first but lack all the features relative to In parallel, we also expect NVM to appear in the a later arrival, and the important market players 6. IEEE COMPUTER SOCIETY main memory space, initially as a memory exten- IN 2022...... 120 could switch to the better media. sion, similar to the efforts already appearing today 7. SUMMARY AND NEXT STEPS.....124 that combine Flash and DRAM [Diablo Memory 8. AUTHORS...... 126 APPENDIX I. 23 Technologies Coverage 40 in IEEE Publications...... 134 CONTENTS 1. INTRODUCTION...... 5 2. SEAMLESS INTELLIGENCE Channel]. This is when the real disruption will involves fighting the inertia in existing legacy con- SCENARIO...... 8 occur, since the presence of NVM in the memory straints. While a complete replacement would be 3. 23 TECHNOLOGIES IN 2022...... 14 space has the potential to fundamentally change the best technical solution, market forces and in- 3.1 Security Cross-Cutting Issues...... 14 the way in which we persist information in the ertia will resist and possibly delay adoption beyond 3.2 The Open Intellectual storage layer: we no longer need to think in terms our prediction. Property Movement...... 17 of serialization and deserialization. 3.3 Sustainability...... 20 3.4 Massively Online Open Courses...... 25 As NVM universal memory appears, there will be 3.8.5 Disruptions 3.5 Quantum Computing ...... 28 consequences in other components of the com- The availability of a much larger byte-addressable 3.6 Device and Nanotechnology...... 31 pute infrastructure, the first being the memory and persistent physical memory will cause a major 3.7 3D Integrated Circuits...... 33 controller. Since about 2005, memory controllers re-thinking and re-architecting of end user applica- CONTENTS3.8 Universal Memory...... 38 3.9 Multicore...... 43 have been integrated with the CPU chip, so that tions and algorithms, as well as the operating sys- 3.10 Photonics...... 47 the memory itself (DRAM and DIMMs) is no more tem components (such as file systems and object 3.11 Networking and Interconnectivity... 52 than a passive slave to the microprocessor. We ex- stores) that are more closely related to storage. 3.12 Software-Defined Networks...... 55 pect that this will change with NVM: because of the The nonvolatility of a large physical memory of- 3.13 High-Performance Computing technology and architecture variability, the most fers several benefits to application writers. At a (HPC) ...... 63 logical evolution will involve breaking NVM access 3.14 Cloud Computing...... 68 certain point in size, NVM can be viewed and used functionality into a high-level asynchronous pro- 3.15 The Internet of Things...... 74 as the union of “memory” and “storage.” Today’s tocol controller (which remains with the CPU) and 3.16 Natural User Interfaces...... 77 hard disks have very high random-access latency, 3.17 3D Printing ...... 81 a low-level media controller (integrated as part of relatively low bandwidth, access via I/O calls that 3.18 Big Data and Analytics ...... 84 the memory system itself). This way, the high-level require a long code path and OS context switch- 3.19 Machine Learning and interface can only specify the “intent” (e.g., read, Intelligent Systems ...... 89 es, and block-based semantics. With sufficient write, copy requests), and the memory will be free persistent physical address space, a large fraction 3.20 Computer Vision and Pattern to optimize and re-order low-level operations to Recognition...... 92 of what today sits on a hard disk can be moved 3.21 Life Sciences...... 96 better match media properties. As these interfaces to NVM. The advantages are compelling: a short 3.22 Computational Biology and standardize over time, they will also allow for some load/store path and simplified random-access Bioinformatics...... 102 level of computation to move into the memory semantics to access file systems or object stores, 3.23 Medical Robotics...... 108 system itself. and possibly minimal or no OS involvement. Even 4. DRIVERS AND DISRUPTORS...... 112 Finally, it is important to observe that any incre- more importantly, data structures can be natively 5. TECHNOLOGY COVERAGE IN mental approach to remedying the current mem- persisted without the need for serializing them to a IEEE XPLORE AND BY IEEE SOCIETIES...... 115 ory ecosystem and accommodating NVM will disk-friendly block-based format—all with a band- be suboptimal and only delay the inevitable. The width comparable to memory and a latency several 6. IEEE COMPUTER SOCIETY IN 2022...... 120 best thing the industry can, and should, do is to orders of magnitude lower than today’s storage. fully re-architect the memory ecosystem, but that 7. SUMMARY AND NEXT STEPS.....124 For transactional applications that rely on high 8. AUTHORS...... 126 APPENDIX I. 23 Technologies Coverage 41 in IEEE Publications...... 134 CONTENTS 1. INTRODUCTION...... 5 2. SEAMLESS INTELLIGENCE IOPS rates (I/O operations per second), we expect Because NVM technologies combine the fast SCENARIO...... 8 NVM solutions to gradually start replacing today’s access patterns of DRAM and the persistence and 3. 23 TECHNOLOGIES IN 2022...... 14 caching and I/O acceleration solutions. While these capacity of disks, they will cause a collapsing of 3.1 Security Cross-Cutting Issues...... 14 are quite successful today, they require more pow- the memory-storage hierarchy that will permeate 3.2 The Open Intellectual er, yield lower performance, add maintenance com- all the way into the software we write across the Property Movement...... 17 plexity, and have additional points of failure relative board, from operating systems to middleware to 3.3 Sustainability...... 20 3.4 Massively Online Open Courses...... 25 to an in-memory file system or object store. applications. This will be a deep-reaching funda- 3.5 Quantum Computing ...... 28 Finally, the packaging of this large amount of mental, powerful, and beneficial change. 3.6 Device and Nanotechnology...... 31 memory requires further consideration. Due to the 3.7 3D Integrated Circuits...... 33 much lower power of NVM, we anticipate that the 3.8.7 References CONTENTS3.8 Universal Memory...... 38 [Hos05] M. Hosomi et al., “A Novel Nonvolatile Memory 3.9 Multicore...... 43 trend of stacking a large number of die slices within with Spin Torque Transfer Magnetization Switching: 3.10 Photonics...... 47 the same component will continue and accelerate. Spin-Ram,” IEDM IEEE Int’l Technical Digest, vol. 5, no. 5, 3.11 Networking and Interconnectivity... 52 Coupled with the increased silicon-level density 2005, pp. 459-462. 3.12 Software-Defined Networks...... 55 advantage of NVM media, we can expect line-of- 3.13 High-Performance Computing sight of 20 to 40x greater density per part, most [Lee09] B.C. Lee et al., “Architecting Phase Change (HPC) ...... 63 likely very different from today’s DIMMs and prob- Memory As a Scalable DRAM Alternative,” Proc. 36th 3.14 Cloud Computing...... 68 ably more similar to an evolution of the recently Int’l Symp. Computer Architecture (ISCA ’09), 2009, pp. 3.15 The Internet of Things...... 74 2-13. 3.16 Natural User Interfaces...... 77 proposed HMC 3D structure. 3.17 3D Printing ...... 81 [Mut13] O. Mutlu, “Memory Scaling: A Systems Archi- 3.18 Big Data and Analytics ...... 84 3.8.6 Summary tecture Perspective,” Proc. 5th Int’l Memory Workshop (IMW), 2013; http://users.ece.cmu.edu/~omutlu/pub/ 3.19 Machine Learning and As DRAM approaches its end of life, we are witness- Intelligent Systems ...... 89 mutlu_memory-scaling_memcon13_talk.pdf. 3.20 Computer Vision and Pattern ing the emergence of new NVM technologies that Recognition...... 92 have the potential to address DRAM’s scaling and [Paw11] J.T. Pawlowski, ‘‘Micron Hybrid Memory Cube 3.21 Life Sciences...... 96 capacity issues. We expect a gradual replacement will (HMC),’’ HotChips 23, 2011. 3.22 Computational Biology and occur between now and 2022. These new NVM tech- [Ran11] P. Ranganathan et al., “From Microprocessors to Bioinformatics...... 102 nologies have a set of characteristics that will make Nanostores: Rethinking Data-Centric Systems,” Com- 3.23 Medical Robotics...... 108 them amenable to becoming a “universal” memory puter, vol. 44, no. 1, 2011, pp. 39-48. 4. DRIVERS AND DISRUPTORS...... 112 that takes over the entire hierarchy from main mem- [Rao08] S. Raoux et al., “Phase-Change Random Ac- 5. TECHNOLOGY COVERAGE IN ory to storage (or at least the top tier of storage). cess Memory: A Scalable Technology,” IBM J. Research IEEE XPLORE AND BY IEEE SOCIETIES...... 115 This will cause disk and Flash technology to move to and Development, vol. 52, no. 4/5, 2008. lower-level tiers, a transition of similar disruptive mag- 6. IEEE COMPUTER SOCIETY [Rib12] G.M. Ribeiro et al., “Designing Memristors: Phys- IN 2022...... 120 nitude to what happened when tape was ubiquitously ics, Materials Science and Engineering,” IEEE Int’l Symp. replaced by spinning hard disks. 7. SUMMARY AND NEXT STEPS.....124 Circuits and Systems (ISCAS), 2012, pp.2513-2516. 8. AUTHORS...... 126 APPENDIX I. 23 Technologies Coverage 42 in IEEE Publications...... 134 CONTENTS 1. INTRODUCTION...... 5 2. SEAMLESS INTELLIGENCE [Stru8] D.B. Strukov et al., “The Missing Memristor 1/16. Also, we should consider static power caused SCENARIO...... 8 Found,” Nature, vol. 453, no. 7191, 2008, pp. 80–83. by leakage current. To reduce static power, power 3. 23 TECHNOLOGIES IN 2022...... 14 [Xie10] Y. Xie, “Modeling, Architecture, and Applications gating on non-active parts is effective. In multicore, 3.1 Security Cross-Cutting Issues...... 14 for Emerging Memory Technologies,” IEEE Design and power gating can be applied to each small proces- 3.2 The Open Intellectual Test of Computers, Special Issues on Memory Technolo- sor core and its local memories. Property Movement...... 17 gies, 2010. 3.3 Sustainability...... 20 However, to obtain good multicore performance, 3.4 Massively Online Open Courses...... 25 software is key for decomposing an original se- 3.5 Quantum Computing ...... 28 quential program into parallel program parts and 3.6 Device and Nanotechnology...... 31 assigning them to processor cores, to minimize 3.7 3D Integrated Circuits...... 33 the execution time (including the data transfer CONTENTS3.8 Universal Memory...... 38 3.9 Multicore...... 43 and synchronization overheads among processor 3.10 Photonics...... 47 cores). So far, such parallelization has been per- 3.11 Networking and Interconnectivity... 52 formed by application programmers, but it is very 3.12 Software-Defined Networks...... 55 difficult, takes a long time, and has a high cost. 3.13 High-Performance Computing Therefore, to use multicore in a wide variety of (HPC) ...... 63 applications, automatic parallelization tools such as 3.14 Cloud Computing...... 68 compilers will be very important [1]. 3.15 The Internet of Things...... 74 3.16 Natural User Interfaces...... 77 3.17 3D Printing ...... 81 3.9.2 State of the Art 3.18 Big Data and Analytics ...... 84 3.9 Multicore There are many options for the low-power em- 3.19 Machine Learning and Intelligent Systems ...... 89 bedded multicore processors on smartphones and 3.20 Computer Vision and Pattern 3.9.1 Introduction tablets, such as homogeneous multicores with 2, Recognition...... 92 Multicore has attracted wide attention from the 4, and 8 cores, heterogeneous multicores combin- 3.21 Life Sciences...... 96 embedded systems community in such areas as ing super low-power 4 cores and ordinary 4 cores 3.22 Computational Biology and [2], heterogeneous multicores with low-power Bioinformatics...... 102 automobiles, smartphones, cameras, tablets, PCs, general-purpose processor cores and accelerator 3.23 Medical Robotics...... 108 and medical systems to high-performance com- cores like GPU cores [3]. The accelerators are very 4. DRIVERS AND DISRUPTORS...... 112 puting systems such as cloud servers and supercomputers. It is known widely that the consumed important for realizing low-power computation 5. TECHNOLOGY COVERAGE IN since accelerators give us high performance with IEEE XPLORE AND BY dynamic power is proportional to the clock fre- IEEE SOCIETIES...... 115 quency cube. So, if we lower the frequency to 1/4, low clock frequency. However, programming for GPUs is often difficult and time-consuming, and 6. IEEE COMPUTER SOCIETY the dynamic power will go down to 1/64, and if we IN 2022...... 120 increase processor cores 4 times to compensate the communication overhead among general-purpose processor cores and GPU cores is sometimes 7. SUMMARY AND NEXT STEPS.....124 for performance degradation, the power will be 8. AUTHORS...... 126 APPENDIX I. 23 Technologies Coverage 43 in IEEE Publications...... 134 CONTENTS 1. INTRODUCTION...... 5 2. SEAMLESS INTELLIGENCE SCENARIO...... 8 3. 23 TECHNOLOGIES IN 2022...... 14 3.1 Security Cross-Cutting Issues...... 14 3.2 The Open Intellectual Property Movement...... 17 3.3 Sustainability...... 20 3.4 Massively Online Open Courses...... 25 3.5 Quantum Computing ...... 28 3.6 Device and Nanotechnology...... 31 3.7 3D Integrated Circuits...... 33 CONTENTS3.8 Universal Memory...... 38 3.9 Multicore...... 43 3.10 Photonics...... 47 3.11 Networking and Interconnectivity... 52 3.12 Software-Defined Networks...... 55 3.13 High-Performance Computing (HPC) ...... 63 3.14 Cloud Computing...... 68 3.15 The Internet of Things...... 74 3.16 Natural User Interfaces...... 77 3.17 3D Printing ...... 81 3.18 Big Data and Analytics ...... 84 3.19 Machine Learning and Intelligent Systems ...... 89 Figure 8. Multicore. many-cores landscape. 3.20 Computer Vision and Pattern Recognition...... 92 very large. Coping with these problems will be 16-core general-purpose multicores are connect- 3.21 Life Sciences...... 96 crucial for the next generation of heterogeneous ed with high-performance GPGPUs, including 3.22 Computational Biology and multicores. more than tens of Pflops supercomputers. In these Bioinformatics...... 102 For routers, servers, and supercomputers relatively high-performance systems, the most difficult prob- 3.23 Medical Robotics...... 108 lem is how to efficiently program many processor 4. DRIVERS AND DISRUPTORS...... 112 high-performance multicores and many-cores are becoming available. For example, 8- to 16-cores and accelerator cores. 5. TECHNOLOGY COVERAGE IN IEEE XPLORE AND BY homogeneous multicores [4] or more than 50-core Another important problem is how to realize IEEE SOCIETIES...... 115 co-processors [5] are available for servers, and low-power hardware and software combinations. 6. IEEE COMPUTER SOCIETY more than 100-core homogenous many-core pro- The most advanced low-power technology is IN 2022...... 120 cessors have been planned for network processors compiler control [1][7][8]. So far, low-power soft- 7. SUMMARY AND NEXT STEPS.....124 [6]. As heterogeneous multiprocessors, the 8- to ware has been realized in the operating system by 8. AUTHORS...... 126 APPENDIX I. 23 Technologies Coverage 44 in IEEE Publications...... 134 CONTENTS 1. INTRODUCTION...... 5 2. SEAMLESS INTELLIGENCE power gating idle processors through virtualization wearable IT systems, smartphones, cameras, SCENARIO...... 8 among different application programs. In the latest games, automobiles, medical systems such as 3. 23 TECHNOLOGIES IN 2022...... 14 technology, each application program accomplish- drinkable inner-cameras for health diagnosis, can- 3.1 Security Cross-Cutting Issues...... 14 es low power through a parallelizing compiler—for cer treatment systems that use carbon ions or pro- 3.2 The Open Intellectual example, a program is parallelized by the compiler, tons, and solar-powered cloud servers to exascale Property Movement...... 17 which inserts DVFS (dynamic voltage and frequen- supercomputers for super-low-power high-perfor- 3.3 Sustainability...... 20 3.4 Massively Online Open Courses...... 25 cy scaling), clock gating, or power gating APIs into mance computation. Multicores and many-cores 3.5 Quantum Computing ...... 28 programs to slowly operate or completely stop light will allow us to recharge our smartphones just once 3.6 Device and Nanotechnology...... 31 load or busy-waiting processors for synchroniza- a month or even enable solar-power recharging. CONTENTS3.7 3D Integrated Circuits...... 33 tion. Especially in real-time computation, program 3.8 Universal Memory...... 38 parts on the critical path are slowly executed by 3.9.5 Potential Disruptions 3.9 Multicore...... 43 DVFS to satisfy a given deadline. With this control, A few technology innovations could disrupt multi- 3.10 Photonics...... 47 program parts not on the critical path have more 3.11 Networking and Interconnectivity... 52 core/many-core systems: chances to be slowly executed or stopped. In other 3.12 Software-Defined Networks...... 55 • Automatic multigrain parallelizing and low-pow- 3.13 High-Performance Computing words, speedup by parallel processing gives us (HPC) ...... 63 low-power execution for real-time computation, er compilers. Multicore or multiprocessor 3.14 Cloud Computing...... 68 such as moving picture applications. systems have been researched or used for 3.15 The Internet of Things...... 74 several decades with varying degrees of diffi- 3.16 Natural User Interfaces...... 77 3.9.3 Challenges culty. Before 2022, automatic multigrain par- 3.17 3D Printing ...... 81 allelizing compilers that use coarse-grain task 3.18 Big Data and Analytics ...... 84 The top challenges for multicore are as follows: parallelization, traditional loop parallelization, 3.19 Machine Learning and Intelligent Systems ...... 89 • low-power scalable homogeneous and hetero- and fine- or near-fine-grain parallelization will 3.20 Computer Vision and Pattern geneous architectures and their programming; be available for most multicore or many-core Recognition...... 92 • hard real-time architectures with local memory processors. The compilers will automatically 3.21 Life Sciences...... 96 and their programming; insert DVFS and clock-gating APIs for dynam- 3.22 Computational Biology and • automatic parallelization and low power ic power reduction and power-gating APIs Bioinformatics...... 102 control; for static power reduction with the efficient 3.23 Medical Robotics...... 108 • debugging and tuning tools; use of nonvolatile memory. The compiler will 4. DRIVERS AND DISRUPTORS...... 112 • reliable architectures and software; and let application developers parallelize in a few 5. TECHNOLOGY COVERAGE IN • solar-powered multicores for everything from minutes compared to several months of careful IEEE XPLORE AND BY IEEE SOCIETIES...... 115 embedded to high-performance computation. hand-written programing. Furthermore, the automatic power reduction with clock and power 6. IEEE COMPUTER SOCIETY 3.9.4 Where We Think It Will Go IN 2022...... 120 gating will reduce or entirely eliminate dead- In 2022, multicore will be everywhere, from locks caused by manual power tuning. 7. SUMMARY AND NEXT STEPS.....124 8. AUTHORS...... 126 APPENDIX I. 23 Technologies Coverage 45 in IEEE Publications...... 134 CONTENTS 1. INTRODUCTION...... 5 2. SEAMLESS INTELLIGENCE • Many-cores for super low-power execution. 3.9.7 References SCENARIO...... 8 Many-cores will be used not only for high-pow- [1] Jun Shirako, Nato Oshiyama, Yasutaka Wada, 3. 23 TECHNOLOGIES IN 2022...... 14 er computation but also for super-low-power Hiroaki Shikano, Keiji Kimura, Hironori Kasaha- 3.1 Security Cross-Cutting Issues...... 14 computation. For example, a small capsule ra, “Compiler Control Power Saving Scheme for 3.2 The Open Intellectual could contain a camera that a person can easily Property Movement...... 17 Multi Core Processors,” Proc. 18th International swallow and continuously take pictures for 7 or 3.3 Sustainability...... 20 Workshop on Languages and Compilers for Parallel 3.4 Massively Online Open Courses...... 25 8 hours; the processor cores could perform pat- Computing (LCPC2005), Oct. 2005. 3.5 Quantum Computing ...... 28 tern matching for cancer detection using a very [2] http://www.arm.com/ja/products/processors/ 3.6 Device and Nanotechnology...... 31 small battery inside the capsule. This kind of 3.7 3D Integrated Circuits...... 33 application requires a 1/100 to 1/1,000 proces- cortex-a/cortex-a17-processor.php CONTENTS3.8 Universal Memory...... 38 sor power reduction, which would require low [3] http://www.nvidia.com/object/nvidia-kepler. 3.9 Multicore...... 43 clock frequency and low voltage via speedup html 3.10 Photonics...... 47 through many-core parallel processing. More 3.11 Networking and Interconnectivity... 52 [4] http://www.amd.com/en-us/products/ processor cores are important not only for high 3.12 Software-Defined Networks...... 55 server/6000/6300# 3.13 High-Performance Computing performance but also for low power. (HPC) ...... 63 • Low-power multicores will be everywhere. As [5] http://www.intel.com/content/www/us/en/ 3.14 Cloud Computing...... 68 killer micros took over almost all computer processors/xeon/xeon-phi-detail.html 3.15 The Internet of Things...... 74 fields, lower-power multicore processors in- [6] http://www.tilera.com/products/processors 3.16 Natural User Interfaces...... 77 cluding low-power embedded multicores will be 3.17 3D Printing ...... 81 used by almost all IT systems in everyday life. [7] Keiji Kimura, Cecilia Gonzales-Alvarez, Akihiro 3.18 Big Data and Analytics ...... 84 Hayashi, Hiroki Mikami, Mamoru Shimaoka, Jun 3.19 Machine Learning and Shirako, Hironori Kasahara, “OS CAR API v2.1: Intelligent Systems ...... 89 3.9.6 Summary 3.20 Computer Vision and Pattern Multicores and many-cores will be everywhere Extensions for an Advanced Accelerator Con- Recognition...... 92 from wearable devices, cameras, smartphones, trol Scheme to a Low-Power Multicore API,” 17th 3.21 Life Sciences...... 96 automobiles, medical systems, cloud servers to Workshop on Compilers for Parallel Computing 3.22 Computational Biology and (CPC2013), Lyon, France, Jul . 2013. Bioinformatics...... 102 exa-scale supercomputers in 2022. Those multi- 3.23 Medical Robotics...... 108 core architectures will be designed with the auto- [8] http://www.youtube.com/ matic parallelizing and power lowering compilers 4. DRIVERS AND DISRUPTORS...... 112 watch?v=M63W2RAjXfc and multiplatform API to make parallel program- 5. TECHNOLOGY COVERAGE IN IEEE XPLORE AND BY ming easier and power consumption lower. Such IEEE SOCIETIES...... 115 low-power multicores will open a road to a solar 6. IEEE COMPUTER SOCIETY powered electronics society. IN 2022...... 120 7. SUMMARY AND NEXT STEPS.....124 8. AUTHORS...... 126 APPENDIX I. 23 Technologies Coverage 46 in IEEE Publications...... 134 CONTENTS 1. INTRODUCTION...... 5 2. SEAMLESS INTELLIGENCE IEEE CS 2022 , Report DRAFT increasingly require23 Technologies wider bandwidths, in 2022 and the 1/26/2014 5:18 PM SCENARIO...... 8 communication energy growing in a nonlinear way, 3. 23 TECHNOLOGIES IN 2022...... 14 3.10 Photonics with bandwidth, makes the problem even worse. 3.1 Security Cross-Cutting Issues...... 14 3.10 Photonics 3.2 The Open Intellectual At the processor-memory level, tighter integration Property Movement...... 17 3.10.1 Introductionof memory and processor using 3D-ICs will address 3.3 Sustainability...... 20 3.10.1 Introduction The technology roadmap several for of the communication data communication challenges. faces three Looking 3.4 Massively Online Open Courses...... 25 The technology roadmap for datachallenges: communication achieving energy at past efficie trendsncy, shows scaling bandwidth that evolutionary to electronic 3.5 Quantum Computing ...... 28 faces three challenges: achievingtrack energy processor efficiency , roadmaps, solutions and will neither low delivering latency reduce the across data communica- 3.6 Device and Nanotechnology...... 31 scaling bandwidth to track processorsystems roadmaps,, with exponentially tions energy increasing nor substantially numbers increase of cores and the band- 3.7 3D Integrated Circuits...... 33 and delivering low latency acrossprocessors systems, with [Moo11]. widths, and certainly not both at the same time. CONTENTS3.8 Universal Memory...... 38 exponentially increasing numbers of cores and 3.9 Multicore...... 43 The energy to move For communication data today beyond exceeds the the individual energy to actually socket, processors [Moo11]. 3.10 Photonics...... 47 compute on the data photonic itself [Dal10], interconnects and this trend offer the is best expected to path to low- 3.11 Networking and Interconnectivity... 52 The energy to move data todaycontinue. exceeds the For energy today’s bit high-‐end transfer systems, energies the and fraction the bandwidth of power scaling 3.12 Software-Defined Networks...... 55 to actually compute on the data itself [Dal10], and needed to track increases in CPU performance 3.13 High-Performance Computing and cost for communications is comparable to processors or memory. Hence, data communication (HPC) ...... 63 this trend is expected to continue.efficiencies For today’s must high- be [Ast09]. sought Emerging at essentially silicon photonic every scale from execution technologies of instructions for within the processor 3.14 Cloud Computing...... 68 end systems, the fraction to the machine room floor. 3.15 The Internet of Things...... 74 of power and cost for 3.16 Natural User Interfaces...... 77 communications is com- 3.17 3D Printing ...... 81 parable to processors 3.18 Big Data and Analytics ...... 84 or memory. Hence, data 3.19 Machine Learning and Intelligent Systems ...... 89 communication efficien- 3.20 Computer Vision and Pattern cies must be sought at Recognition...... 92 essentially every scale 3.21 Life Sciences...... 96 from execution of instruc- 3.22 Computational Biology and tions within the processor Bioinformatics...... 102 to the machine room 3.23 Medical Robotics...... 108 floor. 4. DRIVERS AND DISRUPTORS...... 112 5. TECHNOLOGY COVERAGE IN The fact that data com- IEEE XPLORE AND BY munication is energy IEEE SOCIETIES...... 115 inefficient relative to 6. IEEE COMPUTER SOCIETY computation and storage Figure 9 . Data movement cost: the unbalance of computing vs. moving data energy efficiency [Sha13]. IN 2022...... 120 (Figure 9) is only part of Figure 9. Data movement cost: the unbalance of computing vs. moving data energy efficiency 7. SUMMARY AND NEXT STEPS.....124 The fact that data communication is energy inefficient relative to computation and Figure storage ( 9 ) is the problem. Systems [Sha13]. 8. AUTHORS...... 126 only part of the problem. Systems increasingly require wider bandwidths, and the communication APPENDIX I. 23 Technologies Coverage energy growing in a nonlinear way, with bandwidth, makes the problem even worse. 47 in IEEE Publications...... 134 At the processor-‐memory level, tighter integration of memory and processor using 3D-‐ICs will address several of the communication challenges. Looking at past trends shows that evolutionary electronic solutions will neither reduce the data communications energy nor substantially increase the bandwidths, and certainly not both at the same time.

For communication beyond the individual socket, photonic interconnects offer the best path to low-‐bit transfer energies and the bandwidth scaling needed to CPU track increases in performance [Ast09]. Emerging silicon photonic technologies for interconnect fabrics have radically different performance characteristics when compared to existing CMOS electronics solutions. Silicon photonics offers lower

47 IEEE CS 2022 , Report DRAFT 23 Technologies in 2022 1/26/2014 5:18 PM

power and higher bandwidth nsity, de and eliminates -‐ the link length restrictions associated with electronic interconnects.

There are compelling arguments showing that silicon photonics is a foundational technology for high-‐ end systems [Bea11]. In the 2022 timeframe, high-‐end computing is expected to be in the exascale range. If an exa-‐operation application requires a communication ratio of only 0.04 bytes/operation, and each message goes on average through three hops (a very aggressive estimate), the total communication rate adds up to 40 TB/s, which at 4 pJ/bit per hop is about 4 MW, or 20 percent of the expected 20-‐MW system energy budget. Aiming at 4 pJ/bit per hop implies that the link energy has to be CONTENTS within 1 pJ/bit. Pervasive silicon photonics, all the way down to the compute elements, is the only technology that can reach this Figure objective ( 10 ). 1. INTRODUCTION...... 5 2. SEAMLESS INTELLIGENCE interconnect fabrics have radically SCENARIO...... 8 different performance character- 3. 23 TECHNOLOGIES IN 2022...... 14 istics when compared to existing 3.1 Security Cross-Cutting Issues...... 14 CMOS electronics solutions. Sili- 3.2 The Open Intellectual con photonics offers lower power Property Movement...... 17 and higher bandwidth density, and 3.3 Sustainability...... 20 3.4 Massively Online Open Courses...... 25 eliminates the link-length restric- 3.5 Quantum Computing ...... 28 tions associated with electronic 3.6 Device and Nanotechnology...... 31 interconnects. 3.7 3D Integrated Circuits...... 33 There are compelling arguments CONTENTS3.8 Universal Memory...... 38 3.9 Multicore...... 43 showing that silicon photonics is a 3.10 Photonics...... 47 foundational technology for high- 3.11 Networking and Interconnectivity... 52 end systems [Bea11]. In the 2022 3.12 Software-Defined Networks...... 55 timeframe, high-end computing Figure 10.Figure Rule-of-thumb10 . Rule-‐of of-‐thumb using photonics of using vs. electronics photonics based vs. on distance electronics and based on distance and required bandwidth. 3.13 High-Performance Computing is expected to be in the exascale requiredWhat bandwidth. emerges What from emerges the s graph i from that the graph roughly is that above roughly above 100 100 Gbps Gbps per meter, photonics is clearly a win. Below 10 Gbps (HPC) ...... 63 range. If an exa-operation appli- perper meter, meter, photonics electronics is clearly a is win. Below clearly 10 Gbps a per win meter, (and electronics a gray is clearly area in between). As we move to exascale and massively scale-‐out 3.14 Cloud Computing...... 68 cation requires a communication asystems, win (and a the gray area pressure in between). for As we more move bandwidth to exascale increases, and so and does massively the scale- appeal to use photonics for shorter-‐distance 3.15 The Internet of Things...... 74 outcommunication. systems, the pressure for more bandwidth increases, and so does the appeal to 3.16 Natural User Interfaces...... 77 ratio of only 0.04 bytes/operation, use photonics for shorter-distance communication. and each message goes on av- 3.17 3D Printing ...... 81 3.10.2 State of the Art erage through three hops (a very 3.18 Big Data and Analytics ...... 84 In high-‐end systems players are today, improving most the optical cost effectiveness interconnects use VCSEL-‐based of transmitters, large area 3.19 Machine Learning and aggressive estimate), the total communication rate detectors, and these multimode interconnects fibers through [Bow13]. simplifications Several in players he are improving t cost effectiveness of these Intelligent Systems ...... 89 adds up to 40 TB/s, which at 4 pJ/bit per hop is interconnects through packaging simplifications and by increasing in their packaging bandwidth and with by increasing their bandwidth with 3.20 Computer Vision and Pattern about 4 MW, or 20 percent of the expected 20-MW Recognition...... 92 improvements improvements in VSCEL technology.in VSCEL technology. system energy budget. Aiming at 4 pJ/bit per hop 3.21 Life Sciences...... 96 Silicon photonic transmitters with limited wave- 3.22 Computational Biology and implies that the link energy has to be withinSilicon 1 pJ/ photonic transmitters with limited wavelength division multiplexing (WDM) capabilities have length division multiplexing (WDM) capabilities Bioinformatics...... 102 bit. Pervasive silicon photonics, all the waybeen down demonstrated to by several companies (such as Luxtera, Intel, and IBM) by using Mach-‐Zender-‐based have been demonstrated by several companies 3.23 Medical Robotics...... 108 the compute elements, is the only technologymodulators. that Since these devices exploit a weak effect to modulate the light, they have to be relatively (such as Luxtera, Intel, and IBM) by using Mach- 4. DRIVERS AND DISRUPTORS...... 112 can reach this objective (Figure 10). large in silicon real estate, which leads to high power requirements and limited scope for integration. Zender-based modulators. Since these devices 5. TECHNOLOGY COVERAGE IN Alternate approaches using resonant structures such as micro-‐rings as modulators have also been exploit a weak effect to modulate the light, they IEEE XPLORE AND BY 3.10.2 State of the Art demonstrated by several companies (such as IBM, Sun, and HP). These resonators are more compact have to be relatively large in silicon real estate, IEEE SOCIETIES...... 115 and enable dense wavelength division multiplexing (DWDM), but tuning the resonators and matching In high-end systems today, most optical intercon- which leads to high power requirements and 6. IEEE COMPUTER SOCIETY them to an appropriate laser source remain unsolved technical challenges. IN 2022...... 120 nects use VCSEL-based transmitters, large area limited scope for integration. Alternate approach- detectors, and multimode fibers [Bow13]. Several 7. SUMMARY AND NEXT STEPS.....124 es using resonant structures such as micro-rings 8. AUTHORS...... 126 APPENDIX I. 23 Technologies Coverage 48 48 in IEEE Publications...... 134 CONTENTS 1. INTRODUCTION...... 5 2. SEAMLESS INTELLIGENCE as modulators have also been demonstrated by the overall cost. SCENARIO...... 8 several companies (such as IBM, Sun, and HP). Although photonic interconnects are in principle 3. 23 TECHNOLOGIES IN 2022...... 14 These resonators are more compact and enable more power efficient than electronic interconnects 3.1 Security Cross-Cutting Issues...... 14 dense wavelength division multiplexing (DWDM), for rack-to-rack distances and beyond, the use of 3.2 The Open Intellectual but tuning the resonators and matching them to an Property Movement...... 17 active optical cables (AOCs) has negated much of appropriate laser source remain unsolved technical 3.3 Sustainability...... 20 this advantage. The full benefit of optical intercon- challenges. 3.4 Massively Online Open Courses...... 25 nect can only be realized when the entire physical 3.5 Quantum Computing ...... 28 Finally, hybrid silicon ring lasers [Lia11] use rings of link path is designed for photonics. 3.6 Device and Nanotechnology...... 31 silicon waveguide as resonators and as a laser cavi- A complete integrated photonic link requires de- 3.7 3D Integrated Circuits...... 33 ty stimulated by a layer of III/V material bonded to CONTENTS3.8 Universal Memory...... 38 tectors, optical drop filters, and a range of wave- the silicon. As the laser’s wavelength is determined 3.9 Multicore...... 43 guide and coupling technologies to suit different 3.10 Photonics...... 47 by cavity geometry, several highly compact lasers applications, which in turn involves ecosystem and 3.11 Networking and Interconnectivity... 52 in a range of wavelengths can be formed on the supply-chain issues that are being addressed. An 3.12 Software-Defined Networks...... 55 same substrate, simply by varying the diameter of additional problem lies in coupling light on and off 3.13 High-Performance Computing the resonant cavity. Samples of these devices have the integrated photonic die: while several ap- (HPC) ...... 63 been shown to be capable of direct modulation 3.14 Cloud Computing...... 68 proaches have been shown (including tapers and at 10 Gbps. The directly modulated ring laser has 3.15 The Internet of Things...... 74 grating couplers), significant challenges remain. 3.16 Natural User Interfaces...... 77 several advantages: no requirement for an exter- 3.17 3D Printing ...... 81 nal laser source and optical power distribution, a Finally, all photonics device technologies require 3.18 Big Data and Analytics ...... 84 greatly simplified tuning, and power proportionality innovative packaging that allows large numbers of 3.19 Machine Learning and (the devices can be powered off when not in use). single-mode waveguide connections to be made Intelligent Systems ...... 89 between devices and subassemblies. The develop- 3.20 Computer Vision and Pattern ment of appropriate packaging and connector and Recognition...... 92 3.10.3 Challenges waveguide technologies is an obvious area where 3.21 Life Sciences...... 96 While limited WDM is possible with VCSELs, the additional development is necessary. 3.22 Computational Biology and cost of such links is still proportional to the band- Bioinformatics...... 102 width, as each additional wavelength requires ad- 3.23 Medical Robotics...... 108 ditional components. Silicon photonics has the po- 3.10.4 Where We Think It Will Go 4. DRIVERS AND DISRUPTORS...... 112 tential for much higher bandwidth density through The design of packet switches for processor net- 5. TECHNOLOGY COVERAGE IN WDM and lower power through the use of low-loss, works is constrained by the bandwidth density IEEE XPLORE AND BY available at the chip perimeter. As the connections IEEE SOCIETIES...... 115 single-mode fiber and waveguide detectors. With silicon photonics, bandwidth can be scaled by to switches span distances ranging 10 cm to 20 6. IEEE COMPUTER SOCIETY m, the link-length independence of photonics is IN 2022...... 120 adding very compact transmitters and detectors to an integrated photonic die, at a minimal increase in particularly attractive. For these reasons, we be- 7. SUMMARY AND NEXT STEPS.....124 lieve that switches will be the first to exploit the 8. AUTHORS...... 126 APPENDIX I. 23 Technologies Coverage 49 in IEEE Publications...... 134 IEEE CS 2022 , Report DRAFT 23 Technologies in 2022 1/26/2014 5:18 PM

CONTENTS 1. INTRODUCTION...... 5 2. SEAMLESS INTELLIGENCE benefit of close integra- SCENARIO...... 8 tion between high-den- 3. 23 TECHNOLOGIES IN 2022...... 14 sity CMOS logic and 3.1 Security Cross-Cutting Issues...... 14 silicon photonic commu- 3.2 The Open Intellectual nication. Figure 11 shows Property Movement...... 17 the time progression of 3.3 Sustainability...... 20 3.4 Massively Online Open Courses...... 25 photonics technologies, 3.5 Quantum Computing ...... 28 from active cables (single 3.6 Device and Nanotechnology...... 31 wavelength) to compo- 3.7 3D Integrated Circuits...... 33 nent-based photonics CONTENTS3.8 Universal Memory...... 38 (CWDM) and on-chip 3.9 Multicore...... 43 interconnects (DWDM). 3.10 Photonics...... 47

3.11 Networking and Interconnectivity... 52 DWDM silicon photonics Figure 11. Roadmap of industrial photonics technologies. (source: HP) 3.12 Software-Defined Networks...... 55 will enable the develop- Figure 11 . Roadmap of industrial photonics technologies. (source: HP) 3.13 High-Performance Computing ment of very high-radix 3.10.5 Disruptions (HPC) ...... 63 switch components while continuingDWDM silicon to scale photonics will enable the development of very high-‐radix switch components while 3.14 Cloud Computing...... 68 continuing to scale switch A critical port application bandwidths of photonics to will track be to improvements build in processor performance. Increasing switch port bandwidths to track improvements 3.15 The Internet of Things...... 74 the radix of switches highly allows low-‐diameter energy-efficient networks router to components be deployed that are with consequent advantages in in processor performance. Increasing the radix 3.16 Natural User Interfaces...... 77 latency, energy efficiency, interconnected and reliability. By optically, exploiting use CMOS silicon electronics photonics, switch components with 3.17 3D Printing ...... 81 of switches allows low-diameter networks to be bandwidth up to 50 internally Tbit/s for will packet become processing a reality. and -‐ An attraction of highbuffering,radix network and topologies is that 3.18 Big Data and Analytics ...... 84 deployed with consequent advantages in laten- the switches are distributed connect to with high-performance the processors computing in a engines regular omplexity way, avoiding the wiring c of 3.19 Machine Learning and cy, energy efficiency, and reliability. By exploiting more efficiently than what can be achieved by Intelligent Systems ...... 89 silicon photonics, switch componentscentralized with switches. band - 3.20 Computer Vision and Pattern co-packaging. Micro-solder bumps and face-to- width up to 50 Tbit/s will become a reality. An Recognition...... 92 By incorporating additional face copper logic bonds in the allow switch much smaller fabric, connections the network will become more intelligent, and attraction of high-radix network topologies is that 3.21 Life Sciences...... 96 operations such as collective between devices, broadcast allowing and arrays reduction, of closely whose packed importance increases with larger node the switches are distributed with the processors 3.22 Computational Biology and counts, will become a reality transceivers through to intelligent be bonded to support the CMOS in switch the network itself. Bioinformatics...... 102 in a regular way, avoiding the wiring complexity of device. This increases the CMOS device’s effective 3.23 Medical Robotics...... 108 centralized switches. 3.10.5 Disruptions chip-edge bandwidth, a performance bottleneck in 4. DRIVERS AND DISRUPTORS...... 112 A critical application of photonics will be to build highly energy-‐efficient router components that are By incorporating additional logic in the switch fab- today’s systems, and enables the development of 5. TECHNOLOGY COVERAGE IN interconnected optically, use CMOS electronics internally for packet processing and buffering, and ric, the network will become more intelligent, and higher port count switches without reducing port IEEE XPLORE AND BY connect to high-‐performance computing engines more efficiently than what can -‐ be achieved by co operations such as collective broadcast and reduc- bandwidths. Higher port count switches further IEEE SOCIETIES...... 115 packaging. Micro-‐solder bumps and face-‐to-‐face copper bonds allow much smaller connections between tion, whose importance increases with larger node contribute to lower communication energy by en- 6. IEEE COMPUTER SOCIETY devices, allowing arrays of closely packed transceivers to be bonded to the CMOS switch device. This IN 2022...... 120 counts, will become a reality through intelligent abling networks with a lower diameter to be con- increases the CMOS device’s effective chip-‐edge bandwidth, a performance bottleneck in today’s support in the network itself. structed that require fewer retransmissions 7. SUMMARY AND NEXT STEPS.....124 systems, and enables the development of higher port reducing count switches without port bandwidths. 8. AUTHORS...... 126 Higher port count switches further contribute to lower communication energy by enabling networks APPENDIX I. 23 Technologies Coverage with a lower diameter to be that constructed require fewer retransmissions 50 in IEEE Publications...... 134 The availability of these new switches will enable whole new classes of network topologies that combine the ease of deployment of grids and meshes with the high levels of path diversity logarithmic networks such as fat trees.

50 CONTENTS 1. INTRODUCTION...... 5 2. SEAMLESS INTELLIGENCE The availability of these new switches will enable one connecting chips in a supercomputer. This will SCENARIO...... 8 whole new classes of network topologies that com- create disruption for telco manufacturers [Sar14]. 3. 23 TECHNOLOGIES IN 2022...... 14 bine the ease of deployment of grids and meshes 3.1 Security Cross-Cutting Issues...... 14 with the high levels of path diversity of logarithmic 3.10.7 References 3.2 The Open Intellectual networks such as fat trees. Property Movement...... 17 [Ahn09] D. Vantrease et al., “Corona: System Impli- 3.3 Sustainability...... 20 Network topologies such as HyperX [Ahn09] and cations of Emerging Nanophotonic Technology,” 35th 3.4 Massively Online Open Courses...... 25 flattened butterfly are ideally suited to high-radix Int’l Symp. Computer Architecture (ISCA ’08), 2008, pp. 3.5 Quantum Computing ...... 28 switch components: with a high degree of path 153–164. 3.6 Device and Nanotechnology...... 31 diversity, they have the potential to provide a highly [Ast09] G. Astfalk, “Why Optics and Why Now?,” Ap- 3.7 3D Integrated Circuits...... 33 resilient interconnect fabric scaling to millions of plied Physics A, vol. 95, 2009, pp. 933-940. CONTENTS3.8 Universal Memory...... 38 3.9 Multicore...... 43 nodes. [Beau11] R.G. Beausoleil, “Large-Scale Integrated 3.10 Photonics...... 47 Photonics for High-Performance Interconnects,” ACM J. 3.11 Networking and Interconnectivity... 52 3.10.6 Summary Emerg. Technol. Comput. Syst., vol. 7, 2011. 3.12 Software-Defined Networks...... 55 Silicon photonics will be a fundamental technolo- [Bin11] N. Binkert et al., “The Role of Optics in Future 3.13 High-Performance Computing (HPC) ...... 63 gy to address the bandwidth, latency, and energy High Radix Wwitch Design,” Proc. 38th Annual Int’l 3.14 Cloud Computing...... 68 challenges in the fabric of high-end systems. Symp. Computer Architecture (ISCA ‘11), ACM, 2011, pp. 437–448. 3.15 The Internet of Things...... 74 This area opens up the opportunity to build 3.16 Natural User Interfaces...... 77 [Bow13] J. Bowers, “Trends, Possibilities and Limitations 3.17 3D Printing ...... 81 high-radix switches, with integrated support for important collective operations such as multi- of Silicon Photonic Integrated Circuits and Devices,” 3.18 Big Data and Analytics ...... 84 A Tutorial at the IEEE Custom Integrated Circuits Conf., cast barriers, reductions, scatter, and gather. In 3.19 Machine Learning and 2013. Intelligent Systems ...... 89 the 2022 timeframe, such a photonics-enabled 3.20 Computer Vision and Pattern high-radix switch could reach 64 to 128 ports, 640 [Dal10] B. Dally, “To ExaScale and Beyond (presenta- Recognition...... 92 Gbps per port, and 1 pJ/bit of link energy, which will tion),” Supercomputing 2010 keynote, 2010; http://www. 3.21 Life Sciences...... 96 enable connecting 1 million ports. nvidia.com/content/PDF/sc_2010/theater/Dally_SC10. 3.22 Computational Biology and pdf. Bioinformatics...... 102 Bringing photonics inside chips has another effect: 3.23 Medical Robotics...... 108 [Lia11] D. Liang et al., “Low Threshold Electrical- it gets rid of distance constraints, which in turns ly-Pumped Hybrid Silicon Microring Lasers,” IEEE J. 4. DRIVERS AND DISRUPTORS...... 112 leads to flatter networks. A full photonics-based Sel. Topics Quantum Electron., vol. 17, no. 6, 2011, pp. 5. TECHNOLOGY COVERAGE IN network is nothing but a giant supercomputer, 1528–1533. IEEE XPLORE AND BY where processing units are distributed geograph- IEEE SOCIETIES...... 115 [Moo11] C. Moore, “Data Processing in ExaScale-Class ically. This is going to change the software archi- Computer Systems (presentation),” Salishan Conf., 6. IEEE COMPUTER SOCIETY tecture of the switches in a telecommunications IN 2022...... 120 2011; http://www.lanl.gov/orgs/hpc/salishan/salis- network and will eventually collapse telecom han2011/3moore.pdf. 7. SUMMARY AND NEXT STEPS.....124 networks onto the computer’s inner network, the 8. AUTHORS...... 126 APPENDIX I. 23 Technologies Coverage 51 in IEEE Publications...... 134 CONTENTS 1. INTRODUCTION...... 5 2. SEAMLESS INTELLIGENCE [Sha13] J. Shalf, “Active Power Management Technology networks connecting datacenter machines and, of SCENARIO...... 8 Challenges and Implications for Programming Mod- course, the Internet. 3. 23 TECHNOLOGIES IN 2022...... 14 els (presentation),” Teratec Forum, 2014; http://www. teratec.eu/library/pdf/forum/2013/Pr%C3%A9senta- 3.1 Security Cross-Cutting Issues...... 14 3.11.2 State of the Art 3.2 The Open Intellectual tions/A4_01_John_Shalf_LBNL_FT2013.pdf. Property Movement...... 17 Wherever power is plentiful, there have been great [Sar14] Saracco, R., Personal Communication, 2014. 3.3 Sustainability...... 20 strides in communication technology. Datacenter 3.4 Massively Online Open Courses...... 25 networks, once ruled by 1-Gbyte Ethernet links, are 3.5 Quantum Computing ...... 28 fast evolving to 40-Gbyte+ interconnects, per- 3.6 Device and Nanotechnology...... 31 haps by 2020, even integrated into the same chips 3.7 3D Integrated Circuits...... 33 where data is stored and processed. It remains CONTENTS3.8 Universal Memory...... 38 3.9 Multicore...... 43 unclear, though, whether those on-chip intercon- 3.10 Photonics...... 47 nect technologies will be proprietary or open like 3.11 Networking and Interconnectivity... 52 Ethernet. Rapid improvements are also seen for 3.12 Software-Defined Networks...... 55 home and urban networking, where Wi-Fi technol- 3.13 High-Performance Computing ogy is improving rapidly, with many sites in cities (HPC) ...... 63 now well-connected, even subways and buses. Yet 3.14 Cloud Computing...... 68 3.11 Networking and there still remains a steep difference in connectivi- 3.15 The Internet of Things...... 74 Interconnectivity 3.16 Natural User Interfaces...... 77 ty availability between more versus less developed 3.17 3D Printing ...... 81 3.11.1 Introduction countries and within countries, between urban and 3.18 Big Data and Analytics ...... 84 rural areas. Nonetheless, between 2006 and 2011, Computer networks have become the basis for a 3.19 Machine Learning and for instance, the number of countries with com- Intelligent Systems ...... 89 broad set of critical applications, including sen- mercially available fixed broadband grew from 166 3.20 Computer Vision and Pattern sor networks and their extension to the Internet Recognition...... 92 to 206 (www.itu.int/en/ITU-D/Statistics/Pages/ of Things; Wi-Fi networks in homes, offices, and publications/wtid.aspx), and the number of nation- 3.21 Life Sciences...... 96 stores; high-end networks for large-scale parallel 3.22 Computational Biology and al broadband plans and policies in the world has Bioinformatics...... 102 machines and on-chip versions of those linking the more than doubled since 2009. Yet the ITU also 3.23 Medical Robotics...... 108 many cores and memories of many-core plat- estimates that while roughly 2.5 billion people used 4. DRIVERS AND DISRUPTORS...... 112 forms; and, of course, the Internet itself, which is the Internet in 2012, this included only a quarter composed of a federation of networks connecting 5. TECHNOLOGY COVERAGE IN of people in the developing world, and in the US, IEEE XPLORE AND BY sites across the Earth and even reaching into the there are still 19 million Americans who cannot buy IEEE SOCIETIES...... 115 solar system. fixed broadband Internet service. Wireless is more 6. IEEE COMPUTER SOCIETY In other words, computer systems cannot operate widely available, but even in October 2012, there IN 2022...... 120 without connectivity between their components, were still 1.9 million Americans without access, and 7. SUMMARY AND NEXT STEPS.....124 whether on chip or in larger form factors like the many rural access speeds remain low. As a result, 8. AUTHORS...... 126 APPENDIX I. 23 Technologies Coverage 52 in IEEE Publications...... 134 CONTENTS 1. INTRODUCTION...... 5 2. SEAMLESS INTELLIGENCE while technology permits us to connect with high a single chip, we need vastly more energy/bit to SCENARIO...... 8 bandwidths, access to technology remains a limit- move data between CPU and cache versus CPU 3. 23 TECHNOLOGIES IN 2022...... 14 ing factor [bustamante13]. and DRAM memory, and this ratio is getting worse 3.1 Security Cross-Cutting Issues...... 14 All this said, however, the world now operates with as technology progresses (although on-chip op- 3.2 The Open Intellectual tical interconnects offer some hope for increased Property Movement...... 17 cellphones and, more and more, with smartphones, bandwidth). Power issues are even worse when 3.3 Sustainability...... 20 where in less developed countries, this former operating in cyber-physical environments, an ex- 3.4 Massively Online Open Courses...... 25 luxury is sometimes critically important to the 3.5 Quantum Computing ...... 28 daily lives of their citizens. Cell towers, in fact, are ample being nanotech devices in your body having 3.6 Device and Nanotechnology...... 31 easier to construct than landlines, and the sudden to communicate with the outside world: this likely 3.7 3D Integrated Circuits...... 33 connectivity they have created is replacing exist- cannot be done without the infusion of outside CONTENTS3.8 Universal Memory...... 38 energy, e.g., via radio beams directed at your body. 3.9 Multicore...... 43 ing industries with new ones, one example being financial transactions via phones versus physical Or imagine sensors in rivers or in the forest: water 3.10 Photonics...... 47 movement or wind/solar energy may permit them 3.11 Networking and Interconnectivity... 52 banks. In 2020, even more of our world will be to communicate, but such energy must be harvest- 3.12 Software-Defined Networks...... 55 smartphone-connected, with higher bandwidth ed effectively. 3.13 High-Performance Computing connections in developed countries, and much (HPC) ...... 63 more connectivity everywhere else, particularly At larger scales, communications require infra- 3.14 Cloud Computing...... 68 in urban settings. In fact, for the first time, in 2013, structure, an example being the aforementioned 3.15 The Internet of Things...... 74 3.16 Natural User Interfaces...... 77 sales of smartphones were no longer dominated in cell towers, but there are interesting evolutions in 3.17 3D Printing ...... 81 total volume by the US and European markets. this infrastructure. Specifically, regarding the ability 3.18 Big Data and Analytics ...... 84 of end devices to interact with the cloud, there is 3.19 Machine Learning and 3.11.3 Challenges, Opportunities, ongoing evolution from the current model—end Intelligent Systems ...... 89 and What Will Likely Happen devices either directly interact with each other, via 3.20 Computer Vision and Pattern peering, or with the Internet, via nearby commu- Recognition...... 92 Many questions and issues remain concerning nication endpoints like routers or cell towers—to 3.21 Life Sciences...... 96 modern communication systems and infrastruc- a new model offering an additional intermediate 3.22 Computational Biology and tures. For using them in extremely small devices, Bioinformatics...... 102 layer, such as micro-cells, smarter home gateways, there are power issues, which make it unclear 3.23 Medical Robotics...... 108 or public access points in, say, coffee shops. Those whether communications can occur continuously 4. DRIVERS AND DISRUPTORS...... 112 flexible infrastructure components will not only or whether there will only be intermittent connec- offer the communication support already present in 5. TECHNOLOGY COVERAGE IN tivity, perhaps when devices can acquire addition- IEEE XPLORE AND BY current systems, but they may also provide useful al energy and/or dock with other systems. The IEEE SOCIETIES...... 115 computational or storage services offloaded from question arises because the network technologies 6. IEEE COMPUTER SOCIETY but still interacting with today’s giant datacenters. needed for such communications are inherent- IN 2022...... 120 An obvious example is data caching, but there ly limited by their energy consumption. Even on 7. SUMMARY AND NEXT STEPS.....124 are other, more interesting opportunities, such as 8. AUTHORS...... 126 APPENDIX I. 23 Technologies Coverage 53 in IEEE Publications...... 134 CONTENTS 1. INTRODUCTION...... 5 2. SEAMLESS INTELLIGENCE orientation services warning SCENARIO...... 8 cars away from streets cur- 3. 23 TECHNOLOGIES IN 2022...... 14 rently under construction or 3.1 Security Cross-Cutting Issues...... 14 experiencing a traffic jam. 3.2 The Open Intellectual Thus, we are likely to evolve Property Movement...... 17 to a world in which communi- 3.3 Sustainability...... 20 3.4 Massively Online Open Courses...... 25 cations become more tightly 3.5 Quantum Computing ...... 28 bound with services than 3.6 Device and Nanotechnology...... 31 those in the strictly layered 3.7 3D Integrated Circuits...... 33 systems now in place. This CONTENTS3.8 Universal Memory...... 38 same trend, in fact, is evident 3.9 Multicore...... 43 in datacenter systems, where 3.10 Photonics...... 47 there is a rapid ongoing evolu- 3.11 Networking and Interconnectivity... 52 tion from traditional to soft- 3.12 Software-Defined Networks...... 55 3.13 High-Performance Computing ware-defined networks: the (HPC) ...... 63 idea is to make networks more to applications much beyond that offered by cur- 3.14 Cloud Computing...... 68 programmable to meet existing or future applica- 3.15 The Internet of Things...... 74 tion needs. We do not further comment on those rent mostly homogeneous many-core chips. There 3.16 Natural User Interfaces...... 77 rich developments, as they are already reaching are additional promises derived from more closely 3.17 3D Printing ...... 81 into standards bodies, but note that the broader integrating services with communications. A prom- 3.18 Big Data and Analytics ...... 84 topic of software-defined systems or datacenters ising beginning is in software-defined networks and 3.19 Machine Learning and the many network appliances in use and being de- Intelligent Systems ...... 89 is now an active field of study. veloped for datacenter systems. Their services go 3.20 Computer Vision and Pattern beyond just assisting with communications to deep Recognition...... 92 3.11.4 Potential Disruptions 3.21 Life Sciences...... 96 packet inspection, data cleaning, threat detection 3.22 Computational Biology and Interesting developments underway today may and mitigation, and more. Will future services ac- Bioinformatics...... 102 lead to significant leaps forward in networking and tually mine incoming data, continuously, to extract 3.23 Medical Robotics...... 108 communications. There is the promise of silicon useful information and/or to discard data otherwise 4. DRIVERS AND DISRUPTORS...... 112 photonics, able to move data at energy costs sub- imposing undue load on back-end systems? An 5. TECHNOLOGY COVERAGE IN stantially less than with current technologies and example of the latter is spam: Why should we first IEEE XPLORE AND BY to assist both with on-chip and rack-level commu- store it, paying those costs, to only then recognize IEEE SOCIETIES...... 115 nications. With these technologies, we can almost its lack of value and discard it? 6. IEEE COMPUTER SOCIETY envision a rack of machines able to act like a single IN 2022...... 120 many-core chip, leading to levels of diversity in the In the near future, within this decade, halo nets will 7. SUMMARY AND NEXT STEPS.....124 processing and/or storage components available become a significant player in the creation of the 8. AUTHORS...... 126 APPENDIX I. 23 Technologies Coverage 54 in IEEE Publications...... 134 CONTENTS 1. INTRODUCTION...... 5 2. SEAMLESS INTELLIGENCE communication fabric. Indeed, one can say that SCENARIO...... 8 the evolution is from telecommunications infra- 3. 23 TECHNOLOGIES IN 2022...... 14 structures (designed top down and owned by a few 3.1 Security Cross-Cutting Issues...... 14 operators, requiring great CAPEX) to communica- 3.2 The Open Intellectual tion fabric (aggregated bottom up and owned by Property Movement...... 17 a variety of players with a variety of business and 3.3 Sustainability...... 20 3.4 Massively Online Open Courses...... 25 sustainability models). Terminals such as smart- 3.5 Quantum Computing ...... 28 phones will generate these halo nets, creating a 3.6 Device and Nanotechnology...... 31 continuously evolving network at the edges of the 3.7 3D Integrated Circuits...... 33 big backbones. [Saracco14]. CONTENTS3.8 Universal Memory...... 38 3.12 Software-Defined Networks 3.9 Multicore...... 43 3.11.5 Summary 3.10 Photonics...... 47 3.12.1 Introduction 3.11 Networking and Interconnectivity... 52 Communications and interconnects are seeing new 3.12 Software-Defined Networks...... 55 opportunities, open issues, and potential disrup- Scott Shenker, University of California, Berkeley, 3.13 High-Performance Computing tions from new technologies (silicon photonics), was one of the core contributors to the OpenFlow (HPC) ...... 63 new use cases (online data mining), new challeng- protocol and has a one-sentence description of 3.14 Cloud Computing...... 68 es (the increasingly high energy costs of moving software-defined networking: an SDN is a set of 3.15 The Internet of Things...... 74 abstractions for the control plane in networking. 3.16 Natural User Interfaces...... 77 data), and infrastructure investments (like those in 3.17 3D Printing ...... 81 developing countries). Developments at all levels This is very much a computer scientist’s descrip- 3.18 Big Data and Analytics ...... 84 of the network stack, from interconnects on single tion, for abstraction is the key tool in the system 3.19 Machine Learning and chips to the worldwide networks connecting data- computer scientist’s toolkit. It means separating Intelligent Systems ...... 89 center systems, will continue to drive the research 3.20 Computer Vision and Pattern the function of a device from its implementation. Recognition...... 92 and the Internet economy. This permits independent development of the 3.21 Life Sciences...... 96 implementation, and applications, of a device—pro- 3.22 Computational Biology and 3.11.6 References gramming language, operating system, architec- Bioinformatics...... 102 [bustamante13] F.E. Bustamante, “Broadly Available ture, or protocol. 3.23 Medical Robotics...... 108 Broadband,” IEEE Internet Computing, vol. 17, no. 5, 2013, Of course, this is ubiquitous in computer science 4. DRIVERS AND DISRUPTORS...... 112 pp. 3-5. 5. TECHNOLOGY COVERAGE IN and information technology, and examples are easy [Saracco14] Saracco, R., Personal Communication, IEEE XPLORE AND BY to enumerate: the x86 architecture (or any stan- IEEE SOCIETIES...... 115 2014. dardized ISA), the Linux kernel, the OpenStack API 6. IEEE COMPUTER SOCIETY in cloud services, any programming language, and IN 2022...... 120 any programming language’s standard library. In 7. SUMMARY AND NEXT STEPS.....124 fact, computer systems generally are hierarchies of 8. AUTHORS...... 126 APPENDIX I. 23 Technologies Coverage 55 in IEEE Publications...... 134 CONTENTS 1. INTRODUCTION...... 5 2. SEAMLESS INTELLIGENCE abstract architectures, with each one built on top codes, a two-phase commit protocol to the device, SCENARIO...... 8 of the stack below. Changes in implementation are and so on. We do this for network control proto- 3. 23 TECHNOLOGIES IN 2022...... 14 essentially hidden from upper layers. For example, cols, however, and we do it all the time—so often 3.1 Security Cross-Cutting Issues...... 14 changes in the implementation of the Linux kernel that we barely notice we’re doing it. For example, 3.2 The Open Intellectual are invisible to application programs, programs RFC 2328 specifies the Open Shortest Path First Property Movement...... 17 port easily between one x86 chip and another, and Protocol, the basic routing algorithm of the In- 3.3 Sustainability...... 20 3.4 Massively Online Open Courses...... 25 so on. Abstraction is so ubiquitous that we just ternet. RFC 2328 runs to 250 pages, of which 13 3.5 Quantum Computing ...... 28 assume it. are devoted to the method used to calculate the 3.6 Device and Nanotechnology...... 31 Given that, it is remarkable that, to this date, the appropriate paths; the remaining 237 pages spec- 3.7 3D Integrated Circuits...... 33 network control plane has had no similar set of ify details of how local information is propagated CONTENTS3.8 Universal Memory...... 38 to neighbors in the network graph, maintenance 3.9 Multicore...... 43 abstractions. The network data plane, in contrast, has an extremely successful layer of abstractions: of distributed state, security and authentication 3.10 Photonics...... 47 provisions, and so on. Given a fully developed set 3.11 Networking and Interconnectivity... 52 the familiar standard OSI protocol stack, with its of abstractions for the control plane, OSPF—and 3.12 Software-Defined Networks...... 55 physical, media access, Internet, transmission virtually every other Layer 2 and Layer 3 control 3.13 High-Performance Computing control, and application layers. This set of abstrac- (HPC) ...... 63 tions has been standardized for a generation and protocol—will be shortened by a significant degree. 3.14 Cloud Computing...... 68 has been stunningly successful: a Web server is We have only begun the journey toward standard- 3.15 The Internet of Things...... 74 3.16 Natural User Interfaces...... 77 (largely) independent of the Transmission Control izing and specifying a set of abstractions for the 3.17 3D Printing ...... 81 Protocol implementation upon which it relies, and control plane. Shenker specified three general 3.18 Big Data and Analytics ...... 84 the physical layer is completely invisible to applica- layers: the switch protocol, the network operat- 3.19 Machine Learning and tion programs. ing system, and the specification layer—a.k.a., the Intelligent Systems ...... 89 However, for all of the success of abstraction in the programming language. The initial switch protocol, 3.20 Computer Vision and Pattern OpenFlow, is now fairly mature; the network oper- Recognition...... 92 network data plane, there is no set of accompa- 3.21 Life Sciences...... 96 nying abstractions for the network control plane. ating system is becoming so, with the emergence 3.22 Computational Biology and And, to quote Shenker, “this is crazy.” With every of a number of open source controllers. The final Bioinformatics...... 102 new control protocol, network engineers have to step, the specification language, is fairly nascent, 3.23 Medical Robotics...... 108 re-specify and re-implement the general methods with the emergence of the FreNetic language from 4. DRIVERS AND DISRUPTORS...... 112 common to all control protocols: propagation of Princeton and Cornell. 5. TECHNOLOGY COVERAGE IN distributed state, failover, recovery from error, and IEEE XPLORE AND BY 3.12.2 State of the Art IEEE SOCIETIES...... 115 so on. If this were done for (for example) storage systems, the filesystem API wouldn’t exist, and The state of the art in OpenFlow networks is the 6. IEEE COMPUTER SOCIETY IN 2022...... 120 each application writer would have to re-speci- implementation of a standard protocol, in which a switch is reduced to a simple forwarding table. 7. SUMMARY AND NEXT STEPS.....124 fy the layout of blocks on disk, error-correcting 8. AUTHORS...... 126 APPENDIX I. 23 Technologies Coverage 56 in IEEE Publications...... 134 CONTENTS 1. INTRODUCTION...... 5 2. SEAMLESS INTELLIGENCE The table matches incoming packets based on has been moved off the individual switches and SCENARIO...... 8 specifications of the packet header—values of bits routers, and centralized and opened up to the 3. 23 TECHNOLOGIES IN 2022...... 14 in specific fields, with wildcards. On match, one of network administrator. And thus the network as a 3.1 Security Cross-Cutting Issues...... 14 four actions can be taken: send the packet out on whole is more transparent and more controllable 3.2 The Open Intellectual a port, ask the off-board controller for help, drop by the network administrator. The switch is a sim- Property Movement...... 17 the packet, or send the packet through the switch’s ple forwarding table. 3.3 Sustainability...... 20 normal processing pipeline (so-called “hybrid 3.4 Massively Online Open Courses...... 25 Since forwarding rules and policies, not the phys- 3.5 Quantum Computing ...... 28 mode”). The most recent implemented version of ical topology of a network, essentially defines 3.6 Device and Nanotechnology...... 31 the protocol offers the ability to match a packet what we mean by a “network,” this factoring of 3.7 3D Integrated Circuits...... 33 multiple times, offering the prospect of multiple the control plane offers the possibility of virtual CONTENTS3.8 Universal Memory...... 38 actions on a single packet. 3.9 Multicore...... 43 networks. A virtual network is an application- or 3.10 Photonics...... 47 The preceding description of OpenFlow suggests purpose-specific network with its own forwarding 3.11 Networking and Interconnectivity... 52 two things: one, this is within the capabilities of rules, segregated address and rulespace, quali- 3.12 Software-Defined Networks...... 55 almost any switch on the market today, and two, ty-of-service guarantees, and admission control 3.13 High-Performance Computing there is less need for on-switch software in a “soft- that can be set up and torn down on a dynamic ba- (HPC) ...... 63 ware-defined network” than in today’s networks. sis. This is dream in classic networks: an easy reali- 3.14 Cloud Computing...... 68 Both these observations are correct: OpenFlow ty in an OpenFlow network. You simply identify the 3.15 The Internet of Things...... 74 3.16 Natural User Interfaces...... 77 was deliberately designed to be implementable on virtual network by some combination of address 3.17 3D Printing ...... 81 the current generation of switches, and there is space, VLAN tag, ethertype, protocol, and port and 3.18 Big Data and Analytics ...... 84 less software on a pure OpenFlow switch than in a write the forwarding rules for this virtual network in 3.19 Machine Learning and standard switch. For the first, the original authors a specification to the controller; the controller then Intelligent Systems ...... 89 of the OpenFlow protocol observe, “While each forwards these rules to the individual switches. 3.20 Computer Vision and Pattern vendor’s flow-table is different, we’ve identified Recognition...... 92 This capability of OpenFlow of isolated virtual 3.21 Life Sciences...... 96 an interesting common set of functions that run in networks suggests per-virtual-network control- 3.22 Computational Biology and many switches and routers. OpenFlow exploits this lers. This in turn leads to the concept of a network Bioinformatics...... 102 common set of functions.” hypervisor, which plays the same role for multiple 3.23 Medical Robotics...... 108 For the second, there is no more software in network controllers that a hypervisor does for 4. DRIVERS AND DISRUPTORS...... 112 an SDN than in a classic network. An Open- separate virtual machines on a single common 5. TECHNOLOGY COVERAGE IN Flow-based network routes and forwards packets substrate. The prototype network hypervisor is the IEEE XPLORE AND BY IEEE SOCIETIES...... 115 via on-switch hardware, and provides no more FlowVisor, which partitions the matching header services than any other classic L2/L3 network. The spaces among component controllers and permits 6. IEEE COMPUTER SOCIETY IN 2022...... 120 software in an SDN, defining the rules and policies each controller to write rules over its own header 7. SUMMARY AND NEXT STEPS.....124 concerning packet forwarding and transmission, space (the “flowspace”). Rules from the controllers 8. AUTHORS...... 126 APPENDIX I. 23 Technologies Coverage 57 in IEEE Publications...... 134 IEEE CS 2022 , Report DRAFT 23 Technologies in 2022 1/26/2014 5:18 PM

This capability of OpenFlow of isolated virtual networks suggests per-‐virtual-‐network controllers. This in turn leads to the concept of a network hypervisor, which plays the same role for multiple network controllers that a hypervisor does for separate virtual machines on a single common substrate. The prototype twork ne hypervisor is the FlowVisor, which partitions the matching header spaces among component controllers and permits each controller to write rules over its own header space (the “flowspace”). Rules from the controllers are sent to the FlowVisor, which checks to ensure that each controller is writing rules over its own flowspace and then transmits those rules to the switches.

Further, since the network itself is reduced to a collection of forwarding tables in an SDN, verification of network rules is much simpler. Verification of networks is difficult today because the control plane is embedded in the network and consists of switches running a distributed, Turing-‐complete computation; verification of this is undecidable. However, packets are always forwarded by the data plane; checking the packet-‐forwarding rules for compliance with desired properties verifies the network. Mathematically, verifying a forwarding ruleset is identical to verifying a loop-‐free logic circuit. This problem is far easier fying than veri Turing machines (it is NP-‐complete instead of undecidable) and has been successfully attacked over -‐ a 25 year period in the VLSI industry.

3.12.3 Challenges The primary challenge in implementing OpenFlow is that current-‐generation switch hardware was built for a very different networking use case, and existing switch ASIC pipelines fall far short of that required to implement a pure OpenFlow protocol. In order to explore alternatives to traditional forwarding with off-‐the-‐shelf hardware, we need to explain how that hardware is built to exploit the realities of existing CONTENTS forwarding rules and whether it is suitable for new forwarding requirements. 1. INTRODUCTION...... 5 2. SEAMLESS INTELLIGENCE are sent to the FlowVisor, SCENARIO...... 8 which checks to ensure 3. 23 TECHNOLOGIES IN 2022...... 14 that each controller is 3.1 Security Cross-Cutting Issues...... 14 writing rules over its 3.2 The Open Intellectual own flowspace and then Property Movement...... 17 transmits those rules to 3.3 Sustainability...... 20 3.4 Massively Online Open Courses...... 25 the switches. 3.5 Quantum Computing ...... 28 Further, since the net- 3.6 Device and Nanotechnology...... 31 work itself is reduced to 3.7 3D Integrated Circuits...... 33 a collection of forwarding CONTENTS3.8 Universal Memory...... 38 3.9 Multicore...... 43 tables in an SDN, verifi- FigureFigure 12.12 A. representative A representative switch switch ASIC ASIC pipeline.pipeline. 3.10 Photonics...... 47 cation of network rules is In a traditional (pre-‐SDN “legacy”) forwarding L2 environment, basically all switches are the same—they 3.11 Networking and Interconnectivity... 52 much simpler. Verification In order to explore alternatives to traditional for- have slightly different pipelines depending on how versatile a single piece of silicon is (e.g., whether it 3.12 Software-Defined Networks...... 55 of networks is difficult today built because to only the control be a switch, warding or with whether off-the-shelf it can also hardware,rewalls, be sold for load balancers, fi etc.), we need but to they have 3.13 High-Performance Computing plane is embedded in the network and consists of explain how that hardware is built to exploit the

(HPC) ...... 63 switches running a distributed, Turing-complete realities of existing forwarding rules and whether it 3.14 Cloud Computing...... 68 computation; verification of this is undecidable. 3.15 The Internet of Things...... 74 is suitable for new forwarding requirements. However, packets are always forwarded by the 58 3.16 Natural User Interfaces...... 77 In a traditional (pre-SDN “legacy”) L2 forwarding data plane; checking the packet-forwarding rules 3.17 3D Printing ...... 81 environment, basically all switches are the same— for compliance with desired properties verifies the 3.18 Big Data and Analytics ...... 84 they have slightly different pipelines depending 3.19 Machine Learning and network. Mathematically, verifying a forwarding on how versatile a single piece of silicon is (e.g., Intelligent Systems ...... 89 ruleset is identical to verifying a loop-free logic cir- whether it is built to only be a switch, or whether it 3.20 Computer Vision and Pattern cuit. This problem is far easier than verifying Turing Recognition...... 92 can also be sold for load balancers, firewalls, etc.), machines (it is NP-complete instead of undecid- 3.21 Life Sciences...... 96 but they have effectively baked IEEE and IETF able) and has been successfully attacked over a 3.22 Computational Biology and standards into the silicon. To the extent that you Bioinformatics...... 102 25-year period in the VLSI industry. want to do something that is nontraditional, you 3.23 Medical Robotics...... 108 have limited flexibility. 4. DRIVERS AND DISRUPTORS...... 112 3.12.3 Challenges Pre-SDN standard L2 forwarding is fairly sim- 5. TECHNOLOGY COVERAGE IN The primary challenge in implementing OpenFlow IEEE XPLORE AND BY ple and based on MAC learning and L2 destina- is that current-generation switch hardware was IEEE SOCIETIES...... 115 tion-based forwarding. The network device learns built for a very different networking use case, and 6. IEEE COMPUTER SOCIETY what MAC addresses are connected to each port in existing switch ASIC pipelines fall far short of that IN 2022...... 120 the L2 MAC learning table (per-VLAN in a high-end required to implement a pure OpenFlow protocol. 7. SUMMARY AND NEXT STEPS.....124 device) and then builds a table for all forwarding 8. AUTHORS...... 126 APPENDIX I. 23 Technologies Coverage 58 in IEEE Publications...... 134 CONTENTS 1. INTRODUCTION...... 5 2. SEAMLESS INTELLIGENCE decisions based on this information. This results useful for L3 and L4 matching. The mechanical SCENARIO...... 8 in a fairly simple high-capacity forwarding table, implications of adding a “don’t care” bit to every 3. 23 TECHNOLOGIES IN 2022...... 14 which is optimized for a single field (typically, desti- possible part of a match include the very real prob- 3.1 Security Cross-Cutting Issues...... 14 nation MAC address) and no wildcards. lem that the die space required for TCAM is signifi- 3.2 The Open Intellectual cantly larger than that required for CAM, limiting Property Movement...... 17 ASIC pipelines have reasonably strict rules about the amount available (and the expense of table 3.3 Sustainability...... 20 how you get into various tables in the packet flow; space in general). The effective table size is further 3.4 Massively Online Open Courses...... 25 they are somewhat configurable, but we must keep 3.5 Quantum Computing ...... 28 in mind that the vendor-imagined packet flow has limited if L3 matching allows for IPv6, as this now 3.6 Device and Nanotechnology...... 31 a strong fidelity to existing standards. Thus, lever- significantly increases the number of bits required 3.7 3D Integrated Circuits...... 33 aging L3 matching tables often involves a previ- for each match line. CONTENTS3.8 Universal Memory...... 38 3.9 Multicore...... 43 ous stage in the pipeline indicating that a packet Given that existing devices are built for existing 3.10 Photonics...... 47 needs to be routed (often by marking a packet for a L2/L3 protocols, this table is further limited by the 3.11 Networking and Interconnectivity... 52 destination MAC address modification). This means expectation of the vendor. A TCAM is not designed 3.12 Software-Defined Networks...... 55 that if you want to use the L3 table, you must mark for standard protocol-based L3 forwarding—it is 3.13 High-Performance Computing a packet in this way, and if you mark a packet in exclusively intended for use by the user or firm- (HPC) ...... 63 this way, you should be aware that your packet is ware developer to “correct” a limited number of 3.14 Cloud Computing...... 68 now going to traverse the L3 table, which may have standard forwarding behaviors. Because this table 3.15 The Internet of Things...... 74 3.16 Natural User Interfaces...... 77 adverse effects on your real intent. It is also import- most directly maps to the 12-tuple OpenFlow 3.17 3D Printing ...... 81 ant to keep in mind that regardless of the rewrite match structure, it is very commonly the only table 3.18 Big Data and Analytics ...... 84 you may actually want to execute, there are various exposed by the firmware developer to the Open- 3.19 Machine Learning and L3 capabilities that must be preserved, specifically Flow control channel. However, for many reasons, Intelligent Systems ...... 89 in the L3 multicast case, for normal host and switch including those listed above, this table is very small, 3.20 Computer Vision and Pattern Recognition...... 92 function to work properly. making it an ineffective place to make any kind of 3.21 Life Sciences...... 96 As such, it is relevant to discuss common SDN nuanced forwarding decisions at scale. 3.22 Computational Biology and (typically OpenFlow) firmware, which attempt to To some extent, this limitation can be overcome in Bioinformatics...... 102 avoid the majority of these issues the easy way—by a controller by using sophisticated algorithms to 3.23 Medical Robotics...... 108 leveraging the table of last resort, the ACL table. map OpenFlow rules to existing switch hardware 4. DRIVERS AND DISRUPTORS...... 112 This table is a small TCAM, usually a few megabits while preserving semantics, maximizing the use of 5. TECHNOLOGY COVERAGE IN in size, and thus has a very limited match capability cheap, large memories, and minimizing the use of IEEE XPLORE AND BY IEEE SOCIETIES...... 115 (and as discussed previously, the rewrite capabil- TCAMs. However, to date, these algorithms are not ity of the device may also be significantly limited). implemented in any controller, though their use has 6. IEEE COMPUTER SOCIETY IN 2022...... 120 A TCAM is capable of wildcard matches (unlike a been explored in the literature. CAM, which must make exact matches), so is very 7. SUMMARY AND NEXT STEPS.....124 Two other challenges are secure communication 8. AUTHORS...... 126 APPENDIX I. 23 Technologies Coverage 59 in IEEE Publications...... 134 CONTENTS 1. INTRODUCTION...... 5 2. SEAMLESS INTELLIGENCE between the controller and the switch and mak- substantial savings in rulespace are achievable SCENARIO...... 8 ing the controller robust against network failures using these techniques. 3. 23 TECHNOLOGIES IN 2022...... 14 and outages. Secure communication between the We expect significant improvement in the control- 3.1 Security Cross-Cutting Issues...... 14 switch and controller is in the OpenFlow spec, ler space, implementing features in the literature 3.2 The Open Intellectual based on a Transport Layer Security implementa- Property Movement...... 17 and addressing commonly recognized shortcom- tion on both switch and controller. However, few 3.3 Sustainability...... 20 ings. Controller technology has progressed rapidly, commercial switches today implement TLS. 3.4 Massively Online Open Courses...... 25 from Nox, which was a thin overlay on the simple 3.5 Quantum Computing ...... 28 A robust, high-availability, distributed controller OpenFlow API, to Floodlight, a scalable imple- 3.6 Device and Nanotechnology...... 31 is a sine qua non for real OpenFlow deployments: mentation, to OpenDaylight, which incorporates 3.7 3D Integrated Circuits...... 33 no network can take a single point of failure. For- a number of optimizations for specific switch CONTENTS3.8 Universal Memory...... 38 3.9 Multicore...... 43 tunately, the design of high-availability, robust, families. We expect future controllers to have the 3.10 Photonics...... 47 distributed (HARD) software systems is now well HARD properties and to implement the algorithms 3.11 Networking and Interconnectivity... 52 understood. These lessons must be applied to the for both network verification and safe update that 3.12 Software-Defined Networks...... 55 controller space. have appeared in the literature. 3.13 High-Performance Computing (HPC) ...... 63 We also expect that the hypervisor-like capabilities 3.14 Cloud Computing...... 68 3.12.4 Where We Think It Will Go of FlowVisor and FOAM will become an integral 3.15 The Internet of Things...... 74 Many of the shortcomings that we discussed in the part of controller and network operating system 3.16 Natural User Interfaces...... 77 previous section are due to the relative immaturity design. In this picture, the developer of a distrib- 3.17 3D Printing ...... 81 of OpenFlow and the time constants inherent in new uted system will develop a virtual network as an 3.18 Big Data and Analytics ...... 84 ASIC designs. OpenFlow today runs on a generation integral part of his system, where the network con- 3.19 Machine Learning and Intelligent Systems ...... 89 of switches designed for a far different use case; a trol is simply a part of his application. The network 3.20 Computer Vision and Pattern new generation of switches, with flexible pipelines operating system will then check to ensure that his Recognition...... 92 and larger TCAMs, will run OpenFlow more efficiently generated rules work only over his virtual network 3.21 Life Sciences...... 96 and far more effectively. Already, a number of existing and will mediate communication between the 3.22 Computational Biology and vendors and startups are working on switches opti- physical network and application. Bioinformatics...... 102 mized for OpenFlow deployment. 3.23 Medical Robotics...... 108 If this picture seems exotic, note that it already ex- 4. DRIVERS AND DISRUPTORS...... 112 The new generation of switches will be aided by ists for all other system devices. In today’s world, a 5. TECHNOLOGY COVERAGE IN improvements in controller technology, designed system developer creates and manipulates a virtual IEEE XPLORE AND BY to optimize the ruleset. As we mentioned above, filesystem for his application, consisting of the files IEEE SOCIETIES...... 115 optimization technologies based on the algorithms and directories he reads and writes over the course 6. IEEE COMPUTER SOCIETY used to minimize digital logic circuits are known of the application. This capability will simply bring IN 2022...... 120 but not yet incorporated in existing controllers. the network up to the programmability of other 7. SUMMARY AND NEXT STEPS.....124 Preliminary investigations have indicated that application resources. 8. AUTHORS...... 126 APPENDIX I. 23 Technologies Coverage 60 in IEEE Publications...... 134 CONTENTS 1. INTRODUCTION...... 5 2. SEAMLESS INTELLIGENCE We expect that the controller API will become switch market is today dominated by complex, ex- SCENARIO...... 8 standardized, much as the operating system API pensive devices that are highly feature-rich. In the 3. 23 TECHNOLOGIES IN 2022...... 14 has become standardized through POSIX, and a future, that functionality will move to the controller, 3.1 Security Cross-Cutting Issues...... 14 number of different implementations with a com- and switches and routers will become commodi- 3.2 The Open Intellectual mon API will emerge. ty devices. Virtually every control plane protocol Property Movement...... 17 can be realized in a logically centralized software 3.3 Sustainability...... 20 Finally, we expect that the next few years will see controller running on a standardized platform. This 3.4 Massively Online Open Courses...... 25 the integration of the controller API with a distrib- 3.5 Quantum Computing ...... 28 uted cloud controller, which will site virtual ma- is a far more developer- and administrator-friendly 3.6 Device and Nanotechnology...... 31 chines across a wide-area computing fabric. This is environment, offering far greater visibility, control- 3.7 3D Integrated Circuits...... 33 similar to the existing GENI mesoscale deployment lability, and verifiability over the network. CONTENTS3.8 Universal Memory...... 38 3.9 Multicore...... 43 in the United States, and we believe that this will SDN is likely to happen first at the edges of the 3.10 Photonics...... 47 become standardized and ubiquitous. The fabric of network rather than in the telecom operators’ big 3.11 Networking and Interconnectivity... 52 the future will be a network of virtual machines, sit- networks (although a few are experimenting with 3.12 Software-Defined Networks...... 55 ed close to data sources and users, interconnected SDN). Furthermore, SDN goes hand in hand with 3.13 High-Performance Computing by SDN-enabled virtual networks. NFV, or network function virtualization. They are (HPC) ...... 63 not the same but are likely to leverage one another. 3.14 Cloud Computing...... 68 3.15 The Internet of Things...... 74 3.12.5 Potential Disruptions 3.16 Natural User Interfaces...... 77 From the foregoing, it’s clear that SDNs will be 3.12.6 Summary 3.17 3D Printing ...... 81 the most disruptive force in networking since the OpenFlow and SDN are the greatest advances in 3.18 Big Data and Analytics ...... 84 standardization of the OSI stack in the late 1970s. networking in a generation, and will change the 3.19 Machine Learning and Intelligent Systems ...... 89 SDNs are the perfect complement to the OSI stack fundamental activity from configuring the network 3.20 Computer Vision and Pattern in the sense that they introduce a set of abstrac- to programming it. This will make the network far Recognition...... 92 tions for the control plane to match the OSI stack’s more secure, transparent, flexible, verifiable, and 3.21 Life Sciences...... 96 set of abstractions for the control plane. However, functional. Fully achieving this promise will take 3.22 Computational Biology and the impact goes far beyond this evident symme- some years; a new generation of switches must Bioinformatics...... 102 try. OpenFlow and SDN herald a disruption of the emerge, and the promise of SDNs must be incorpo- 3.23 Medical Robotics...... 108 networking market similar to the disruption in the rated into the controllers. This will not be automatic 4. DRIVERS AND DISRUPTORS...... 112 server market caused by the introduction of the or quick, but we know how to do it, and over time, 5. TECHNOLOGY COVERAGE IN Linux OS on the x86 platform. Within a decade, the these changes will take place and SDNs will ex- IEEE XPLORE AND BY IEEE SOCIETIES...... 115 Linux-on-x86 had largely displaced the proprietary ceed even the high expectations of today. server stacks that had dominated the IT world: 6. IEEE COMPUTER SOCIETY IN 2022...... 120 Solaris-on-SPARC, HPUX-on-PA-RISC, AIX-on- POWER, etc. In a similar sense, the router and 7. SUMMARY AND NEXT STEPS.....124 8. AUTHORS...... 126 APPENDIX I. 23 Technologies Coverage 61 in IEEE Publications...... 134 CONTENTS 1. INTRODUCTION...... 5 2. SEAMLESS INTELLIGENCE 3.12.7 References [8] N. Foster et al., “Frenetic: A Network Programming SCENARIO...... 8 [1] N. McKeown et al., “OpenFlow: Enabling Inno- Language,” Proc. 16th ACM SIGPLAN Int’l Conf. Func- 3. 23 TECHNOLOGIES IN 2022...... 14 vation in Campus Networks,” SIGCOMM Computer tional Programming (ICFP ‘11), ACM, 2011, pp. 279- 3.1 Security Cross-Cutting Issues...... 14 Communications Rev., vol. 38, no. 2, 2008, pp. 69- 291; DOI=10.1145/2034773.2034812 http://doi.acm. 3.2 The Open Intellectual 74; DOI=10.1145/1355734.1355746 http://doi.acm. org/10.1145/2034773.2034812. Property Movement...... 17 org/10.1145/1355734.1355746. [9] P. Kazemian, G. Varghese, and N. McKeown, “Header 3.3 Sustainability...... 20 Space Analysis: Static Checking for Networks,” NSDI, 3.4 Massively Online Open Courses...... 25 [2] S. Zhang, S. Malik, and R. McGeer, “Verification 2012. 3.5 Quantum Computing ...... 28 of Computer Switching Networks: An Overview,” 3.6 Device and Nanotechnology...... 31 Proc. 10th Int’l Conf. Automated Technology for Ver- [10] M. Reitblatt et al., “Abstractions for Network Up- 3.7 3D Integrated Circuits...... 33 ification and Analysis (ATVA’12), S. Chakraborty and date,” Proc. ACM SIGCOMM 2012 Conf. Applications, CONTENTS3.8 Universal Memory...... 38 M. Mukund, eds., Springer-Verlag, 2012, pp. 1-16; Technologies, Architectures, and Protocols for Computer 3.9 Multicore...... 43 DOI=10.1007/978-3-642-33386-6_1 http://dx.doi. Communication, ACM, 2012. 3.10 Photonics...... 47 org/10.1007/978-3-642-33386-6_1. 3.11 Networking and Interconnectivity... 52 [11] S. Gutz et al., “Splendid Isolation: A Slice Abstrac- [3] A. Khurshid et al., “VeriFlow: Verifying Network-Wide 3.12 Software-Defined Networks...... 55 tion for Software-Defined Networks,” Proc. 1st Work- Invariants in Real Time,” Proc. 10th USENIX Conf. Net- 3.13 High-Performance Computing shop on Hot Topics in Software Defined Networks, ACM, (HPC) ...... 63 worked Systems Design and Implementation (nsdi’13), N. 2012. 3.14 Cloud Computing...... 68 Feamster and J. Mogul, eds., USENIX Association, 2013, pp. 15-28. [12] R. McGeer, “A Safe, Efficient Update Protocol for 3.15 The Internet of Things...... 74 OpenFlow Networks,” Proc. 1st Workshop on Hot Topics 3.16 Natural User Interfaces...... 77 [4] R. McGeer and P. Yalagandula, “Minimizing Rulesets in Software Defined Networks, ACM, 2012. 3.17 3D Printing ...... 81 for TCAM Implementation,” INFOCOM 2009, IEEE, 2009. 3.18 Big Data and Analytics ...... 84 [13] R. McGeer, “Verification of Switching Network 3.19 Machine Learning and [5] S. Shenker, “A Gentle Introduction to Software-De- Properties Using Satisfiability,” IEEE Int’l Conf. Commu- Intelligent Systems ...... 89 fined Networking,” Lecture at Technion, 2012; http:// nications (ICC), IEEE, 2012. 3.20 Computer Vision and Pattern www.youtube.com/watch?v=eXsCQdshMr4. Recognition...... 92 [14] The GENI Project. http://www.geni.net 3.21 Life Sciences...... 96 [6] R. Sherwood et al., “FlowVisor: A Network Virtualiza- 3.22 Computational Biology and tion Layer,” OpenFlow Switch Consortium, Tech. Rep., [15] Saracco, R., Personal Communication, 2014. Bioinformatics...... 102 2009. 3.23 Medical Robotics...... 108 4. DRIVERS AND DISRUPTORS...... 112 5. TECHNOLOGY COVERAGE IN IEEE XPLORE AND BY IEEE SOCIETIES...... 115 6. IEEE COMPUTER SOCIETY IN 2022...... 120 7. SUMMARY AND NEXT STEPS.....124 8. AUTHORS...... 126 APPENDIX I. 23 Technologies Coverage 62 in IEEE Publications...... 134 CONTENTS 1. INTRODUCTION...... 5 2. SEAMLESS INTELLIGENCE amounts of data, which is stored possibly in hi- SCENARIO...... 8 erarchies and across sites. Because of the large 3. 23 TECHNOLOGIES IN 2022...... 14 processing power, these machines can be liquid- 3.1 Security Cross-Cutting Issues...... 14 cooled instead of the traditional air-cooling. 3.2 The Open Intellectual Property Movement...... 17 At the systems software layer, operating systems 3.3 Sustainability...... 20 are optimized to reduce any noise, to enable par- 3.4 Massively Online Open Courses...... 25 allelism. The presence of noise (various daemons, 3.5 Quantum Computing ...... 28 TLB and cache flushes, etc.) accumulates and 3.6 Device and Nanotechnology...... 31 3.13 High-Performance aggregates into delay, preventing applications to 3.7 3D Integrated Circuits...... 33 Computing (HPC) scale. HPC operating systems are usually stripped CONTENTS3.8 Universal Memory...... 38 3.9 Multicore...... 43 down or work with lean microkernels and runtimes. 3.10 Photonics...... 47 3.13.1 Introduction 3.11 Networking and Interconnectivity... 52 High-performance computing is a sector that 3.13.2 State of the Art 3.12 Software-Defined Networks...... 55 entails hardware, systems software and tools, State of the art at the very high end of the HPC 3.13 High-Performance Computing and applications/services. It is strategically (HPC) ...... 63 field leap-frogs between the vendors exchanging 3.14 Cloud Computing...... 68 important to many areas in industry, such as the lead in the 10-Pflop range of supercomputers. 3.15 The Internet of Things...... 74 biotechnology, chemical, life sciences, pharma- The most recent entrants on the list are computers 3.16 Natural User Interfaces...... 77 ceutical, national security and homeland de- from China, including the current top computer 3.17 3D Printing ...... 81 fense, automotive, gas and oil, financial, weather 500 computer Tianhe-2, from the National Univer- 3.18 Big Data and Analytics ...... 84 forecasting, computer-aided engineering, and sity of Defense Technology, with almost 55 Pflops 3.19 Machine Learning and many others. Multiple vendors develop and sell Intelligent Systems ...... 89 peak and over 3 million cores. (Reference top 500). HPC equipment, such as Bull, Cray, DDN, Fujitsu, 3.20 Computer Vision and Pattern An even more interesting race becomes in the Recognition...... 92 HP, IBM, Intel, Mellanox, NEC, NVIDIA, and SGI, power efficiency of these computers. Similar to 3.21 Life Sciences...... 96 to name a few. The market is still divided based 3.22 Computational Biology and on server cost (from over $500,000 to less than the large datacenters that host Internet providers Bioinformatics...... 102 $100,000) into supercomputers, division, depart- (Google, Yahoo, etc.), the operating costs start to 3.23 Medical Robotics...... 108 ment, and workgroup servers. outweigh the capital costs. The primary contribu- 4. DRIVERS AND DISRUPTORS...... 112 tor to operating costs is power consumption. Peak At the hardware layer, HPC systems are typically 5. TECHNOLOGY COVERAGE IN usage becomes a bottleneck as it becomes hard to IEEE XPLORE AND BY designed from compute-intensive processor farms bring sufficient power to supercomputers in certain IEEE SOCIETIES...... 115 with memories as large as possible (depending geographical areas. on application footprint) and specially designed 6. IEEE COMPUTER SOCIETY At the lower end of HPC, general-purpose graphics IN 2022...... 120 interconnects to enable low latency and high bandwidth. Storage is optimized to receive large processing units (GPGPUs) are turning worksta- 7. SUMMARY AND NEXT STEPS.....124 tions into supercomputers. By including GPGPUs 8. AUTHORS...... 126 APPENDIX I. 23 Technologies Coverage 63 in IEEE Publications...... 134 IEEE CS 2022 , Report DRAFT 23 Technologies in 2022 1/26/2014 5:18 PM

programming relies on explicit memory equires sharing and r programming wizards to extract parallelism at scale.

File systems have been optimized to sustain high-‐bandwidth throughput—for example, most of the Lustre file system’s design is to work around the bottleneck of disk performance and its mechanical parts (moving head). This will likely -‐ change as solid state disks and other new NVM technologies get CONTENTS introduced. 1. INTRODUCTION...... 5 2. SEAMLESS INTELLIGENCE in personal servers and SCENARIO...... 8 even laptops, it is pos- 3. 23 TECHNOLOGIES IN 2022...... 14 sible to have the power 3.1 Security Cross-Cutting Issues...... 14 of a supercomputer 3.2 The Open Intellectual by using a high level of Property Movement...... 17 parallelism. 3.3 Sustainability...... 20 3.4 Massively Online Open Courses...... 25 Current programming 3.5 Quantum Computing ...... 28 models still rely on estab- 3.6 Device and Nanotechnology...... 31 lished libraries, such as 3.7 3D Integrated Circuits...... 33 MPI (message-passing CONTENTS3.8 Universal Memory...... 38 3.9 Multicore...... 43 interface), although the 3.10 Photonics...... 47 degrees of parallelism will 3.11 Networking and Interconnectivity... 52 require better suited and 3.12 Software-Defined Networks...... 55 finer granularity sharing. 3.13 High-Performance Computing CUDA programming Figure 13 . High-‐performance computing. Figure 13. High-performance computing. (HPC) ...... 63 relies on explicit memory 3.14 Cloud Computing...... 68 3.13.3 Challenges sharing and requires pro- 3.15 The Internet of Things...... 74 The top challenges for HPC for exascale are as follows. would require 50- to 60-fold pow- gramming wizards to extract parallelism at scale. 3.16 Natural User Interfaces...... 77 Scaling within power limits is the er first savings and compared largest challenge, to today’s followed systems closely power by reaching the next levels 3.17 3D Printing ...... 81 File systems have been optimizedof performance, to sustain such as consumption. exascale. This can only be accomplished by careful optimization at all levels. 3.18 Big Data and Analytics ...... 84 Reaching the power budget of 20 MW for exascale would require 50-‐ to 60-‐fold power savings high-bandwidth throughput—forcompared example, to most today’s of systems Interconnect power consumption. bottlenecks are the second major 3.19 Machine Learning and the Lustre file system’s design is to work around Intelligent Systems ...... 89 Interconnect bottlenecks are the challenge. second major With increased challenge. scale, With increased there is scale, a greater there is a greater need 3.20 Computer Vision and Pattern the bottleneck of disk performanceto communicate and its me across - many need systems, to communicate which puts across stress many on d interconnect latency an systems, bandwidth. which Recognition...... 92 chanical parts (moving head). ThisPhotonic will interconnects likely offer puts potential stress here on interconnect3.10 (see Section ). latency and bandwidth. 3.21 Life Sciences...... 96 change as solid-state disks andLow other-‐noise new system NVM software that Photonic enables applications interconnects to increase offer potential parallelism here has (see a similar importance to 3.22 Computational Biology and technologies get introduced. interconnects. At the scale of as 100,000 nodes predicted for exascale, the frequency of failures will be Bioinformatics...... 102 Section 3.10).

3.23 Medical Robotics...... 108 Low-noise system software that enables applica- 4. DRIVERS AND DISRUPTORS...... 112 3.13.3 Challenges tions to increase parallelism has a similar impor- 5. TECHNOLOGY COVERAGE IN The top challenges for HPC are as follows. tance to interconnects.64 At the scale of 100,000 IEEE XPLORE AND BY Scaling within power limits is the first and largest nodes as predicted for exascale, the frequency of IEEE SOCIETIES...... 115 challenge, followed closely by reaching the next failures will be much higher, while requirement to 6. IEEE COMPUTER SOCIETY levels of performance, such as exascale. This can retain similar reliability to today’s systems will re- IN 2022...... 120 only be accomplished by careful optimization at main. This will require new approaches to reliability 7. SUMMARY AND NEXT STEPS.....124 all levels. Reaching the power budget of 20 MW at all levels of the system. 8. AUTHORS...... 126 APPENDIX I. 23 Technologies Coverage 64 in IEEE Publications...... 134 IEEE CS 2022 , Report DRAFT 23 Technologies in 2022 1/26/2014 5:18 PM CONTENTS 1. INTRODUCTION...... 5 2. SEAMLESS INTELLIGENCE SCENARIO...... 8 3. 23 TECHNOLOGIES IN 2022...... 14 3.1 Security Cross-Cutting Issues...... 14 3.2 The Open Intellectual Property Movement...... 17 3.3 Sustainability...... 20 3.4 Massively Online Open Courses...... 25 3.5 Quantum Computing ...... 28 3.6 Device and Nanotechnology...... 31 3.7 3D Integrated Circuits...... 33 CONTENTS3.8 Universal Memory...... 38 3.9 Multicore...... 43 3.10 Photonics...... 47 3.11 Networking and Interconnectivity... 52 3.12 Software-Defined Networks...... 55 3.13 High-Performance Computing (HPC) ...... 63 Figure Figure14 14.. Comparing Comparing classes classes of HPC of and HPC their and feasibility their to feasibility deliver in the to cloud. deliver in the cloud. 3.14 Cloud Computing...... 68 3.15 The Internet of Things...... 74 Ease 3.13.5of use of programmingPotential languages Disruptions and tools with the systems software supporting them will be 3.16 Natural User Interfaces...... 77 that canSome enable technology applications innovations to be more could intuitively disrupt critical the for rise enabling HPC and in leveraging particular, -‐ but also general purpose dark silicon. 3.17 3D Printing ...... 81 writtencomputing. in a parallel fashion are required to enrich Finally, managing the complexity at a large scale, as 3.18 Big Data and Analytics ...... 84 today’s MPI versus shared-memory (multi-threads) well as heterogeneity of CPUs, storage, and soft- 3.19 Machine Learning and models.Low -‐power components, such as ARM processors, are increasingly used to build high-‐end computers. Intelligent Systems ...... 89 They represent a perfect match for the Innovator’s Dilemma ware components model, will where low-‐cost be a challenge products to enter keep the these 3.20 Computer Vision and Pattern New market applications with and lower algorithms characteristics that can lever (performance, - large systems robustness, operational. efficiency, etc.) and start gaining market Recognition...... 92 age parallelism,share based computational on cost. power, As and they large improve in quality, they push the existing market leaders up the food chain. 3.21 Life Sciences...... 96 amountsCoprocessors of available and memory accelerators arealso needed fall in to evolve this category 3.13.4 and Where can We likewise Think It disrupt Will Go the state of the art 3.22 Computational Biology and existing applications and algorithms designed Bioinformatics...... 102 technology. They are important for power Exascale savings is a and goal set for by many optimizing governments underlying hardware for specific in many years ago. Uncertainty quantification, com- 3.23 Medical Robotics...... 108 applications and workloads. the US, Europe, and Asia. Many teams are trying bustion, and many new fields will be enabled, but to achieve this level of computing at a predefined 4. DRIVERS AND DISRUPTORS...... 112 Nonvolatile m emory (NVM) is truly disruptive to computing in general but especially to HPC (see scientist need to learn how to reason about them power envelope—for example, the US Department 5. TECHNOLOGY COVERAGE IN and howSection to program3.8 ). NVM -‐inbased these new benefits fields, for which HPC are in terms -‐ of checkpoint restart, file systems, and memory IEEE XPLORE AND BY of Energy has issued a target of 20 MW. size. The long-‐term execution of ications HPC appl and the inevitable failures that increase with the IEEE SOCIETIES...... 115 were unfeasible to even consider with previous infrastructure’s increasing complexity and scale At the require other end checkpoint of the spectrum, applications a lot of that can then restart HPC 6. IEEE COMPUTER SOCIETY generations of IT. from a previous checkpoint in case of scientists failures. are Because pushing to of move the HPC nonvolatility, memory checkpointing to the is cloud. not IN 2022...... 120 Heterogeneity of the infrastructure with newly intro- needed anymore, only the nonvolatile state, This makes such sense as for flushing embarrassingly caches and parallel processor state. This will 7. SUMMARY AND NEXT STEPS.....124 ducedsubstantially accelerators reduce and how to both maintain checkpoint compatibility and restore times. The files produced as a result of HPC will be 8. AUTHORS...... 126 much more quickly written, removing the bottleneck introduced by disks. Finally, the larger memories APPENDIX I. 23 Technologies Coverage enabled by new NVM technologies will enable new classes of algorithms, replacing previous algorithms 65 in IEEE Publications...... 134 that had to write the data to disks.

Photonic interconnects are critical to HPC because they will enable lower latency and higher bandwidth, decreasing the delays due to communication across parallel components in HPC systems. In addition, photonics will increase the scale and reduce power consumption.

66 CONTENTS 1. INTRODUCTION...... 5 2. SEAMLESS INTELLIGENCE applications or for development at a smaller scale, optimizing underlying hardware for specific appli- SCENARIO...... 8 but at the larger scale and, in particular, for finer cations and workloads. 3. 23 TECHNOLOGIES IN 2022...... 14 granularity sharing, current virtualization technol- Nonvolatile memory (NVM) is truly disruptive 3.1 Security Cross-Cutting Issues...... 14 ogies and multitenancy introduce too much noise, to computing in general but especially to HPC 3.2 The Open Intellectual preventing HPC applications from scaling. Property Movement...... 17 (see Section 3.8). NVM-based benefits for HPC 3.3 Sustainability...... 20 The figure below classifies different types of HPC are in terms of checkpoint-restart, file systems, 3.4 Massively Online Open Courses...... 25 systems and applications, indicating that the very and memory size. The long-term execution of 3.5 Quantum Computing ...... 28 high-end HPC applications will likely remain ex- HPC applications and the inevitable failures that 3.6 Device and Nanotechnology...... 31 ecuting on dedicated supercomputers, just like increase with the infrastructure’s increasing com- 3.7 3D Integrated Circuits...... 33 some of the top banks still run on mainframes, plexity and scale require checkpoint applications CONTENTS3.8 Universal Memory...... 38 3.9 Multicore...... 43 while the lower-end applications will likely move to that can then restart from a previous checkpoint in 3.10 Photonics...... 47 the cloud. The remaining “in between” applications case of failures. Because of the nonvolatility, mem- 3.11 Networking and Interconnectivity... 52 have the potential to move to the cloud, assuming ory checkpointing is not needed anymore, only 3.12 Software-Defined Networks...... 55 that it is adapted for HPC applications in terms of the nonvolatile state, such as flushing caches and 3.13 High-Performance Computing improved interconnects and virtualization (multiple processor state. This will substantially reduce both (HPC) ...... 63 references). checkpoint and restore times. The files produced 3.14 Cloud Computing...... 68 as a result of HPC will be much more quickly writ- 3.15 The Internet of Things...... 74 3.16 Natural User Interfaces...... 77 3.13.5 Potential Disruptions ten, removing the bottleneck introduced by disks. 3.17 3D Printing ...... 81 Some technology innovations could disrupt the Finally, the larger memories enabled by new NVM 3.18 Big Data and Analytics ...... 84 rise HPC in particular, but also general-purpose technologies will enable new classes of algorithms, 3.19 Machine Learning and computing. replacing previous algorithms that had to write the Intelligent Systems ...... 89 data to disks. 3.20 Computer Vision and Pattern Low-power components, such as ARM pro- Recognition...... 92 cessors, are increasingly used to build high-end Photonic interconnects are critical to HPC be- 3.21 Life Sciences...... 96 computers. They represent a perfect match for cause they will enable lower latency and higher 3.22 Computational Biology and bandwidth, decreasing the delays due to communi- Bioinformatics...... 102 the Innovator’s Dilemma model, where low-cost cation across parallel components in HPC systems. 3.23 Medical Robotics...... 108 products enter the market with lower character- In addition, photonics will increase the scale and 4. DRIVERS AND DISRUPTORS...... 112 istics (performance, robustness, efficiency, etc.) reduce power consumption. 5. TECHNOLOGY COVERAGE IN and start gaining market share based on cost. IEEE XPLORE AND BY As they improve in quality, they push the existing Big data analytics continues to be a disruptor for IEEE SOCIETIES...... 115 market leaders up the food chain. Coprocessors many technologies, and HPC is no exception. Tradi- 6. IEEE COMPUTER SOCIETY and accelerators also fall in this category and can tionally, HPC was both a producer and consumer of IN 2022...... 120 likewise disrupt the state of the art of technolo- big data. However, new techniques and algorithms 7. SUMMARY AND NEXT STEPS.....124 gy. They are important for power savings and for for big data analytics may be considered for use 8. AUTHORS...... 126 APPENDIX I. 23 Technologies Coverage 66 in IEEE Publications...... 134 CONTENTS 1. INTRODUCTION...... 5 2. SEAMLESS INTELLIGENCE in HPC. One variant of this is the ability to per- exascale. Yet, the next set of challenges will remain in SCENARIO...... 8 form analytics (or any application processing) programming models at the new levels of scale. 3. 23 TECHNOLOGIES IN 2022...... 14 in real time. As hardware capabilities increase, it 3.1 Security Cross-Cutting Issues...... 14 will be increasingly possible to execute many more 3.13.7 References 3.2 The Open Intellectual algorithms in real time or near-real time. There Property Movement...... 17 [1] http://www.top500.org/ are many examples, such as simulations, financial 3.3 Sustainability...... 20 [2] http://www.green500.org/ 3.4 Massively Online Open Courses...... 25 what-if analyses, fraud detection, etc., that can [3] Summary Report of the Advanced Scientific Com- 3.5 Quantum Computing ...... 28 dramatically change the way in which business is puting Advisory Committee (ASCAC), “The Oppor- 3.6 Device and Nanotechnology...... 31 conducted in many market segments and verticals. tunities and Challenges of Exascale Computing,” 3.7 3D Integrated Circuits...... 33 In many cases, real time is back-end processing Fall 2010. CONTENTS3.8 Universal Memory...... 38 whose results are leveraged in real time, similarly [4] “Magellan Final Report,” U.S. Department of Energy 3.9 Multicore...... 43 to how Google prepares serialization of all pages (DOE), Tech. Rep., 2011. 3.10 Photonics...... 47 on the Internet and then searches the serialized [5] P. Mehrotra et al., “Performance Evaluation of 3.11 Networking and Interconnectivity... 52 structure rather than parsing the Internet. In this Amazon EC2 for NASA HPC Applications,” Proc. 3.12 Software-Defined Networks...... 55 3rd Workshop on Scientific Cloud Computing, ACM, way, a lot of data can be precomputed, such as in- 3.13 High-Performance Computing 2012, pp. 41–50. (HPC) ...... 63 surance quotes, medical results, and others, which [6] K.R. Jackson et al., “Performance Analysis of High 3.14 Cloud Computing...... 68 will further increase the demand for technology Performance Computing Applications on the Ama- 3.15 The Internet of Things...... 74 improvements and enable more functionality to be zon Web Services Cloud,” CloudCom’10, 2010. 3.16 Natural User Interfaces...... 77 executed in real time. [7] “High Performance Computing (HPC) on AWS,” 3.17 3D Printing ...... 81 http://aws.amazon.com/hpc-applications. 3.18 Big Data and Analytics ...... 84 Third-world countries entering the race for HPC [8] T. Hoefler, T. Schneider, and A. Lumsdaine, “Char- 3.19 Machine Learning and will set new types of requirements for HPC, such acterizing the Influence of System Noise on Large- Intelligent Systems ...... 89 as stringent power consumption guidelines, new 3.20 Computer Vision and Pattern Scale Applications by Simulation,” Supercomputing Recognition...... 92 cooling technologies in extremely hot countries, 10, 2010. 3.21 Life Sciences...... 96 different kinds of support and reliability, etc. [9] Gupta et al., “The Who, What, Why and How of High 3.22 Computational Biology and Performance Computing Applications in the Cloud,” Bioinformatics...... 102 3.13.6 Summary Proc. CloudCom 2013, Best Paper Award; Also avail- 3.23 Medical Robotics...... 108 able at http://www.hpl.hp.com/techreports/2013/ High-performance computing is still leading the HPL-2013-49.html. 4. DRIVERS AND DISRUPTORS...... 112 advances in computing, but it is also being commod- 5. TECHNOLOGY COVERAGE IN itized. Power bottlenecks are becoming the biggest IEEE XPLORE AND BY IEEE SOCIETIES...... 115 challenge for advancing the state of the art. But new advances in NVM, photonics, and integrated 6. IEEE COMPUTER SOCIETY IN 2022...... 120 circuits (see Sections 3.8, 3.10, and 3.7 respectively) are promising for overcoming new barriers, such as 7. SUMMARY AND NEXT STEPS.....124 8. AUTHORS...... 126 APPENDIX I. 23 Technologies Coverage 67 in IEEE Publications...... 134 CONTENTS 1. INTRODUCTION...... 5 2. SEAMLESS INTELLIGENCE to accommodate peak demand at any time of day SCENARIO...... 8 or season. The result for Amazon was that many 3. 23 TECHNOLOGIES IN 2022...... 14 machines in its datacenters were idle most of the 3.1 Security Cross-Cutting Issues...... 14 time and only activated during hours of peak de- 3.2 The Open Intellectual mand. The opportunity to leverage these expensive Property Movement...... 17 resources to utilize them for computing services 3.3 Sustainability...... 20 3.4 Massively Online Open Courses...... 25 during off peak hours became obvious. The result 3.5 Quantum Computing ...... 28 was a revolution in computing, where computing 3.6 Device and Nanotechnology...... 31 3.14 Cloud Computing resources became available as a utility service over CONTENTS3.7 3D Integrated Circuits...... 33 the network. 3.8 Universal Memory...... 38 3.14.1 Introduction 3.9 Multicore...... 43 3.14.2 State of the Art 3.10 Photonics...... 47 Computing has transitioned from centralized main- 3.11 Networking and Interconnectivity... 52 frames to clients networked locally to servers to The literature often implies that cloud computing 3.12 Software-Defined Networks...... 55 the current generation of Web services and mobile is no more than virtualization, or that it is simply 3.13 High-Performance Computing applications using service-oriented architectures the next generation of automated hosting services. (HPC) ...... 63 (SOA). A new generation of computing technology But cloud computing is actually the outcome of 3.14 Cloud Computing...... 68 is emerging rapidly, in which computing is made many advances in computing and communication 3.15 The Internet of Things...... 74 3.16 Natural User Interfaces...... 77 available as a virtualized resource, accessible via a technologies, particularly through virtualization 3.17 3D Printing ...... 81 network. This emerging technology is called cloud that unleashes new opportunities for automation 3.18 Big Data and Analytics ...... 84 computing. It is a response to a need for simplifying of what traditionally used to be manual tasks. The 3.19 Machine Learning and the administration and management of physical result is the ability to build IT applications without Intelligent Systems ...... 89 computing resources in order to focus on business having to endure the long cycle of ordering and 3.20 Computer Vision and Pattern Recognition...... 92 logic and utility of the computing resources. deploying equipment, setting up physical space for it locally or remotely, and having to babysit it 24x7 3.21 Life Sciences...... 96 One may argue justifiably that cloud computing has to keep it running, while still dealing with mundane 3.22 Computational Biology and existed since the early 1990s in the form of SOAs Bioinformatics...... 102 issues such as power, cooling, and depreciation. In and network accessed services. There is, however, 3.23 Medical Robotics...... 108 addition, with cloud computing services, IT leaders a new innovation that comes with today’s cloud 4. DRIVERS AND DISRUPTORS...... 112 gain a new option for investing in IT equipment, computing—it combines the virtualization of com- which has become simplified to a pay-as-you-go 5. TECHNOLOGY COVERAGE IN puting resources at all levels through automation, IEEE XPLORE AND BY model, meaning you only pay for the resources making these resources available for assembly and IEEE SOCIETIES...... 115 you use when you use them, enjoying the ability use remotely via networks. This definition of cloud 6. IEEE COMPUTER SOCIETY to scale your resources to support increased or computing was first invented at Amazon. Retail IN 2022...... 120 decreased demand programmatically and almost business is seasonal, and datacenters are designed 7. SUMMARY AND NEXT STEPS.....124 instantaneously. 8. AUTHORS...... 126 APPENDIX I. 23 Technologies Coverage 68 in IEEE Publications...... 134 CONTENTS 1. INTRODUCTION...... 5 2. SEAMLESS INTELLIGENCE The term cloud computing has become a marketing PaaS SCENARIO...... 8 catchall, but it is, in fact, a new technology. We of- PaaS offers a virtualized substrate that appears as 3. 23 TECHNOLOGIES IN 2022...... 14 fer a definition here that we hope will further clarify if it were an integrated single system; the compute, 3.1 Security Cross-Cutting Issues...... 14 it, but we are certain that it will increase the debate storage, and network configuration details are 3.2 The Open Intellectual as to what it is. Property Movement...... 17 further hidden from the developer, and the devel- 3.3 Sustainability...... 20 oper is only concerned with working on that cloud 3.4 Massively Online Open Courses...... 25 Definition computing platform as if it were a server platform. 3.5 Quantum Computing ...... 28 Cloud computing is based on a datacenter-scale The toolset gives the developer access for spec- 3.6 Device and Nanotechnology...... 31 virtualization of computing resources, in which ifying the layout and relationship between the 3.7 3D Integrated Circuits...... 33 through the collective automation of these virtu- components that make up a workload such as the CONTENTS3.8 Universal Memory...... 38 alized resources, a virtualized subset of compute, relationship between the front-end, the application, 3.9 Multicore...... 43 and the back-end layers. Programming becomes 3.10 Photonics...... 47 storage, connectivity, and application/middleware 3.11 Networking and Interconnectivity... 52 services are carved out to serve as a virtualized specific to the respective cloud computing plat- 3.12 Software-Defined Networks...... 55 computing substrate accessed via a network. form, and its portability is limited. Windows Azure 3.13 High-Performance Computing from Microsoft is a good example of this approach. (HPC) ...... 63 There are several forms of manifestation of cloud computing, including infrastructure as a service 3.14 Cloud Computing...... 68 SaaS 3.15 The Internet of Things...... 74 (IaaS), platform as a service (PaaS), and software 3.16 Natural User Interfaces...... 77 as a service (SaaS). They all share the above defini- SaaS offers virtualized finished application/middle- 3.17 3D Printing ...... 81 tion in that each offers access to a virtualized com- ware services. The developer is offered a running, 3.18 Big Data and Analytics ...... 84 puting substrate at a different level of abstraction. virtualized version of the workload required and 3.19 Machine Learning and provisions it to make it accessible to his/her end Intelligent Systems ...... 89 users. If it were just middleware, then the develop- 3.20 Computer Vision and Pattern IaaS er can use the middleware API to receive services Recognition...... 92 IaaS offers a virtualized substrate where the indi- that can be integrated with some other workload 3.21 Life Sciences...... 96 vidual compute, storage, and connectivity resourc- 3.22 Computational Biology and he/she is developing. Google App Engine and es are made more visible to the workload develop- Bioinformatics...... 102 Force.com from Salesforce.com are good examples er. As such, the toolset a developer uses allows for 3.23 Medical Robotics...... 108 of this approach. 4. DRIVERS AND DISRUPTORS...... 112 the instantiation of specific virtual machines with specific versions of an operating system, size of Cloud computing is also about employing auto- 5. TECHNOLOGY COVERAGE IN mation to raise the level of resource utilization in IEEE XPLORE AND BY memory and type of mass storage, network attach- IEEE SOCIETIES...... 115 ment, and network subnets and VLANs all auto- a datacenter to significantly higher levels than 6. IEEE COMPUTER SOCIETY mated to create a mini-datacenter for a computing what traditionally is possible, from as little as 15 IN 2022...... 120 workload. Amazon’s AWS is a good example of this percent without resource virtualization to as much 7. SUMMARY AND NEXT STEPS.....124 approach. as 50 percent with it. Furthermore, utilization can 8. AUTHORS...... 126 APPENDIX I. 23 Technologies Coverage 69 in IEEE Publications...... 134 CONTENTS 1. INTRODUCTION...... 5 2. SEAMLESS INTELLIGENCE be increased to upwards of 85 percent with col- SCENARIO...... 8 lective cloud computing automation of virtualized 3. 23 TECHNOLOGIES IN 2022...... 14 resources. 3.1 Security Cross-Cutting Issues...... 14 3.2 The Open Intellectual Public clouds are publicly available cloud com- Property Movement...... 17 puting services, and private clouds are datacenter 3.3 Sustainability...... 20 implementations that dedicate datacenter resourc- 3.4 Massively Online Open Courses...... 25 es to a private party. Public clouds are multitenant, 3.5 Quantum Computing ...... 28 whereas private clouds are intended for a single 3.6 Device and Nanotechnology...... 31 tenant. A secure subset of a public cloud can be 3.7 3D Integrated Circuits...... 33 used as a dedicated private cloud, such as a VPC CONTENTS3.8 Universal Memory...... 38 (virtual private cloud) in the AWS offering. 3.9 Multicore...... 43 working toward standards in this space, specifically 3.10 Photonics...... 47 Vendors today, including VMware, Cisco, and EMC, on P2301 cloud portability and commonality and 3.11 Networking and Interconnectivity... 52 offer modular building blocks to construct data- P2302 cloud-to-cloud interoperability. 3.12 Software-Defined Networks...... 55 centers that employ cloud computing technology 3.13 High-Performance Computing for building private and public clouds. Public cloud (HPC) ...... 63 Layered Approach to Cloud services are available worldwide through mega-da- 3.14 Cloud Computing...... 68 Computing Infrastructure 3.15 The Internet of Things...... 74 tacenters built by companies like Amazon, Google, 3.16 Natural User Interfaces...... 77 and Microsoft. Since access to cloud services is Cloud computing is about managing a datacenter’s 3.17 3D Printing ...... 81 made via a network, accelerated network services, virtualized resources collectively to present a uni- 3.18 Big Data and Analytics ...... 84 such as content delivery networks (CDNs) and fied system that can be allocated on demand and 3.19 Machine Learning and edge compute networks (ECNs) tend to facilitate managed automatically, to deal with availability and Intelligent Systems ...... 89 resilience and to engage the pay-as-you-go mod- 3.20 Computer Vision and Pattern and augment cloud computing services. Recognition...... 92 el. We envision and propose a community effort to develop a layered model for constructing cloud 3.21 Life Sciences...... 96 3.14.3 Challenges and Opportunities 3.22 Computational Biology and computing infrastructures that will allow the indus- Bioinformatics...... 102 Building today’s computing clouds is fairly com- try to cooperate on creating interoperable modules 3.23 Medical Robotics...... 108 plicated. Standards are scattered and lacking, that complement each other in a manner similar to 4. DRIVERS AND DISRUPTORS...... 112 and do not often follow a methodical approach. the ISO-OSI layered network model developed in 5. TECHNOLOGY COVERAGE IN An effort to develop a “divide-and-conquer” ap- the early 1980s to address the complexity of creat- IEEE XPLORE AND BY proach to standardizing layers of building block ing interoperable networking solutions. Standard- IEEE SOCIETIES...... 115 services can go a long way in accelerating the izing the building block layers of cloud computing 6. IEEE COMPUTER SOCIETY penetration of cloud computing into datacenter infrastructure has the opportunity to achieve the IN 2022...... 120 infrastructures. Today’s solutions are proprietary same results seen in networking over the past 7. SUMMARY AND NEXT STEPS.....124 and rarely interoperable. IEEE and the IEEE CS are three decades. 8. AUTHORS...... 126 APPENDIX I. 23 Technologies Coverage 70 in IEEE Publications...... 134 CONTENTS 1. INTRODUCTION...... 5 2. SEAMLESS INTELLIGENCE Virtual Connectivity for a Virtual Cloud sometimes adapting it to run is proprietary and can SCENARIO...... 8 Computing Substrate to a Workload take a lot of dedicated expertise. Nonetheless, the 3. 23 TECHNOLOGIES IN 2022...... 14 The purpose of clouds is to create and offer to resulting benefits in time to market, total cost of 3.1 Security Cross-Cutting Issues...... 14 developers a virtualized computing substrate on ownership, and increased flexibility, scalability, and 3.2 The Open Intellectual resilience outweigh the cost of the learning curve. Property Movement...... 17 top of which to run a workload. Unfortunately, 3.3 Sustainability...... 20 today’s networking and connectivity technology to The case for PaaS is challenging, the equivalent of 3.4 Massively Online Open Courses...... 25 integrate resources from different clouds as well introducing a new server platform for development. 3.5 Quantum Computing ...... 28 as resources from private enterprise datacenters The battle between Linux and Windows is a good 3.6 Device and Nanotechnology...... 31 falls short in enabling the flexible creation of the illustration of the barriers involved, because this is 3.7 3D Integrated Circuits...... 33 connectivity that could integrate such virtualized an effort only giants like Microsoft can tackle. Al- CONTENTS3.8 Universal Memory...... 38 distributed resources into an integrated substrate. though there are benefits to developing new scal- 3.9 Multicore...... 43 Furthermore, automated enforcement of secu- 3.10 Photonics...... 47 able applications that can take advantage of cloud 3.11 Networking and Interconnectivity... 52 rity measures is lacking, as is federated identity computing’s benefits, the reality is that it is very 3.12 Software-Defined Networks...... 55 management that could enable seamless interop- hard to develop to a new platform without taking 3.13 High-Performance Computing erability among security zones from multiple advantage of existing software. In the absence of (HPC) ...... 63 independent identity providers. Software-defined a secure and virtualized connectivity solution that 3.14 Cloud Computing...... 68 networking solutions such as Nicera’s are steps integrates the new development with existing soft- 3.15 The Internet of Things...... 74 in the right direction, but they do not address the 3.16 Natural User Interfaces...... 77 ware or enterprise data, such a platform is nice to full requirements in integrating resources across 3.17 3D Printing ...... 81 have but not very useful unless the enterprise ded- 3.18 Big Data and Analytics ...... 84 multiple clouds and enterprise datacenters. Proj- icates its entire development to the target platform 3.19 Machine Learning and ect Sydney at Microsoft attempted to achieve and accepts the risks of not being able to migrate Intelligent Systems ...... 89 this objective, but it failed to come to fruition for the application to other environments in the future. 3.20 Computer Vision and Pattern nontechnical reasons. Recognition...... 92 The case for SaaS is more compelling, both for the 3.21 Life Sciences...... 96 service operator and the consumer. The model of Developing for and Migrating to the Cloud 3.22 Computational Biology and finished services offered via a cloud to enterpris- Bioinformatics...... 102 Provisioning a cloud environment is itself a chal- es and users at large is strong, as service margins 3.23 Medical Robotics...... 108 lenge that requires resources with new IT skills that can be higher for the providing operators and the 4. DRIVERS AND DISRUPTORS...... 112 are also familiar with the requirements and capabil- benefits of paying as you go to the consumer—with 5. TECHNOLOGY COVERAGE IN ities of specific IaaS clouds such as AWS. The chal- all the other cloud computing benefits of scal- IEEE XPLORE AND BY lenge is not in developing applications, but rather IEEE SOCIETIES...... 115 ability, time to market, and not having to manage in provisioning the environment to host an applica- infrastructure—are obvious. There is an opportunity 6. IEEE COMPUTER SOCIETY tion along with the resources to support, operate, IN 2022...... 120 to offer middleware SaaS that requires developing and monitor it. For each public cloud, provisioning standard interoperability interfaces that combine 7. SUMMARY AND NEXT STEPS.....124 the environment, deploying the application, and 8. AUTHORS...... 126 APPENDIX I. 23 Technologies Coverage 71 in IEEE Publications...... 134 CONTENTS 1. INTRODUCTION...... 5 2. SEAMLESS INTELLIGENCE the same APIs between shrink-wrapped software computing technologies. Computing as a utility will SCENARIO...... 8 and software offered as a service. not likely become a reality by then. But short time 3. 23 TECHNOLOGIES IN 2022...... 14 to market in introducing application services will 3.1 Security Cross-Cutting Issues...... 14 The Potential of Mash-ups become the norm. The use of hosted application 3.2 The Open Intellectual services in the public cloud will also likely increase Property Movement...... 17 SaaS offers an interesting opportunity dynamically dramatically, as the economic incentives are too 3.3 Sustainability...... 20 compose new services from existing ones. Publish- 3.4 Massively Online Open Courses...... 25 ing such services’ APIs enables developers to build high to ignore, which will drive the market to seek 3.5 Quantum Computing ...... 28 more sophisticated software applications by uti- expert IT resources that can work with clouds. 3.6 Device and Nanotechnology...... 31 lizing the APIs for finished services through mash- 3.7 3D Integrated Circuits...... 33 ups. The resulting solutions can be developed and 3.14.5 Potential Disruptions CONTENTS3.8 Universal Memory...... 38 brought to market in a much shorter timeframe 3.9 Multicore...... 43 Cloud computing is a disruptive technology. It 3.10 Photonics...... 47 and at a much smaller cost. Dependency on SaaS changes the economic dynamics of how datacen- 3.11 Networking and Interconnectivity... 52 service continuity and how to assure continuity ters are built and operated, which impacts their 3.12 Software-Defined Networks...... 55 are a risk here. Standards may offer an answer. If total cost of ownership and drives server, storage, 3.13 High-Performance Computing standards are developed for such APIs, then the and network vendors to transition from the form (HPC) ...... 63 presence of multiple sources can alleviate this risk. factors and technologies they produce today to 3.14 Cloud Computing...... 68 technologies that are more suited to the cloud. The 3.15 The Internet of Things...... 74 3.16 Natural User Interfaces...... 77 Big Data and Analytics challenge/opportunity we presented above regard- 3.17 3D Printing ...... 81 The emergence of clouds enables enterprises to ing the layering of cloud infrastructure will drive 3.18 Big Data and Analytics ...... 84 utilize commodity computing to run applications new standards and present opportunities for new 3.19 Machine Learning and that in the past were prohibitively expensive. Tech- technology vendors to introduce new disruptive Intelligent Systems ...... 89 nologies like Hadoop and MapReduce let research- solutions to today’s existing vendor solutions. 3.20 Computer Vision and Pattern Recognition...... 92 ers tackle problems that were previously impossi- Disruption can also be felt on shrink-wrapped 3.21 Life Sciences...... 96 ble. But IT expertise in using such tools is still rare software products, as they will be challenged by 3.22 Computational Biology and and presents opportunities to introduce education- offerings that can be hosted in the cloud as a ser- Bioinformatics...... 102 al programs that can meet this need. vice. Traditional applications offered as cloud-host- 3.23 Medical Robotics...... 108 ed services have not yet matched in quality and 4. DRIVERS AND DISRUPTORS...... 112 3.14.4 Where We Think It Will Go versatility the traditional shrink-wrapped versions, 5. TECHNOLOGY COVERAGE IN such as Microsoft Office. Cloud-hosted services IEEE XPLORE AND BY We expect cloud computing to continue to im- IEEE SOCIETIES...... 115 prove, depending on how quickly the related also face the challenging requirement that network connectivity is robust and high in performance, 6. IEEE COMPUTER SOCIETY challenges and opportunities are addressed and IN 2022...... 120 resolved. By 2022, it is likely that most new in- which is still not the case everywhere around the world. There is an opportunity for such vendors to 7. SUMMARY AND NEXT STEPS.....124 stallations of datacenters will be based on cloud 8. AUTHORS...... 126 APPENDIX I. 23 Technologies Coverage 72 in IEEE Publications...... 134 CONTENTS 1. INTRODUCTION...... 5 2. SEAMLESS INTELLIGENCE introduce cloud-enhanced shrink-wrapped appli- 3.14.6 Summary SCENARIO...... 8 cations, where when connecting to the cloud, the By 2022, cloud computing will likely become more 3. 23 TECHNOLOGIES IN 2022...... 14 user gets significant feature enhancements that entrenched, and a significantly larger segment of 3.1 Security Cross-Cutting Issues...... 14 can leverage the cloud’s power without compro- computing workloads will be run on cloud com- 3.2 The Open Intellectual mising the power and sophistication of native appli- Property Movement...... 17 puting infrastructures, whether public or private. cations on PCs and tablets. 3.3 Sustainability...... 20 This promising market faces many challenges and 3.4 Massively Online Open Courses...... 25 opportunities. Standardization and inter-vendor 3.5 Quantum Computing ...... 28 Traditional Hosting cooperation on breaking up the puzzle of building 3.6 Device and Nanotechnology...... 31 Traditional hosting services face major disruption and managing cloud infrastructures is a major chal- 3.7 3D Integrated Circuits...... 33 if they do not transition to the cloud comput- lenge, and successes here can drive this market CONTENTS3.8 Universal Memory...... 38 ing model of operation, as they will not be able toward expansion much more rapidly. 3.9 Multicore...... 43 3.10 Photonics...... 47 to compete economically against the benefits The real promise of cloud computing is the way 3.11 Networking and Interconnectivity... 52 of total cost of ownership resulting from using that it changes the game for software develop- 3.12 Software-Defined Networks...... 55 cloud computing technology in the datacenter. ment. Once IT administrators and developers have 3.13 High-Performance Computing Furthermore, traditional datacenter architectures the ability to create true virtual datacenter infra- (HPC) ...... 63 are changing, and with the traditional server, structure substrates, where resources are connect- 3.14 Cloud Computing...... 68 storage and network form factors and solutions 3.15 The Internet of Things...... 74 ed virtually across clouds and premises, and devel- are no longer suitable for datacenters that are 3.16 Natural User Interfaces...... 77 opers are able to tap into APIs of services to mash 3.17 3D Printing ...... 81 built based on cloud computing technologies and up applications and middleware from different 3.18 Big Data and Analytics ...... 84 managed through the cloud. providers, there is the potential for experiencing a 3.19 Machine Learning and Cambrian-like explosion in the next generation of Intelligent Systems ...... 89 Time to Market 3.20 Computer Vision and Pattern software. The sophistication of newly developed Recognition...... 92 Cloud computing will invigorate the ability of en- offerings that leverage already developed SaaS 3.21 Life Sciences...... 96 trepreneurs to create and deliver to the market at can potentially exceed our wildest imaginations. 3.22 Computational Biology and a much lower cost and in a much shorter time the Bioinformatics...... 102 solutions that today’s vendors took years and huge 3.23 Medical Robotics...... 108 investment to develop. This new phenomenon will 4. DRIVERS AND DISRUPTORS...... 112 shake up the market and create a new dynamic for 5. TECHNOLOGY COVERAGE IN competition. IEEE XPLORE AND BY IEEE SOCIETIES...... 115 6. IEEE COMPUTER SOCIETY IN 2022...... 120 7. SUMMARY AND NEXT STEPS.....124 8. AUTHORS...... 126 APPENDIX I. 23 Technologies Coverage 73 in IEEE Publications...... 134 CONTENTS 1. INTRODUCTION...... 5 2. SEAMLESS INTELLIGENCE computing has been industrial applications, for SCENARIO...... 8 example, to track fleets of trucks on the road, 3. 23 TECHNOLOGIES IN 2022...... 14 perform detailed mapping of environments (recall 3.1 Security Cross-Cutting Issues...... 14 Google’s 2013 efforts to map sites like the Grand 3.2 The Open Intellectual Canyon for Google Earth), engage in environmental Property Movement...... 17 sensing and monitoring that utilizes either spe- 3.3 Sustainability...... 20 3.4 Massively Online Open Courses...... 25 cial devices or the many sensors now integrated 3.5 Quantum Computing ...... 28 into and/or available for smartphones or small 3.6 Device and Nanotechnology...... 31 3.15 The Internet of Things form-factor tablets, and locating and tracking CONTENTS3.7 3D Integrated Circuits...... 33 products in warehouses, transit, and stores. In 3.8 Universal Memory...... 38 3.15.1 State of the Art sports, there have been attempts to instrument 3.9 Multicore...... 43 balls and/or players to help improve goal-kicking Technology drivers for the Internet of Things (IoT) 3.10 Photonics...... 47 accuracy, and numerous studies of swimmers’ 3.11 Networking and Interconnectivity... 52 include sensor and actuator evolution and ubiqui- abilities use sensors built into sports devices or 3.12 Software-Defined Networks...... 55 ty, along with increased interconnectivity for such clothes. Recent research has explored ways to 3.13 High-Performance Computing devices with each other and with compute and (HPC) ...... 63 reduce re-stocking costs, using vision processing memory capacities. To understand the IoT, there- 3.14 Cloud Computing...... 68 by wandering robots. Related work in datacenter fore, we must begin by understanding devices and 3.15 The Internet of Things...... 74 systems attempts to track temperature profiles device connectivity. 3.16 Natural User Interfaces...... 77 for improved cooling efficiency. More recent work 3.17 3D Printing ...... 81 Electronic devices and sensors are becoming both aims to find less intrusive ways of monitoring or 3.18 Big Data and Analytics ...... 84 increasingly cheap and common, and device minia- reacting to inputs, such as commands from ges- 3.19 Machine Learning and Intelligent Systems ...... 89 turization is ongoing relentlessly. An early driver for ture recognition. Games like Microsoft’s Kinect 3.20 Computer Vision and Pattern these technologies has been the military’s need for and Nintendo’s Wii have seen substantial market Recognition...... 92 cheap, ubiquitous sensing. “Smart dust” technol- acceptance; underlining the upcoming importance 3.21 Life Sciences...... 96 ogy was funded by DARPA and the US military in of these technologies for broad sets of consumer 3.22 Computational Biology and the early 1990s, often based on MEMS devices, and Bioinformatics...... 102 electronics, Intel bought Omek Interactive, a ges- smart fabrics or e-textiles date back to a similar 3.23 Medical Robotics...... 108 ture recognition company, in 2013. period of time and sometimes called “wearable 4. DRIVERS AND DISRUPTORS...... 112 Finally, we have all heard about consumer appli- computers.” In fact, these technologies have be- cations such as smart homes, which can monitor 5. TECHNOLOGY COVERAGE IN come sufficiently mature to result in annual meet- IEEE XPLORE AND BY electricity consumption and adjust to current ings dedicated to discussing smart fabrics and their IEEE SOCIETIES...... 115 pricing, give owners remote access for monitoring roadmaps, for entirely commercial applications, 6. IEEE COMPUTER SOCIETY their properties, etc. These are already deployed in whether for fashion, sports, or medical purposes. IN 2022...... 120 European countries, along with smart grids, smart 7. SUMMARY AND NEXT STEPS.....124 A second driver for ubiquitous sensing and city facilities for security and monitoring, and many 8. AUTHORS...... 126 APPENDIX I. 23 Technologies Coverage 74 in IEEE Publications...... 134 CONTENTS 1. INTRODUCTION...... 5 2. SEAMLESS INTELLIGENCE other such facilities. In fact, self-monitored oil and someone or profilers using the data to update SCENARIO...... 8 gas pipelines predated many of these technologies, credit reports or inform potential loan agents. 3. 23 TECHNOLOGIES IN 2022...... 14 already deployed in the late 1980s to help watch for Yes, you can measure your gait and activity level, 3.1 Security Cross-Cutting Issues...... 14 pipeline failures. continuously, and such information can be invalu- 3.2 The Open Intellectual able to your doctor, but what if your health insur- Property Movement...... 17 ance rates rise because you did not move around 3.3 Sustainability...... 20 3.15.2 Where We Think It Will Go enough last month? In other words, concerns with 3.4 Massively Online Open Courses...... 25 But where are sensors/actuators and devices 3.5 Quantum Computing ...... 28 going? And what is the future of the IoT? Cer- privacy and security will affect the IoT’s growth 3.6 Device and Nanotechnology...... 31 tainly, by 2020, many heretofore manual business and acceptance, particularly in lieu of differences in 3.7 3D Integrated Circuits...... 33 processes will have been automated, whether via government actions and in the legal environments CONTENTS3.8 Universal Memory...... 38 of the US versus Europe versus Asia. 3.9 Multicore...... 43 active devices built into products and supply chains 3.10 Photonics...... 47 or via external sensors such as cameras. It will be Setting aside privacy concerns, amazing possibil- 3.11 Networking and Interconnectivity... 52 possible to dress in clothes that completely and ities arise not just for future wearable fabrics but 3.12 Software-Defined Networks...... 55 thoroughly monitor the wearer’s activities, which is also for other materials, such as weight-bearing 3.13 High-Performance Computing evidently useful in sports and sports training, in the supports for bridges monitoring their own status (HPC) ...... 63 arts (e.g., instrumented dancers whose moves are and reporting it, and smart airplane wings adjust- 3.14 Cloud Computing...... 68 amplified and displayed), and in medical settings. ing their surface structure to current air flow and 3.15 The Internet of Things...... 74 3.16 Natural User Interfaces...... 77 flight requirements. Further, the IoT can improve 3.17 3D Printing ...... 81 3.15.3 Challenges and Opportunities the lives of citizens and tourists alike, with cities 3.18 Big Data and Analytics ...... 84 What about our everyday actions, however, and guiding visitors to important sites, and the sites 3.19 Machine Learning and self-narrating their histories and importance; rather Intelligent Systems ...... 89 the privacy concerns raised by the IoT? Yes, it is convenient to walk into a coffee shop, have your than going by guidebooks and what everyone “has 3.20 Computer Vision and Pattern to” see, tourists can re-live what their friends found Recognition...... 92 wearable glasses recognize customers’ faces and 3.21 Life Sciences...... 96 tell you that the person sitting by the window is exciting or re-trace the paths of an ancestor who 3.22 Computational Biology and “Dave Smith,” someone you last met at some busi- used to live in that city. City dwellers can bypass Bioinformatics...... 102 ness meeting. This way, you won’t be embarrassed crowds or avoid traffic jams, guided on the current 3.23 Medical Robotics...... 108 by having forgotten his name and affiliation. How- best routes to their destinations, or, when using 4. DRIVERS AND DISRUPTORS...... 112 ever, do you really want the “cloud” to know where public transportation or walking, guided to meet 5. TECHNOLOGY COVERAGE IN you are right now, with whom you are talking, and friends along the way. In suburbia, car sharing may IEEE XPLORE AND BY be automated, with self-directed cars arranging IEEE SOCIETIES...... 115 who else is around? Numerous commercial opportunities are raised by such monitoring, making it of for meeting places, and for overland trips, tedious 6. IEEE COMPUTER SOCIETY long hours of driving along country roads replaced IN 2022...... 120 interest to many businesses, but such monitoring is also rife for abuse, as with stalkers hounding by a self-driving car narrating when beacons are 7. SUMMARY AND NEXT STEPS.....124 passed, without the need for constant hands-on 8. AUTHORS...... 126 APPENDIX I. 23 Technologies Coverage 75 in IEEE Publications...... 134 CONTENTS 1. INTRODUCTION...... 5 2. SEAMLESS INTELLIGENCE driving. Even better, city traffic jams SCENARIO...... 8 become more bearable (and more 3. 23 TECHNOLOGIES IN 2022...... 14 efficient) when cars move automat- 3.1 Security Cross-Cutting Issues...... 14 ically when the light turns green and 3.2 The Open Intellectual when cars move at some common, Property Movement...... 17 safe speed. 3.3 Sustainability...... 20 3.4 Massively Online Open Courses...... 25 In nature settings, hikers encounter- 3.5 Quantum Computing ...... 28 ing wildlife may result in, say, bears 3.6 Device and Nanotechnology...... 31 being tagged, virtually, to track their 3.7 3D Integrated Circuits...... 33 movements, via image and face CONTENTS3.8 Universal Memory...... 38 3.9 Multicore...... 43 recognition technologies. Long-term 3.10 Photonics...... 47 research on bear behavior and pres- 3.11 Networking and Interconnectivity... 52 ervation of their natural habitat can 3.12 Software-Defined Networks...... 55 use such data as well. Anglers may 3.13 High-Performance Computing be directed to where the fish are, not (HPC) ...... 63 only by their own sonar fish find- 3.14 Cloud Computing...... 68 ers, but because all such finders are Such data can be used to identify, e.g., industri- 3.15 The Internet of Things...... 74 linked, via the IoT, with back-end processing com- al, urban, or home pollution sources, not only in 3.16 Natural User Interfaces...... 77 extent but also in their effects on the environment, 3.17 3D Printing ...... 81 puting likely fish concentrations and possibly en- informing enforcement or lawmakers. It can also 3.18 Big Data and Analytics ...... 84 forcing fishing quotas. In fact, it is the global scope 3.19 Machine Learning and of the IoT—its knowledge about the many sensors help attract and maintain tourism. Intelligent Systems ...... 89 deployed in many different settings—that presents 3.20 Computer Vision and Pattern 3.15.4 Potential Disruptions Recognition...... 92 entirely new opportunities for enriching or facili- 3.21 Life Sciences...... 96 tating our daily lives, saving the planet, and helping It should be apparent from the descriptions above 3.22 Computational Biology and research. In one demo, a phone company, simply that the IoT is more than just the plethora of dis- Bioinformatics...... 102 by tracking its cellphones’ current locations, was tributed sensors and actuators embedded into 3.23 Medical Robotics...... 108 able to draw a precise map of London, including home, urban, or natural settings, but that it is 4. DRIVERS AND DISRUPTORS...... 112 boats in the river, demonstrating progress in the critically enabled by and dependent on the tremen- 5. TECHNOLOGY COVERAGE IN mapping technologies enabled by the IoT. Or, in dous data collections and compute capacities in IEEE XPLORE AND BY science, by understanding water temperature and the back-end machines and datacenters that use IEEE SOCIETIES...... 115 pollutants, along with fish concentrations, and by such data to understand our world and improve 6. IEEE COMPUTER SOCIETY tracking the types of fish being caught, we can it. Its growth and continued evolution, therefore, IN 2022...... 120 understand for, say, the entire state of Minnesota, depends on our ability to deploy the computational 7. SUMMARY AND NEXT STEPS.....124 how and where fish populations thrive or suffer. and storage capacities needed to support the IoT. 8. AUTHORS...... 126 APPENDIX I. 23 Technologies Coverage 76 in IEEE Publications...... 134 CONTENTS 1. INTRODUCTION...... 5 2. SEAMLESS INTELLIGENCE That, however, depends on the potential profits and SCENARIO...... 8 gains obtained from building and paying for such 3. 23 TECHNOLOGIES IN 2022...... 14 infrastructure. It appears today that these profits 3.1 Security Cross-Cutting Issues...... 14 and gains exist, but there are many potential dis- 3.2 The Open Intellectual ruptions. First, what entities will obtain those gains Property Movement...... 17 and will, therefore, be willing to make continued 3.3 Sustainability...... 20 3.4 Massively Online Open Courses...... 25 investments? Currently, large companies like Goo- 3.5 Quantum Computing ...... 28 gle, etc., are capitalizing on end user services, but 3.6 Device and Nanotechnology...... 31 are there other models? Certainly, yes, as evident 3.7 3D Integrated Circuits...... 33 from developing countries’ use of “microservices” 3.16 Natural User Interfaces CONTENTS3.8 Universal Memory...... 38 as in banking via cellphones versus using local and 3.9 Multicore...... 43 unreliable banks. But if such developments become 3.16.1 Introduction 3.10 Photonics...... 47 more ubiquitous, the will to continue infrastructure Since the beginning of the computing age, the 3.11 Networking and Interconnectivity... 52 investment and the large entities needed for doing public has dreamed of computers that could 3.12 Software-Defined Networks...... 55 interact in a natural way using speech, gestures, 3.13 High-Performance Computing so may be reduced. Second, as already evident (HPC) ...... 63 from past experiences with peer-to-peer technol- and intelligence, and interface with humans in 3.14 Cloud Computing...... 68 ogy, there are legal and governmental issues, as the same natural ways as we communicate with 3.15 The Internet of Things...... 74 well, which if not resolved, can substantially slow one another. But what was once in the realm of 3.16 Natural User Interfaces...... 77 progress. A notable current case in point in the US popular science fiction culture is rapidly becom- 3.17 3D Printing ...... 81 is the legal case against someone wearing Google ing a part of everyday life. Using techniques from 3.18 Big Data and Analytics ...... 84 Glass while driving a car receiving a traffic fine in touch and gesture to speech recognition lets users 3.19 Machine Learning and Intelligent Systems ...... 89 lieu of laws passed against texting while driving. increasingly interact with computing devices just 3.20 Computer Vision and Pattern as naturally as they interact with each other. Such Recognition...... 92 3.15.5 Summary natural ways to interact with new technologies in- 3.21 Life Sciences...... 96 tend to make it easier to operate them and speed In summary, the IoT is here to stay, driven, among 3.22 Computational Biology and up their adoption. Bioinformatics...... 102 others, by device technology advances, the op- 3.23 Medical Robotics...... 108 portunities created by the billions of smartphones Since the first appearance of the graphical user 4. DRIVERS AND DISRUPTORS...... 112 with their rich built-in sensors, Internet connectivity interface (GUI) almost 40 years ago, engineers 5. TECHNOLOGY COVERAGE IN to fixed facilities, increased mobile connectivity, have envisioned increasingly natural ways to inter- IEEE XPLORE AND BY the new functionalities it enables, and business act with the systems they design and develop. But IEEE SOCIETIES...... 115 reasons, such as the desire to reduce cost through what has become known as the natural user in- 6. IEEE COMPUTER SOCIETY automation, reduced loss/wastage, and shorter terface (NUI) was in reality until recently only little IN 2022...... 120 durations for supply chains. more than an enhanced GUI. The capture of a lim- 7. SUMMARY AND NEXT STEPS.....124 ited range of human interactions such as speech, 8. AUTHORS...... 126 APPENDIX I. 23 Technologies Coverage 77 in IEEE Publications...... 134 CONTENTS 1. INTRODUCTION...... 5 2. SEAMLESS INTELLIGENCE handwriting with pen, and simple touch was used laptops with multitouch screens are transforming SCENARIO...... 8 as an alternative way to click a button or hit a key how we interact with our surroundings. With recent 3. 23 TECHNOLOGIES IN 2022...... 14 on a keyboard. Speech recognition took the form developments in connectivity and cloud services, 3.1 Security Cross-Cutting Issues...... 14 of command and control, and pen-based input devices today have continuous access to unimag- 3.2 The Open Intellectual was essentially about character recognition. With inable computing resources and mind-boggling Property Movement...... 17 relatively poor dependability, these fragmented el- amounts of data. The developer ecosystems evolv- 3.3 Sustainability...... 20 3.4 Massively Online Open Courses...... 25 ements of a NUI did not experience wide adoption. ing around these devices are empowered to flight 3.5 Quantum Computing ...... 28 However, today’s NUI is in the midst of a big tran- their innovations with millions of real users while 3.6 Device and Nanotechnology...... 31 sition. New display technologies turn any surface monitoring detailed usage in real time. This feedback 3.7 3D Integrated Circuits...... 33 into an interactive screen. Megapixel cameras loop creates a cycle of increasingly more refined NUI CONTENTS3.8 Universal Memory...... 38 interactions. Smartphones and other mobile devices 3.9 Multicore...... 43 and microphones are embedded in every device, enabling seamless understanding of speech and have led to the creation of the largest workbench 3.10 Photonics...... 47 in human history, and it is churning out numerous 3.11 Networking and Interconnectivity... 52 gestures. The incumbent keyboard and mouse are applications that are exploring new ground in NUIs, 3.12 Software-Defined Networks...... 55 giving way to gesture, touch, and the spoken lan- whether that is real-time machine translation ser- 3.13 High-Performance Computing guage. The traditional desktop and laptop comput- (HPC) ...... 63 ing devices are being supplemented, if not directly vices that break down language barriers between 3.14 Cloud Computing...... 68 displaced, by an aggregate of powerful connected people or applications that let your eyes control your 3.15 The Internet of Things...... 74 phone without actually touching it. 3.16 Natural User Interfaces...... 77 devices giving a sense of ambient intelligence. [1] 3.17 3D Printing ...... 81 There is another dynamic driving NUI development 3.18 Big Data and Analytics ...... 84 3.16.2 State of the Art that is coming from a transformation of the limited 3.19 Machine Learning and interface of individual computing devices to the Intelligent Systems ...... 89 Microsoft’s Kinect [2] is revolutionizing more than interactive gaming. One of the fastest selling con- opportunities arising from turning our living rooms, 3.20 Computer Vision and Pattern workspaces, and vehicles into computing spaces. Recognition...... 92 sumer devices of all times is being adapted for a 3.21 Life Sciences...... 96 wide range of applications outside the living room. While many devices become increasingly smaller, 3.22 Computational Biology and At the heart of the Kinect experience is its ability to these computing spaces are inherently physically Bioinformatics...... 102 larger than the user. Whether it is a self-driving car, a video analyze and process images, gestures, and voice. 3.23 Medical Robotics...... 108 conferencing room, or immersive video gaming, touch, These inputs can be used to create a NUI that can 4. DRIVERS AND DISRUPTORS...... 112 gesture, and speech have been integrated to provide a change the way users interact with computers. more natural and efficient way of interacting with these 5. TECHNOLOGY COVERAGE IN Example applications range from American Sign complex, multisensory systems. IEEE XPLORE AND BY IEEE SOCIETIES...... 115 Language readers to shopping applications that analyze your body shape and select a pair of jeans 6. IEEE COMPUTER SOCIETY 3.16.3 Challenges IN 2022...... 120 that both fits and flatters. But for all the progress and developments, the vast 7. SUMMARY AND NEXT STEPS.....124 Smartphones, tablets, and a new generation of majority of our interactions with technology remain 8. AUTHORS...... 126 APPENDIX I. 23 Technologies Coverage 78 in IEEE Publications...... 134 CONTENTS 1. INTRODUCTION...... 5 2. SEAMLESS INTELLIGENCE little changed. The software productivity tools user in the most appropriate way. Can utilize SCENARIO...... 8 at the core of our working lives take only limited information about a user’s current and past 3. 23 TECHNOLOGIES IN 2022...... 14 advantage of a NUI. Most of our home appliances environments, such as proximity to other phys- 3.1 Security Cross-Cutting Issues...... 14 are still firmly rooted in the dark ages of VCR pro- ical devices and resources, geo-location, and 3.2 The Open Intellectual gramming. Modern technologies are often crudely movement, to assist with task completion. Property Movement...... 17 integrated and appear too often more unintelligent • Language and inference. Understand natural 3.3 Sustainability...... 20 3.4 Massively Online Open Courses...... 25 than we would expect in this day and age. language with ability to correctly infers users’ 3.5 Quantum Computing ...... 28 To successfully leverage NUI trends and oppor- intentions and goals, and engage in a dialog to 3.6 Device and Nanotechnology...... 31 tunities, applications must perform one particular resolve ambiguity and simplify collaboration. 3.7 3D Integrated Circuits...... 33 function exceptionally well: in the context of a • Augmented reality. Create the most natural CONTENTS3.8 Universal Memory...... 38 extension to the reality in which the user op- 3.9 Multicore...... 43 specific task, they must enable the user to interact with the application as if the user were interacting erates. The ability to capture the surrounding 3.10 Photonics...... 47 natural environment and create an augmented 3.11 Networking and Interconnectivity... 52 with a capable human assistant. Accomplishing this environment tailored to complete the task at 3.12 Software-Defined Networks...... 55 goal involves a complete rethinking of the interac- hand. 3.13 High-Performance Computing tion between human and computer. A truly natural (HPC) ...... 63 interface goes beyond the interface between the By designing computing systems around these 3.14 Cloud Computing...... 68 user and the system and focuses on what the inter- interaction paradigms, we redefine the relationship 3.15 The Internet of Things...... 74 3.16 Natural User Interfaces...... 77 face actually enables and the processes required between users and their computing devices. We 3.17 3D Printing ...... 81 for this to happen. Applications that employ a NUI will no longer force our users into unintuitive and 3.18 Big Data and Analytics ...... 84 have to address these challenging “intelligent” arcane interactions typically required by existing 3.19 Machine Learning and interaction paradigms to be successful: computers. We do not envision NUIs to be static; Intelligent Systems ...... 89 they will dynamically adapt to the user and fine- • Predictive, anticipatory, and adaptive. Use 3.20 Computer Vision and Pattern tune themselves as usage evolves. This NUI vision Recognition...... 92 past and current user actions to assist with task 3.21 Life Sciences...... 96 completion or perhaps even automation. Pre- has the potential of enabling applications to play a 3.22 Computational Biology and dict user behavior and act in synergy with the greater role in tasks such as driving, walking, and Bioinformatics...... 102 user. reading, where traditional interfaces would either 3.23 Medical Robotics...... 108 be a distraction or even intolerable. • Contextual awareness. Understand the user’s 4. DRIVERS AND DISRUPTORS...... 112 context similarly to what would be expected of With a NUI, we envision a world where no device 5. TECHNOLOGY COVERAGE IN a human assistant. Capture intent and emotion is an island, and ambient intelligence is achieved IEEE XPLORE AND BY IEEE SOCIETIES...... 115 to better aid with task completion. through natural interactions and integrated soft- • Multisensory input. Ability to capture multi- ware architecture. Devices will work together and 6. IEEE COMPUTER SOCIETY IN 2022...... 120 modal input including but not limited to speech, perhaps more importantly, they will both feed into 7. SUMMARY AND NEXT STEPS.....124 touch, and gestures; ability to respond to the and take advantage of cloud services to exhibit the 8. AUTHORS...... 126 APPENDIX I. 23 Technologies Coverage 79 in IEEE Publications...... 134 CONTENTS 1. INTRODUCTION...... 5 2. SEAMLESS INTELLIGENCE human-like intelligence required for a truly natural interfaces, and speech recognition and natural SCENARIO...... 8 interface. This mesh of devices and computing language processing. 3. 23 TECHNOLOGIES IN 2022...... 14 services will constantly collect a wide range of sen- The NUI is starting to make its way into the main- 3.1 Security Cross-Cutting Issues...... 14 sory information and combine this with contextual stream, but the work to make it real has been 3.2 The Open Intellectual data via cloud services. In turn, cloud services will Property Movement...... 17 going on for years. We should expect innovation to interpret aggregated datasets against patterns to 3.3 Sustainability...... 20 continue, with the emergence of entirely new kinds anticipate tasks, activities, and events, and in this 3.4 Massively Online Open Courses...... 25 of computing form factors combined with a wide 3.5 Quantum Computing ...... 28 way, adapt its behavior to the particular devices range of significant hardware and software tech- 3.6 Device and Nanotechnology...... 31 used in a given situation. nological breakthroughs leading to far more radical 3.7 3D Integrated Circuits...... 33 The promise of a brave NUI world is the emergence types of NUI. This an amazing opportunity for both CONTENTS3.8 Universal Memory...... 38 3.9 Multicore...... 43 of applications that will make possible a new model researchers in academia and for the technology in- 3.10 Photonics...... 47 of interaction between human and computer. Our dustry to create even more exciting products with 3.11 Networking and Interconnectivity... 52 “intelligent” interaction patterns and continued the NUI at their core. 3.12 Software-Defined Networks...... 55 new hardware innovations are a challenge, but they 3.13 High-Performance Computing will surely enable a multitude of computing devices 3.16.5 References (HPC) ...... 63 that know so much about us that they can increas- 3.14 Cloud Computing...... 68 [1] E.H.L. Aarts and J.L. Encarnação, True Visions: The ingly work on our behalf. 3.15 The Internet of Things...... 74 Emergence of Ambient Intelligence, Springer, 2006. 3.16 Natural User Interfaces...... 77 [2] G. Goth, “Brave NUI World,” Comm. ACM, vol. 54, no. 3.17 3D Printing ...... 81 3.16.4 Summary 12, 2011. 3.18 Big Data and Analytics ...... 84 After years of being the Next Big Thing on the tech- 3.19 Machine Learning and Intelligent Systems ...... 89 nology horizon, NUIs are rapidly becoming main- 3.20 Computer Vision and Pattern stream. Interactions between human and machine Recognition...... 92 become more natural and intuitive when people 3.21 Life Sciences...... 96 can use touch, gesture, and speech to interact with 3.22 Computational Biology and their computing devices. Bioinformatics...... 102 3.23 Medical Robotics...... 108 Hardware prices are falling rapidly and capabilities 4. DRIVERS AND DISRUPTORS...... 112 rising at an even faster pace. These developments 5. TECHNOLOGY COVERAGE IN are making it easier to embed sensors, extreme IEEE XPLORE AND BY processing power, and connectivity into devices IEEE SOCIETIES...... 115 and surroundings. The software that runs these 6. IEEE COMPUTER SOCIETY technologies is the result of years of research into IN 2022...... 120 computer vision, machine learning, big data, user 7. SUMMARY AND NEXT STEPS.....124 8. AUTHORS...... 126 APPENDIX I. 23 Technologies Coverage 80 in IEEE Publications...... 134 CONTENTS 1. INTRODUCTION...... 5 2. SEAMLESS INTELLIGENCE commercialized several decades ago. Carl Deck- SCENARIO...... 8 ard and Joseph Beaman invented a selective laser 3. 23 TECHNOLOGIES IN 2022...... 14 sintering printer at the University of Texas, Austin, 3.1 Security Cross-Cutting Issues...... 14 in 1986. That same year, Charles Hull received a 3.2 The Open Intellectual patent on stereolithography, a method for building Property Movement...... 17 up a solid object by depositing successive thin lay- 3.3 Sustainability...... 20 3.4 Massively Online Open Courses...... 25 ers of a liquid polymer that could be cured (solid- 3.5 Quantum Computing ...... 28 ified) by exposure to ultraviolet light. These basic 3.6 Device and Nanotechnology...... 31 techniques have been refined and form the basis of 3.7 3D Integrated Circuits...... 33 3.17 3D Printing many commercial-grade 3D printers. CONTENTS3.8 Universal Memory...... 38 3D printing is currently used in a wide variety of 3.9 Multicore...... 43 3.17.1 Introduction small-scale and custom fabrication jobs. For exam- 3.10 Photonics...... 47 3D printing promises a revolution in fabrication. In ple, in the movie The Hobbit: An Unexpected Jour- 3.11 Networking and Interconnectivity... 52 today’s manufacturing, products are usually assem- 3.12 Software-Defined Networks...... 55 ney, most of the animatronics for goblin eyeballs, bled from components that are separately created 3.13 High-Performance Computing facial muscles, lips, and tongues were 3D printed. using specialized machinery. The scale required to (HPC) ...... 63 Such props were traditionally made by hand, re- 3.14 Cloud Computing...... 68 make such specialized manufacturing cost effec- quiring weeks of work by a skilled artisan, whereas 3.15 The Internet of Things...... 74 tive often requires a network of far-flung suppliers they can now be prototyped, refined, and produced 3.16 Natural User Interfaces...... 77 and complex supply chains. With 3D printers capa- in days. Hobbyists and artists fabricate small items 3.17 3D Printing ...... 81 ble of handling multiple materials, it may become 3.18 Big Data and Analytics ...... 84 in personal printers or send them to 3D print- possible to fabricate many such items entirely in 3.19 Machine Learning and ing services for items requiring printers that can one place, close to the consumer. 3D printers can Intelligent Systems ...... 89 handle multiple materials or larger scale. Dental already create many shapes using combinations of 3.20 Computer Vision and Pattern labs produce custom dental crowns, bridges, and Recognition...... 92 materials that would be very difficult to create with orthodontic appliances in hours using digital oral 3.21 Life Sciences...... 96 conventional machining methods. Moreover, they scanning, specialized CAD/CAM software, and 3D 3.22 Computational Biology and can handle products from a few inches to many Bioinformatics...... 102 printers. Large manufacturers, such as aircraft and feet in size, and materials ranging from plastic to 3.23 Medical Robotics...... 108 automobile companies, use 3D printing for rapidly metal to edible foodstuffs to stem cells for creating 4. DRIVERS AND DISRUPTORS...... 112 prototyping parts and for producing specialized living tissue. The possibilities of what can be made 5. TECHNOLOGY COVERAGE IN production parts such as jigs for aircraft assembly. IEEE XPLORE AND BY with 3D printers are endless. IEEE SOCIETIES...... 115 Inexpensive 3D printers typically use plastic as the building material, but there are experimental 6. IEEE COMPUTER SOCIETY 3.17.2 State of the Art modifications to print with food pastes such as IN 2022...... 120 Also known as additive manufacturing, the ba- cookie dough and frosting to produce elaborate 7. SUMMARY AND NEXT STEPS.....124 sic 3D printing technology was invented and 8. AUTHORS...... 126 APPENDIX I. 23 Technologies Coverage 81 in IEEE Publications...... 134 CONTENTS 1. INTRODUCTION...... 5 2. SEAMLESS INTELLIGENCE confections. Industrial 3D printers can handle a produce through machining. Other techniques such SCENARIO...... 8 variety of materials, including ceramics and metals as surface modeling borrow from video game and 3. 23 TECHNOLOGIES IN 2022...... 14 such as bronze, steel, tungsten, and titanium. Tech- animation design, but these methods are primarily 3.1 Security Cross-Cutting Issues...... 14 nologies include deposition-based methods, where intended for modeling the surface of the solid, not 3.2 The Open Intellectual material is deposited in paper-thin layers to build the interior. Even surface modeling requires consid- Property Movement...... 17 up the desired shape and methods where lasers eration of surface properties such as color, texture, 3.3 Sustainability...... 20 3.4 Massively Online Open Courses...... 25 melt and fuse powdered metals or polymers. These reflectivity, and hardness. Solid forms have many 3.5 Quantum Computing ...... 28 printers can produce shapes and forms that are aspects that are difficult to specify and optimize 3.6 Device and Nanotechnology...... 31 difficult or impossible to create using conventional using current software, including strength require- 3.7 3D Integrated Circuits...... 33 techniques. For example, it is possible to create ments, rigidity, weight, and center of gravity. In ad- CONTENTS3.8 Universal Memory...... 38 a mesh consisting of seamless interlocking rings, dition, for industrial use, the design must consider 3.9 Multicore...... 43 which is difficult using conventional processes. other aspects such as cost, ease of manufacturing, 3.10 Photonics...... 47 Precisely shaped internal voids can be created by and workflow if multiple steps are required. Overall, 3.11 Networking and Interconnectivity... 52 filling the space with material that can later be re- there are significant challenges in designing soft- 3.12 Software-Defined Networks...... 55 3.13 High-Performance Computing moved, such as a gel or unfused powder. Multi-ma- ware that will make it easy for users to visualize, (HPC) ...... 63 terial objects can be formed by filling voids in a refine, and realize forms from their imagination into 3.14 Cloud Computing...... 68 strong structure with a different material later, an input for a 3D printer. 3.15 The Internet of Things...... 74 yielding the possibility of strong yet light composite Besides the technological issues, 3D printing brings 3.16 Natural User Interfaces...... 77 structures. 3.17 3D Printing ...... 81 with it several social, legal, and ecological challeng- 3.18 Big Data and Analytics ...... 84 es. Given appropriate design files, it is possible to 3.19 Machine Learning and 3.17.3 Challenges print weapons and contraband items such as coun- Intelligent Systems ...... 89 A major challenge in using 3D printing to its full terfeit goods and paraphernalia for manufacturing 3.20 Computer Vision and Pattern Recognition...... 92 potential is that the available software and algo- illegal drugs, making it difficult to regulate such 3.21 Life Sciences...... 96 rithms for driving the hardware are still nascent items. With improvements in methods for deriving 3.22 Computational Biology and and limited. Input to 3D printers is typically in the designs from 3D scanning, it will become easier Bioinformatics...... 102 form of an STL (Standard Tesselation Language) to reproduce proprietary designs, thus evading 3.23 Medical Robotics...... 108 file generated by a CAD package. While sever- intellectual property regulations. Parts made from 4. DRIVERS AND DISRUPTORS...... 112 al CAD packages are designed for 3D printing, uncertified designs, such as replacement parts for 5. TECHNOLOGY COVERAGE IN generating a correct, printable object description automobiles, may be dangerous in use. Taxation IEEE XPLORE AND BY from a concept in a user’s mind is a fairly complex of product sales will become harder if users can IEEE SOCIETIES...... 115 process. Solid modeling software often uses met- simply purchase designs that they can print them- 6. IEEE COMPUTER SOCIETY aphors adapted from machine shop operations, selves. Inexpensive and easily printed products IN 2022...... 120 such as drilling and milling; however, 3D printers made from non-recyclable and non-biodegradable 7. SUMMARY AND NEXT STEPS.....124 can generate many forms that are difficult to materials may lead to more pollution and other 8. AUTHORS...... 126 APPENDIX I. 23 Technologies Coverage 82 in IEEE Publications...... 134 CONTENTS 1. INTRODUCTION...... 5 2. SEAMLESS INTELLIGENCE ecological issues. New legal regulations may be purchase or otherwise obtain designs for items SCENARIO...... 8 required to address these issues. over the Internet and print them either in their 3. 23 TECHNOLOGIES IN 2022...... 14 homes or in local print shops, then there may be 3.1 Security Cross-Cutting Issues...... 14 3.17.4 Where We Think It Will Go less need for retail personnel. On the other hand, it 3.2 The Open Intellectual may also create design jobs and a need for teach- Property Movement...... 17 As 3D printer hardware and software improve and ers and equipment to help train designers. 3.3 Sustainability...... 20 become cheaper and more widely available, we 3.4 Massively Online Open Courses...... 25 expect that more products will be customized to 3.5 Quantum Computing ...... 28 specific use cases. Custom prosthetics and even 3.17.6 Summary 3.6 Device and Nanotechnology...... 31 replacements for body parts may be created with 3D printing, also known as additive manufactur- 3.7 3D Integrated Circuits...... 33 3D printers—initial research already shows that ing, is a technology with enormous potential that CONTENTS3.8 Universal Memory...... 38 3.9 Multicore...... 43 cartilage can be formed using 3D printed molds. will allow us to produce objects with designs that 3.10 Photonics...... 47 Combining different materials will allow the cre- in the past would have been prohibitively expen- 3.11 Networking and Interconnectivity... 52 ation of composite materials with new properties, sive or impossible to manufacture. As the printing 3.12 Software-Defined Networks...... 55 such as the ability to heal after failure. The ability hardware and design software improve, we expect 3.13 High-Performance Computing to print batteries and sensors directly onto objects that a wide variety of products will be manufac- (HPC) ...... 63 will enable more “smart” mechanisms that can tured mostly or even entirely using 3D printers in a 3.14 Cloud Computing...... 68 sense changes in surrounding temperature and manufacturing plant, at local printing services, or in 3.15 The Internet of Things...... 74 3.16 Natural User Interfaces...... 77 light levels, as well as impending failure. By adding the consumer’s home. These changes may be quite 3.17 3D Printing ...... 81 articulated joints and electrical connections, it will disruptive because the increased automation may 3.18 Big Data and Analytics ...... 84 become possible to print complete, fully function- reduce jobs in manufacturing, assembly, freight 3.19 Machine Learning and ing devices, both electronic and mechanical, rather transportation, and retailing. Changes will be Intelligent Systems ...... 89 than assembling them from parts produced sep- required in education to train a new generation of 3.20 Computer Vision and Pattern Recognition...... 92 arately. Manufacture of many products will move designers, as well as in laws to manage new issues 3.21 Life Sciences...... 96 from large, centralized factories to local workshops in intellectual property, taxation, and certification 3.22 Computational Biology and and even the user’s home. of product safety and effectiveness. Bioinformatics...... 102 3.23 Medical Robotics...... 108 3.17.5 Disruptions 3.17.7 References 4. DRIVERS AND DISRUPTORS...... 112 3D printing may be highly disruptive because it 1. H. Lipson and M. Kurman, Fabricated: The New 5. TECHNOLOGY COVERAGE IN could make many jobs obsolete. For example, if World of 3D Printing, Wiley, 2013. IEEE XPLORE AND BY IEEE SOCIETIES...... 115 entire mechanisms can be created directly by 3D 2. S. Bradshaw, A. Bowyer, and P. Haufe, “The In- printing, then this may eliminate many assembly 6. IEEE COMPUTER SOCIETY tellectual Property Implications of Low-Cost 3D IN 2022...... 120 jobs. With local manufacturing of goods, there Printing,” SCRIPTed, vol. 7, no. 1, Apr. 2010. may be less freight to transport. If consumers can 7. SUMMARY AND NEXT STEPS.....124 3. Saracco, R., Personal Communication, 2014. 8. AUTHORS...... 126 APPENDIX I. 23 Technologies Coverage 83 in IEEE Publications...... 134 CONTENTS 1. INTRODUCTION...... 5 2. SEAMLESS INTELLIGENCE this data with long-term information such as pre- SCENARIO...... 8 vious user reviews of the different dishes in this 3. 23 TECHNOLOGIES IN 2022...... 14 restaurant, your weight loss goals, food preference, 3.1 Security Cross-Cutting Issues...... 14 and sensitiveness, and perhaps even your individ- 3.2 The Open Intellectual ual genetic properties. In milliseconds, it makes its Property Movement...... 17 top three suggestions, from which you choose your 3.3 Sustainability...... 20 3.4 Massively Online Open Courses...... 25 meal. In the meantime, data is collected anony- 3.5 Quantum Computing ...... 28 mously in the aggregate about your choices and 3.6 Device and Nanotechnology...... 31 decisions, as well as other patrons’. The restaurant 3.7 3D Integrated Circuits...... 33 manager can use it to adjust the menu. Research- CONTENTS3.8 Universal Memory...... 38 3.18 Big Data and Analytics ers can use it to better understand the relationship 3.9 Multicore...... 43 between nutrition, fitness, and health. Your friends 3.10 Photonics...... 47 3.18.1 Executive Summary can use it to obtain personalized food recommen- 3.11 Networking and Interconnectivity... 52 More data is collected, shared, and analyzed every dations, and you can use it to track your progress 3.12 Software-Defined Networks...... 55 3.13 High-Performance Computing day. The growing availability of data and demand toward your health goals. (HPC) ...... 63 for its insights hold great potential to improve many This scenario, already more science than fiction, 3.14 Cloud Computing...... 68 data-driven decisions, from the mundane to the exemplifies how so-called “big data” can be used 3.15 The Internet of Things...... 74 strategic. But this growth also poses significant 3.16 Natural User Interfaces...... 77 to seamlessly affect decisions from the prosaic challenges, both technological and societal. To (your choice of lunch) to the strategic (the FDA’s 3.17 3D Printing ...... 81 harness the deluge of data to beneficial use, we will 3.18 Big Data and Analytics ...... 84 nutrition guidelines). It represents but one of many need to address rapid changes in data acquisition, 3.19 Machine Learning and opportunities envisioned for the large-scale analyt- Intelligent Systems ...... 89 storage, and processing technologies; education, ics of diverse data. Big data is finally transitioning 3.20 Computer Vision and Pattern both of the analytics workforce and everyday us- from the computer science and machine learning Recognition...... 92 ers; and complex privacy issues. 3.21 Life Sciences...... 96 classrooms into numerous real-world scenarios in 3.22 Computational Biology and business, government and military, science, politics, Bioinformatics...... 102 3.18.2 Introduction medicine, climate, and personal analytics—a trend 3.23 Medical Robotics...... 108 Imagine, if you will, the following scenario. You sit that we expect to grow rapidly through 2022. 4. DRIVERS AND DISRUPTORS...... 112 down at a restaurant for lunch, wondering what 5. TECHNOLOGY COVERAGE IN to order. You take a picture of the menu with your 3.18.3 State of the Art IEEE XPLORE AND BY phone, starting a whirlwind of activity on your IEEE SOCIETIES...... 115 Big data is exploding, with no signs of slowing behalf. The software combs through the data it has down. The growth is manifest on two separate 6. IEEE COMPUTER SOCIETY collected during the day about you, such as your IN 2022...... 120 axes: more data is collected, and more data is breakfast, exercise and calorie expenditure, blood shared. Growth along both axes is exponential, and 7. SUMMARY AND NEXT STEPS.....124 pressure and blood sugar levels, etc. It combines 8. AUTHORS...... 126 APPENDIX I. 23 Technologies Coverage 84 in IEEE Publications...... 134 CONTENTS 1. INTRODUCTION...... 5 2. SEAMLESS INTELLIGENCE the combined growth results in a very rapid in- 3.18.4 Where We Think It Will Go SCENARIO...... 8 crease in total available data indeed. IDC estimates We see tremendous opportunities in big data. In 3. 23 TECHNOLOGIES IN 2022...... 14 that the amount of data created and shared on the the sciences, for example, the growth in experi- 2 3.1 Security Cross-Cutting Issues...... 14 Internet will reach around 8 zettabytes by 2015. mental data and in simulations—the fourth para- 3.2 The Open Intellectual Let’s look at a few examples: Property Movement...... 17 digm of science—has already advanced our under- 3.3 Sustainability...... 20 • Photos and videos taken and shared are standing of the universe and of life. The growth in 3.4 Massively Online Open Courses...... 25 growing at an exponential rate,3 a result of ubiquity of mobile computing devices, as well as in 3.5 Quantum Computing ...... 28 three multiplicative trends. One, people are the applications that collect data, means that much 3.6 Device and Nanotechnology...... 31 taking more photos, because more camer- more data is available about a lot more people (and CONTENTS3.7 3D Integrated Circuits...... 33 as are ubiquitously available through the possibly to many more people). This data is often 3.8 Universal Memory...... 38 used in quotidian decisions such as picking a driv- 3.9 Multicore...... 43 proliferation of smart and feature phones. ing route. In business, much more data is collected 3.10 Photonics...... 47 Two, these photos increasingly contain more 3.11 Networking and Interconnectivity... 52 data (pixels) through the rapid growth in on every aspect of operation, increasing efficiency, 4 3.12 Software-Defined Networks...... 55 sensor technology. And three, more people customer marketing, and pivoting to new markets. 3.13 High-Performance Computing share their photos and videos on Facebook, In medicine, big data combined with rapid advanc- (HPC) ...... 63 YouTube, Twitter, Snapchat, and other fast- es medical science can bring us to a point where all 3.14 Cloud Computing...... 68 growth social networks. major health decisions are tailored to an individu- 3.15 The Internet of Things...... 74 al’s situation.5 3.16 Natural User Interfaces...... 77 • Crowd-sourced and individuals’ data from day- 3.17 3D Printing ...... 81 to-day life is proliferating through a variety of If big data fulfills its promise, we think it will have 3.18 Big Data and Analytics ...... 84 mobile applications. Such information includes tremendous impact in reducing uncertainty around 3.19 Machine Learning and service reviews, traffic and geo-tagged check- large domains of decisions, both before they’re Intelligent Systems ...... 89 ins, health and exercise metrics from wearable 3.20 Computer Vision and Pattern made and, retrospectively, afterward, too. Recognition...... 92 devices, etc. (see Section 3.2). 3.21 Life Sciences...... 96 • More studies, more instrumentation, more 3.18.5 Technological Challenges 3.22 Computational Biology and simulations, and more sharing facilitated by the Bioinformatics...... 102 Internet (e.g., CERN’s Grid) is translating into The collection, organization, validation, interpre- 3.23 Medical Robotics...... 108 more scientific data. tation, and management of large datasets present 4. DRIVERS AND DISRUPTORS...... 112 • The increase in resolution of freely available multiple technical and technological challenges. As the amount of data grows rapidly, additional 5. TECHNOLOGY COVERAGE IN elevation data has led to a lot more mapping. IEEE XPLORE AND BY IEEE SOCIETIES...... 115 4 http://www.wired.com/insights/2013/07/putting-a-dollar-value-on 6. IEEE COMPUTER SOCIETY 2 IDC report “Extracting Value from Chaos,” June 2011 -big-data-insights IN 2022...... 120 3 See Kleiner Perkin’s “Internet Trends 2013” report at 5 http://spectrum.ieee.org/tech-talk/computing/software/predictive 7. SUMMARY AND NEXT STEPS.....124 http://www.slideshare.net/kleinerperkins/kpcb-internet-trends-2013 -analytics-and-deciding-who-should-receive-organ-transplants 8. AUTHORS...... 126 APPENDIX I. 23 Technologies Coverage 85 in IEEE Publications...... 134 CONTENTS 1. INTRODUCTION...... 5 2. SEAMLESS INTELLIGENCE computational resources are required to process SCENARIO...... 8 the data in a timely manner. This time and resource 3. 23 TECHNOLOGIES IN 2022...... 14 pressure is increased for ongoing analysis by a 3.1 Security Cross-Cutting Issues...... 14 recurring deadline (such as daily business metrics) 3.2 The Open Intellectual and even more so for interactive and exploratory Property Movement...... 17 data exploration. 3.3 Sustainability...... 20 3.4 Massively Online Open Courses...... 25 Accordingly, computational resources dedicated to 3.5 Quantum Computing ...... 28 big data are growing explosively. The demand for 3.6 Device and Nanotechnology...... 31 data storage and processing can, however, grow 3.7 3D Integrated Circuits...... 33 faster than the underlying technologies. Take the CONTENTS3.8 Universal Memory...... 38 3.9 Multicore...... 43 recent growth of informatics-based scientific disci- 3.10 Photonics...... 47 plines, for example. In genomics, advances in gene For example, CERN is planning for its Wigner 3.11 Networking and Interconnectivity... 52 sequencer technology have brought down the cost datacenter to double its processing and storage 3.12 Software-Defined Networks...... 55 and delay of sequencing to the point where many capability in the next three years. Although impres- 3.13 High-Performance Computing labs around the world can produce copious genet- sive, this rate is a far cry from the growth rate of (HPC) ...... 63 ic data. Worldwide, we can now produce around 3.14 Cloud Computing...... 68 the data to be processed, creating an increasing 15 Pbytes of compressed genetic data per year, 3.15 The Internet of Things...... 74 gap between the amount of data to process and which is growing at a rate of 3x to 5x a year.6 In 3.16 Natural User Interfaces...... 77 the hardware to process it. If cost weren’t an issue 3.17 3D Printing ...... 81 high-energy physics, the Large Hadron Collider and in scaling the hardware, power still remains a stub- 3.18 Big Data and Analytics ...... 84 other instruments at CERN alone produce a similar born constraint, limiting practical datacenter size 3.19 Machine Learning and amount of data annually.7 Intelligent Systems ...... 89 to several MW. And even if that constraint were 3.20 Computer Vision and Pattern Pervasive big data tools such as Hadoop are to be removed by advances in power efficiency, Recognition...... 92 already prevalent in the analysis of these mas- the speed of light effectively limits the scale of a 3.21 Life Sciences...... 96 sive genomic databases. But despite the carefully datacenter to the tolerable limits of latency in data 3.22 Computational Biology and designed scalability of the software tools, they are fetching before computation grinds to a halt. Bioinformatics...... 102 still limited by hardware constraints, such as power, 3.23 Medical Robotics...... 108 acquisition, and operation costs; capacity growth 4. DRIVERS AND DISRUPTORS...... 112 3.18.6 Potential Technological Disruptions in processors, hard disks, and networks; and in- To fully exploit the opportunity promised by big 5. TECHNOLOGY COVERAGE IN creased complexity in management and cooling. IEEE XPLORE AND BY data, we must find ways to bridge the gap between IEEE SOCIETIES...... 115 data growth and processing capability. 6. IEEE COMPUTER SOCIETY 6 IEEE Spectrum 07-2013 “The DNA Data Deluge” On the hardware side, it is a simple matter to IN 2022...... 120 http://spectrum.ieee.org/biomedical/devices/the-dna-data-deluge extrapolate current growth trends to predict 7. SUMMARY AND NEXT STEPS.....124 7 http://home.web.cern.ch/about/computing increased storage density; continued processor 8. AUTHORS...... 126 APPENDIX I. 23 Technologies Coverage 86 in IEEE Publications...... 134 CONTENTS 1. INTRODUCTION...... 5 2. SEAMLESS INTELLIGENCE growth along Moore’s law; and better power human skills. These skills range from the selection, SCENARIO...... 8 efficiency and networks. But the trends set in the preparation, and cleaning of data; the exploration 3. 23 TECHNOLOGIES IN 2022...... 14 past two decades, even if exponential in growth, of different analyses on the data; the application 3.1 Security Cross-Cutting Issues...... 14 are not disruptive enough to close this gap. It of error checking and strong statistical reasoning 3.2 The Open Intellectual would take radical technological shifts to match to reduce bias and type I/II errors; and finally, the Property Movement...... 17 resource growth with data growth, or we could interpretation, visualization (for the higher-band- 3.3 Sustainability...... 20 3.4 Massively Online Open Courses...... 25 experience a significant decline in the current rate width visual sensory), and domain application of 3.5 Quantum Computing ...... 28 of data growth and, with it, the predictive capabili- the results. Although we believe that many import- 3.6 Device and Nanotechnology...... 31 ties of its analysis. ant parts of these processes can and will undergo 3.7 3D Integrated Circuits...... 33 Barring an unpredictable disruptive technology, a increased automation, we do not foresee an elimi- CONTENTS3.8 Universal Memory...... 38 nation of the skilled human element. If anything, big 3.9 Multicore...... 43 more feasible path to closing the gap is innovation in software. Big data software can be considered data analytics falls in line with the workforce migra- 3.10 Photonics...... 47 tion we observed in the past century from labor- to 3.11 Networking and Interconnectivity... 52 still in its infancy, with plenty of opportunities for knowledge-intensive industries. 3.12 Software-Defined Networks...... 55 growth. Areas of possible improvement include 3.13 High-Performance Computing • reducing the amount data to be processed: bet- In a recent study, 88 percent of companies sur- (HPC) ...... 63 veyed have already reported a talent shortage 3.14 Cloud Computing...... 68 ter compression; early detection of irrelevant to successfully execute on big data initiatives.8 3.15 The Internet of Things...... 74 data; and more effective sampling techniques; Because big data is growing at such a rapid rate, 3.16 Natural User Interfaces...... 77 • algorithmic effort/processing reduction: more 3.17 3D Printing ...... 81 efficient machine learning algorithms to pro- along with the demand for data scientists and an- 3.18 Big Data and Analytics ...... 84 duce predictions in less time, etc.; alysts, and because the skills required encompass 3.19 Machine Learning and a wide range of advanced computing, statistics, Intelligent Systems ...... 89 • improving systems effort: better utilization and sharing of available hardware resources; and communication, and domain expertise, we may po- 3.20 Computer Vision and Pattern tentially face a critical shortfall in this workforce.9 Recognition...... 92 • generating insightful analysis: something that 3.21 Life Sciences...... 96 produces much higher-level analytics and an- This challenge needs to be met by a correspond- 3.22 Computational Biology and swers than is standard today. ingly large challenge in workforce education, both Bioinformatics...... 102 in academia and industry. 3.23 Medical Robotics...... 108 Such innovations will not only reduce the require- Other aspects of the big data shift will require 4. DRIVERS AND DISRUPTORS...... 112 ments of labor and expertise from analysts but may in fact drive efficiency through more parsimonious societal response. Perhaps the biggest one is the 5. TECHNOLOGY COVERAGE IN concern about eroding privacy and data leaks, with IEEE XPLORE AND BY representations of data. IEEE SOCIETIES...... 115 8 http://thehiringsite.careerbuilder.com/2013/07/16/ 6. IEEE COMPUTER SOCIETY 3.18.7 Societal Challenges careerbuilder-big-data-study/ IN 2022...... 120 Big data hardware and software are no panacea. In 9 http://spectrum.ieee.org/podcast/at-work/tech-careers/ 7. SUMMARY AND NEXT STEPS.....124 fact, they are currently useless without specialized is-data-science-your-next-career 8. AUTHORS...... 126 APPENDIX I. 23 Technologies Coverage 87 in IEEE Publications...... 134 CONTENTS 1. INTRODUCTION...... 5 2. SEAMLESS INTELLIGENCE a potential for very significant personal, business, them. We may therefore experience a disruption SCENARIO...... 8 or military damage. The concentration of big data from the traditional employment model of accred- 3. 23 TECHNOLOGIES IN 2022...... 14 in the hands of governments also evokes concerns ited employees sharing an office space. Instead, 3.1 Security Cross-Cutting Issues...... 14 about the risk to democracy and civil rights. The we may see the increasing demand for these 3.2 The Open Intellectual challenge is then to find ways to collect, share, and professionals met by companies who successfully Property Movement...... 17 benefit from big data technologies while still pre- adapt to a distributed workforce of varying formal 3.3 Sustainability...... 20 3.4 Massively Online Open Courses...... 25 serving the privacy, trust, and rights of the individu- education. 3.5 Quantum Computing ...... 28 als whose data is collected. Finally, we may find that the explosive success of 3.6 Device and Nanotechnology...... 31 big data may hinge on a significant disruption in the 3.7 3D Integrated Circuits...... 33 3.18.8 Potential Disruptions field of data security and privacy. There is certainly CONTENTS3.8 Universal Memory...... 38 3.9 Multicore...... 43 Academia is already mobilizing to develop new a technological challenge and opportunity here, to 3.10 Photonics...... 47 programs around big data and to train thousands of come up with provable standards of privacy and 10 3.11 Networking and Interconnectivity... 52 new data scientists and analysts. Perhaps a more security. But there is also one that may require 3.12 Software-Defined Networks...... 55 radical approach in workforce education is required legal, normative, and educational changes to place 3.13 High-Performance Computing to meet the rapid demand. Interestingly, one po- acceptable limits on the use of big data. (HPC) ...... 63 tential disruption in the training of the workforce in 3.14 Cloud Computing...... 68 general, and big data in particular, also comes from 3.15 The Internet of Things...... 74 3.18.9 References 3.16 Natural User Interfaces...... 77 a new scale-out field: MOOCs (refer to Section G. Brumfiel, “High-Energy Physics: Down the Petabyte 3.17 3D Printing ...... 81 3.4). Distributed online education, with its various Highway,” Nature, vol. 469, 19 Jan. 2011, pp. 282–83. 3.18 Big Data and Analytics ...... 84 levels of certification and cost, is already training P. Webster, “Supercomputing the Climate: NASA’s Big 3.19 Machine Learning and many thousands of individuals in big data-related Data Mission,” CSC World, Computer Sciences Corpo- Intelligent Systems ...... 89 fields and showing a strong growth trend.11 3.20 Computer Vision and Pattern ration, 2012. Recognition...... 92 One of the characteristics of MOOCs is that suc- S. Shah, A. Horne, and J. Capella, “Good Data Won’t 3.21 Life Sciences...... 96 cessful training no longer requires physical school Guarantee Good Decisions,” Harvard Business Review, 3.22 Computational Biology and attendance and is therefore independent of ge- Bioinformatics...... 102 Apr. 2012. ography. This is just one aspect that may require 3.23 Medical Robotics...... 108 “Data, Data Everywhere”, The Economist, 25 Feb. 2010. employers too to radically adjust to the changing 4. DRIVERS AND DISRUPTORS...... 112 landscape of big data professionals, even if uni- O.J. Reichman, M.B. Jones, and M.P. Schildhauer, “Chal- 5. TECHNOLOGY COVERAGE IN versities and MOOC train an adequate number of lenges and Opportunities of Open Data in Ecology,” Sci- IEEE XPLORE AND BY ence, vol. 331, no. 6018, 2011, pp. 703–705; doi:10.1126/ IEEE SOCIETIES...... 115 10 http://www.wired.com/insights/2013/07/the-growing-need-for-big science.1197962. 6. IEEE COMPUTER SOCIETY -data-workers-meeting-the-challenge-with-training/ IN 2022...... 120 S. Cherry, “Is Data Science Your Next Career?”, IEEE 11 http://gigaom.com/2012/10/14/why-becoming-a-data-scientist Spectrum, 28 May 2013. 7. SUMMARY AND NEXT STEPS.....124 -might-be-easier-than-you-think/ 8. AUTHORS...... 126 APPENDIX I. 23 Technologies Coverage 88 in IEEE Publications...... 134 CONTENTS 1. INTRODUCTION...... 5 2. SEAMLESS INTELLIGENCE The simplified ML process consists of (i) training, SCENARIO...... 8 (ii) test, and (iii) prediction. A dataset is partitioned 3. 23 TECHNOLOGIES IN 2022...... 14 into the training set and the test set, where the 3.1 Security Cross-Cutting Issues...... 14 training set trains the system, and the test set is 3.2 The Open Intellectual kept secret from it. The output of the training pro- Property Movement...... 17 cess is a learned model that will be used for predic- 3.3 Sustainability...... 20 3.4 Massively Online Open Courses...... 25 tion; we use the test set to test the learned model’s 3.5 Quantum Computing ...... 28 accuracy. Because of a wide range of choice of 3.6 Device and Nanotechnology...... 31 training parameters as well as training algorithms, 3.7 3D Integrated Circuits...... 33 the ML process often becomes an iterative one, CONTENTS3.8 Universal Memory...... 38 with the final model selection resulting from an 3.9 Multicore...... 43 empirical process. 3.10 Photonics...... 47 3.19 Machine Learning and 3.11 Networking and Interconnectivity... 52 Intelligent Systems There are different ways of training in ML: 3.12 Software-Defined Networks...... 55 • Supervised learning. In this kind of learning, the 3.13 High-Performance Computing (HPC) ...... 63 3.19.1 Introduction algorithm is given training data that consist of 3.14 Cloud Computing...... 68 In the past decade, we have witnessed a dramatic examples with both the input data and the de- 3.15 The Internet of Things...... 74 increase in the use of machine learning (ML). The sired output, also known as labels. The learner 3.16 Natural User Interfaces...... 77 application of ML plays an increasingly important should be able to generalize from the presented 3.17 3D Printing ...... 81 role in our daily lives, whether it is ranking search previously unseen data. 3.18 Big Data and Analytics ...... 84 results for Google and Bing or recommending • Unsupervised learning. Here, the algorithm is 3.19 Machine Learning and Intelligent Systems ...... 89 products on Amazon.com and movies on Netflix. presented with examples from the input data 3.20 Computer Vision and Pattern Innovative ML techniques have led to emerging only and will fit a model to these observations Recognition...... 92 businesses that are able to identify influential without prior human knowledge. 3.21 Life Sciences...... 96 users on Twitter and capture product sentiments • Reinforcement learning. This algorithm learns 3.22 Computational Biology and how to respond given an observation of the Bioinformatics...... 102 for marketers. environment. Every action has an impact on the 3.23 Medical Robotics...... 108 ML is the discipline of artificial intelligence aimed environment, which provides feedback in the 4. DRIVERS AND DISRUPTORS...... 112 at creating computing systems that can learn from form of positive or negative rewards. 5. TECHNOLOGY COVERAGE IN data. For example, a classical ML system can be IEEE XPLORE AND BY trained on email messages to learn to distinguish 3.19.2 State of the Art IEEE SOCIETIES...... 115 between spam and ham. Optical character recog- With the dramatic increase in processing power and 6. IEEE COMPUTER SOCIETY nition is another example in which printed charac- storage capacity, the field of ML has changed dramat- IN 2022...... 120 ters are recognized automatically based on previ- ically in recent years. This change can be attributed to 7. SUMMARY AND NEXT STEPS.....124 ous examples. 8. AUTHORS...... 126 APPENDIX I. 23 Technologies Coverage 89 in IEEE Publications...... 134 CONTENTS 1. INTRODUCTION...... 5 2. SEAMLESS INTELLIGENCE a deeper understanding of ML algorithms, inventions has led to extraordinary improvements in speech SCENARIO...... 8 such as multicore processors, distributed computing, recognition, as popularized by Apple’s Siri service. 3. 23 TECHNOLOGIES IN 2022...... 14 and new storage technology, as well as to an explo- 3.1 Security Cross-Cutting Issues...... 14 sion of available data from an increasingly connected 3.19.3 Challenges 3.2 The Open Intellectual world—the so-called big data phenomenon. The ag- Property Movement...... 17 Most essential algorithms around us operate in gregate of these changes has resulted in copious new 3.3 Sustainability...... 20 near-linear or better time. We often think of these developments in ML that are fundamentally charac- 3.4 Massively Online Open Courses...... 25 algorithms as meeting the high bar of unlimit- 3.5 Quantum Computing ...... 28 terized by large scale. ed scalability. Unfortunately, the ML field often 3.6 Device and Nanotechnology...... 31 The ML community is vibrant and diverse. While works with super-linear algorithms in both time 3.7 3D Integrated Circuits...... 33 numerous ML techniques have made signifi- and space. Many ML algorithms display quadratic CONTENTS3.8 Universal Memory...... 38 3.9 Multicore...... 43 cant strides recently—too many to be covered or worse behavior and are inherently tailored to 3.10 Photonics...... 47 here—one stands out as a representative of the operate in a single-address space. While learning 3.11 Networking and Interconnectivity... 52 power of both looking to understanding how the algorithms that operate in linear time that allow us 3.12 Software-Defined Networks...... 55 human brain works and utilizing technological to train on very large datasets would be preferable, 3.13 High-Performance Computing advancements, namely, deep learning (DL) [1]. DL in practice, the growing volume of data often pre- (HPC) ...... 63 algorithms and systems have enjoyed remarkable cludes the application of standard single-machine 3.14 Cloud Computing...... 68 success in the speech, language, and image-pro- training algorithms. However, DL has successfully 3.15 The Internet of Things...... 74 3.16 Natural User Interfaces...... 77 cessing fields over the past few years. These new been explored in scaled-up environments involving 3.17 3D Printing ...... 81 algorithms swiftly beat current approaches to clusters of GPU and very large amounts of RAM as 3.18 Big Data and Analytics ...... 84 image analysis, acoustic modeling, and natural lan- well as scaled-out environments of thousands of 3.19 Machine Learning and guage understanding. Several factors have contrib- networked commodity servers. Initial results have Intelligent Systems ...... 89 uted to these achievements, starting with building been promising, and further progress should be 3.20 Computer Vision and Pattern Recognition...... 92 artificial neural networks that mimic the behavior expected in this area. 3.21 Life Sciences...... 96 of the human brain. Much like the brain, these mul- Most ML algorithms come with levers and knobs 3.22 Computational Biology and tilayered networks can capture information and re- for tuning the learning process to the data at Bioinformatics...... 102 spond to it. Arguably, these network architectures hand. Questions often arise about which configu- 3.23 Medical Robotics...... 108 build up an understanding of what image objects ration to use for a particular dataset and learning 4. DRIVERS AND DISRUPTORS...... 112 look, or phonemes sound, like. Across a wide range task. There are also challenges with collecting, 5. TECHNOLOGY COVERAGE IN of application domains, the abilities of DL to learn cleaning, and preparing large datasets for this IEEE XPLORE AND BY from unlabeled data have been broadly useful and IEEE SOCIETIES...... 115 process. New tools are needed to help people have led to significant advances. In acoustic mod- specify what they want to learn and determine 6. IEEE COMPUTER SOCIETY eling, the ability of DL architectures to separate IN 2022...... 120 how to measure the accuracy of the predictions multiple factors of variation in the input, such as made by the learned models. 7. SUMMARY AND NEXT STEPS.....124 speaker-dependent effects on speech acoustics, 8. AUTHORS...... 126 APPENDIX I. 23 Technologies Coverage 90 in IEEE Publications...... 134 CONTENTS 1. INTRODUCTION...... 5 2. SEAMLESS INTELLIGENCE Finally, there is the societal challenge of how to SCENARIO...... 8 guide actions and public policies in a world increas- 3. 23 TECHNOLOGIES IN 2022...... 14 ingly based on large-scale predictions made by 3.1 Security Cross-Cutting Issues...... 14 computing systems that no single human being can 3.2 The Open Intellectual fathom the scope of. Will people trust predictions Property Movement...... 17 from these systems? 3.3 Sustainability...... 20 3.4 Massively Online Open Courses...... 25 3.5 Quantum Computing ...... 28 3.19.4 Where We Think It Will Go 3.6 Device and Nanotechnology...... 31 For ML, the best is yet to come. Improvements in 3.7 3D Integrated Circuits...... 33 ML, including new DL architectures and optimi- CONTENTS3.8 Universal Memory...... 38 3.9 Multicore...... 43 zation strategies, are being explored broadly in the ML field with applications normally exclusively 3.10 Photonics...... 47 part of a new Presidential focus aimed at revolu- 3.11 Networking and Interconnectivity... 52 reserved for humans including facial recognition, tionizing our understanding of the human brain. 3.12 Software-Defined Networks...... 55 image object recognition (tagging), language anal- The objective of this initiative is to produce a new 3.13 High-Performance Computing ysis, conversation, and translation. With computers’ (HPC) ...... 63 ability to process and store vast amounts of infor- dynamic picture of the brain that shows how indi- 3.14 Cloud Computing...... 68 mation at extremely high speeds, we must expect vidual cells and complex neural aggregates interact 3.15 The Internet of Things...... 74 in both time and space. The results from BRAIN 3.16 Natural User Interfaces...... 77 that ML-based computing systems in some cases soon will exceed human capabilities. Combining will fill major gaps in our current knowledge and 3.17 3D Printing ...... 81 create opportunities for exploring exactly how the 3.18 Big Data and Analytics ...... 84 ML-based systems in order to create ensembles brain enables the human body to capture, process, 3.19 Machine Learning and that exhibit human-like intelligent behavior in the Intelligent Systems ...... 89 aforementioned domains has the potential to enact analyze, store, and retrieve vast quantities of infor- 3.20 Computer Vision and Pattern an even greater change to human society than mation. These insights may lead to new hardware Recognition...... 92 and software learning architectures that have the 3.21 Life Sciences...... 96 was already experienced by the computing revolution of the last 50 years. We are facing a unique potential to revolutionize the ML field. The Human 3.22 Computational Biology and Brain Flagship project of the European Commission Bioinformatics...... 102 opportunity to build systems that really become 3.23 Medical Robotics...... 108 empowering agents that fundamentally understand has similar goals [3]. 4. DRIVERS AND DISRUPTORS...... 112 our intent and continue to work on our behalf to We already mentioned that most ML algorithms 5. TECHNOLOGY COVERAGE IN complement us in our daily life. require tradeoffs to execute in reasonable time and IEEE XPLORE AND BY space. We suspect that computers as we know IEEE SOCIETIES...... 115 3.19.5 Potential Disruptions them today are not truly optimized for this class of 6. IEEE COMPUTER SOCIETY problems. This is where quantum computing comes IN 2022...... 120 The NIH Brain Research through Advancing Inno- vative Neurotechnologies (BRAIN) [2] Initiative is into play. By mixing quantum computing—which 7. SUMMARY AND NEXT STEPS.....124 is extremely well suited at finding global minima in 8. AUTHORS...... 126 APPENDIX I. 23 Technologies Coverage 91 in IEEE Publications...... 134 CONTENTS 1. INTRODUCTION...... 5 2. SEAMLESS INTELLIGENCE multidimensional spaces—with traditional comput- SCENARIO...... 8 ing systems, we may experience a true revolution in 3. 23 TECHNOLOGIES IN 2022...... 14 the capabilities of ML. Quantum ML may enable the 3.1 Security Cross-Cutting Issues...... 14 most creative problem-solving process allowed by 3.2 The Open Intellectual the laws of nature. Property Movement...... 17 3.3 Sustainability...... 20 3.4 Massively Online Open Courses...... 25 3.19.6 Summary 3.5 Quantum Computing ...... 28 ML is about building better models of the environ- 3.6 Device and Nanotechnology...... 31 ment in order to make predictions that are more CONTENTS3.7 3D Integrated Circuits...... 33 accurate. If we want to cure diseases, we need 3.8 Universal Memory...... 38 3.20 Computer Vision and 3.9 Multicore...... 43 better models of how they evolve. If we want to 3.10 Photonics...... 47 combat climate change, we need better models of Pattern Recognition 3.11 Networking and Interconnectivity... 52 what is happening to our global climate. 3.12 Software-Defined Networks...... 55 The directions and goals of ML fields are bold. They 3.20.1 Introduction 3.13 High-Performance Computing Computer vision studies methods to extract informa- (HPC) ...... 63 span explorations of the basic science of ML to un- 3.14 Cloud Computing...... 68 derstanding how to best solve practical problems tion from pictures, from video, and from laser ranger 3.15 The Internet of Things...... 74 and perform specific predictions. The development data. Pattern recognition studies methods to prepare 3.16 Natural User Interfaces...... 77 of more efficient and powerful tools to support the features from a range of signals, then exploits them 3.17 3D Printing ...... 81 engineering practices of ML are strongly needed. to tell where the signal came from. While some work 3.18 Big Data and Analytics ...... 84 Tools and methods that let nonexperts do a great clearly belongs to only one or the other discipline, 3.19 Machine Learning and there is significant interaction between the two. For Intelligent Systems ...... 89 job with their own predictive modeling are needed example, the main conference in each discipline is 3.20 Computer Vision and Pattern to truly empower users with machines that learn. Recognition...... 92 the IEEE Computer Vision and Pattern Recognition 3.21 Life Sciences...... 96 3.19.7 References Conference. 3.22 Computational Biology and Bioinformatics...... 102 [1] A. Ng and J. Dean, “Building High-level Features Us- One key topic is recognition, where programs try 3.23 Medical Robotics...... 108 ing Large Scale Unsupervised Learning,” 2012 to produce semantic descriptions of the world from 4. DRIVERS AND DISRUPTORS...... 112 [2] The NIH BRAIN Initiative, http://www.nih.gov/sci- visual information. Recognition systems might aim at detecting which objects are present in a picture and 5. TECHNOLOGY COVERAGE IN ence/brain, 2013. where. Alternatively, systems might try to identify the IEEE XPLORE AND BY [3] The human Brain Project, https://www.humanbrain- IEEE SOCIETIES...... 115 scene (is this a kitchen, forest, or arena?) or estimate project.eu/, 2014. 6. IEEE COMPUTER SOCIETY the free space and the major physical obstacles. As IN 2022...... 120 another example, recognition systems might try to 7. SUMMARY AND NEXT STEPS.....124 detect pedestrians (for vehicle safety applications) or 8. AUTHORS...... 126 APPENDIX I. 23 Technologies Coverage 92 in IEEE Publications...... 134 CONTENTS 1. INTRODUCTION...... 5 2. SEAMLESS INTELLIGENCE identify human activities (for human-computer inter- 3.20.2 State of the Art SCENARIO...... 8 face applications, surveillance applications, and some These disciplines have flourished and become a 3. 23 TECHNOLOGIES IN 2022...... 14 kinds of medical screening). major source of important application technologies. 3.1 Security Cross-Cutting Issues...... 14 Recognition methods can now distinguish between 3.2 The Open Intellectual A second key topic is registration, where programs Property Movement...... 17 take signals obtained in different ways, at different thousands of objects, if the objects appear in unclut- 3.3 Sustainability...... 20 times, or from different locations, and identify corre- tered scenes; a significant error rate is tolerable, and 3.4 Massively Online Open Courses...... 25 sponding structures. Registration is a crucial topic this error rate is falling quickly. Recognition methods 3.5 Quantum Computing ...... 28 in medical applications. For example, registration can also distinguish between hundreds of types 3.6 Device and Nanotechnology...... 31 programs might take magnetic resonance and x-ray of scenes and find tens to hundreds of objects in CONTENTS3.7 3D Integrated Circuits...... 33 tomographic images of the same anatomy, and warp cluttered scenes, again with a significant and object 3.8 Universal Memory...... 38 dependent error rate, and that error rate is falling 3.9 Multicore...... 43 them so that corresponding structures lie on top quickly. Such methods can recover body configu- 3.10 Photonics...... 47 of one another. This is valuable, because different 3.11 Networking and Interconnectivity... 52 imaging modes provide different types of information ration information from range data well enough to 3.12 Software-Defined Networks...... 55 about different structures. support commercial computer game interfaces. 3.13 High-Performance Computing They can also recover accurate, detailed geometric (HPC) ...... 63 Registration methods are an important part of recon- reconstructions of domains on the scale of a city 3.14 Cloud Computing...... 68 struction, where programs try to produce geometric, center from video and recover very detailed geo- 3.15 The Internet of Things...... 74 photometric, or physical models of the world from metric reconstructions of moderately sized objects 3.16 Natural User Interfaces...... 77 visual information. For example, reconstruction pro- from small numbers of photographs. Reconstruction 3.17 3D Printing ...... 81 grams can take tourist photographs or video of a city 3.18 Big Data and Analytics ...... 84 methods can recover tolerable estimates of depth and produce a detailed, three-dimensional geometric from single photographs: they are a routine part of 3.19 Machine Learning and model of it; take a video sequence and recover how Intelligent Systems ...... 89 modern radiography. New applications for estab- 3.20 Computer Vision and Pattern the camera moved during the sequence; and recover lished methods appear regularly, and there is signifi- Recognition...... 92 physical properties of surfaces—corresponding to cant industrial demand for skilled personnel in these 3.21 Life Sciences...... 96 roughness, lightness, shininess, etc. —from pictures. areas. A variety of effects drive this success: 3.22 Computational Biology and Reconstruction methods are often used to support Bioinformatics...... 102 applications in mapping, entertainment, computer • Technical advances. The last 30 years have 3.23 Medical Robotics...... 108 graphics, and virtual reality. seen a huge expansion in understanding of 4. DRIVERS AND DISRUPTORS...... 112 methods and techniques to solve important 5. TECHNOLOGY COVERAGE IN Yet another key topic is biometrics, where programs intermediate problems. For example, there are IEEE XPLORE AND BY try to identify individuals from various important bio- very good constructions for image features; IEEE SOCIETIES...... 115 logical measurements. For example, programs might there are excellent classification and detection 6. IEEE COMPUTER SOCIETY recognize faces from pictures, irises from eye-scan methods; the underlying geometric mathemat- IN 2022...... 120 data, thumbprints from finger-scan data, or the load ics for reconstruction has been worked out in 7. SUMMARY AND NEXT STEPS.....124 that a person is carrying from his or her gait. detail; and there are excellent basic registration 8. AUTHORS...... 126 APPENDIX I. 23 Technologies Coverage 93 in IEEE Publications...... 134 CONTENTS 1. INTRODUCTION...... 5 2. SEAMLESS INTELLIGENCE algorithms. Scientists and engineers attacking 3.20.3 Challenges SCENARIO...... 8 new applications have a deep base of estab- The core problems in these areas are immensely 3. 23 TECHNOLOGIES IN 2022...... 14 lished methods to draw on. difficult. It is not yet known how to formulate them 3.1 Security Cross-Cutting Issues...... 14 • Access to data. The rise of the Internet means correctly. What does it mean to recognize objects? 3.2 The Open Intellectual it’s easier than ever to create and use visual in- Property Movement...... 17 Should we name everything? Wouldn’t there be too formation. As a result, huge collections of train- 3.3 Sustainability...... 20 much in that description? Humans also seem to al- 3.4 Massively Online Open Courses...... 25 ing data are now available to support research. low the same object to have different categories— 3.5 Quantum Computing ...... 28 Furthermore, there is demand for solutions to sometimes it’s a screwdriver, sometimes it’s a lever, 3.6 Device and Nanotechnology...... 31 new problemsModern users want to search and sometimes it’s a punch. This is clearly a useful CONTENTS3.7 3D Integrated Circuits...... 33 for pictures using text descriptions, so there is way to think about objects, but it links recognition 3.8 Universal Memory...... 38 tremendous industry interest in methods that with the goals of action in ways that aren’t yet well 3.9 Multicore...... 43 can describe pictures. Modern users also want understood. As yet another example, there seem 3.10 Photonics...... 47 to remove objects from pictures, a demand that 3.11 Networking and Interconnectivity... 52 to be a very large number of different types of ob- drove deep research on how to select the ob- 3.12 Software-Defined Networks...... 55 ject out there, and examples of each type are rare. 3.13 High-Performance Computing ject to be removed and what to put in its place. This means that systems should likely be able to (a) (HPC) ...... 63 • Computing improvements. Research in comput- learn from few or no examples; (b) transfer infor- 3.14 Cloud Computing...... 68 er vision and pattern recognition poses excep- mation from types of object to other types; and (c) 3.15 The Internet of Things...... 74 tional demands on computation and storage. cope with surprise objects in sensible ways. 3.16 Natural User Interfaces...... 77 Really significant improvements in object rec- 3.17 3D Printing ...... 81 ognition and detection, registration, and recon- 3.18 Big Data and Analytics ...... 84 3.20.4 Where We Think It Will Go struction have closely tracked improvements in 3.19 Machine Learning and The next decade will likely see much more technol- Intelligent Systems ...... 89 available computing resources. ogy from these areas in the hands of consumers. 3.20 Computer Vision and Pattern • Application demand. Many important and Recognition...... 92 interesting problems require processing visual The trend is already established: most consumer 3.21 Life Sciences...... 96 data. For example, autonomous cars will have digital cameras have red-eye removal software; 3.22 Computational Biology and image search is offered by all big search engines; Bioinformatics...... 102 to process visual data (though it remains con- troversial what kinds of data, how it should be and the entertainment industry is a huge consumer 3.23 Medical Robotics...... 108 of reconstruction software. Each started about a 4. DRIVERS AND DISRUPTORS...... 112 processed, and what the results will be). There is also a tremendous pool of valuable applica- decade ago, and each is the result of a multidecade 5. TECHNOLOGY COVERAGE IN research literature that found an application. IEEE XPLORE AND BY tions around watching people, for example, to IEEE SOCIETIES...... 115 support the care of the frail, to build improved The last decade has seen a steady drumbeat in 6. IEEE COMPUTER SOCIETY computer interfaces, or to keep bad people performance improvements in core problems in IN 2022...... 120 out of sensitive places. All these applications computer vision and pattern recognition. Far more 7. SUMMARY AND NEXT STEPS.....124 require some processing of visual data. technologies are ready for application than 10 8. AUTHORS...... 126 APPENDIX I. 23 Technologies Coverage 94 in IEEE Publications...... 134 CONTENTS 1. INTRODUCTION...... 5 2. SEAMLESS INTELLIGENCE years ago. We expect to see routine application 3.20.5 Potential Disruptions SCENARIO...... 8 of recognition and reconstruction technologies in The most likely source of disruption in this area is 3. 23 TECHNOLOGIES IN 2022...... 14 consumer applications. that an unpredictable insight will lead to a sudden 3.1 Security Cross-Cutting Issues...... 14 improvement in performance of recognition or de- 3.2 The Open Intellectual Autonomous actors—vehicles, household care Property Movement...... 17 robots, search and rescue robots, and so on—need tection technologies. The consequences would be 3.3 Sustainability...... 20 to know what is going on around them. Improve- significant: quick, massive improvements in radiol- 3.4 Massively Online Open Courses...... 25 ments in detection, recognition, and reconstruction ogy; quick, massive improvements in cost control 3.5 Quantum Computing ...... 28 technologies will lead to safer, cheaper, and more in construction, the restaurant trade, and similar 3.6 Device and Nanotechnology...... 31 widely deployed autonomous actors. skilled service industries; and quick, massive im- 3.7 3D Integrated Circuits...... 33 provements in the safety of autonomous vehicles. CONTENTS3.8 Universal Memory...... 38 There will be tremendous impact on the data-min- 3.9 Multicore...... 43 ing world, too. Generally, information in pictures is 3.10 Photonics...... 47 locked away from current big data analysis. Im- 3.20.6 Summary 3.11 Networking and Interconnectivity... 52 provements in recognition technologies will expose Computer vision and pattern recognition seek to 3.12 Software-Defined Networks...... 55 the information in collections of visual data, which unlock information in pictures, video, ranger data, 3.13 High-Performance Computing and allied signals. These disciplines are now very (HPC) ...... 63 will allow important scientific fishing expeditions— 3.14 Cloud Computing...... 68 for example, are there structures in medical images large and have had a major but mostly indirect 3.15 The Internet of Things...... 74 that predict prognosis, but haven’t been noticed? impact on consumers. A range of significant tech- 3.16 Natural User Interfaces...... 77 nology improvements are currently in the pipeline Improvements in surveillance technologies will 3.17 3D Printing ...... 81 either between academia and industry, or between 3.18 Big Data and Analytics ...... 84 lead to improved cost-control in a wide variety of industry and consumers. Likely future impact is 3.19 Machine Learning and industries. In the construction industry, much value very great. Intelligent Systems ...... 89 is lost because it’s hard for anyone to ensure that 3.20 Computer Vision and Pattern all workers on a large construction site are usefully Recognition...... 92 engaged at all times (workers may not have what 3.20.7 References 3.21 Life Sciences...... 96 they need, for example, due to a scheduling break- D.A. Forsyth and J. Ponce, Computer Vision: A Modern 3.22 Computational Biology and Approach, 2e, Prentice-Hall, 2011. Bioinformatics...... 102 down). There is strong evidence emerging that 3.23 Medical Robotics...... 108 current recognition tools and future improvements R. Szeliski, Computer Vision: Algorithms and Applica- 4. DRIVERS AND DISRUPTORS...... 112 will revolutionize construction.It is likely that these tions, Springer, 2010. 5. TECHNOLOGY COVERAGE IN remarks apply to skilled service industries with R. Hartley and A. Zisserman, Multiple View Geometry in IEEE XPLORE AND BY large numbers of workers such as the restaurant Computer Vision, 2e, Cambridge University Press, 2003. IEEE SOCIETIES...... 115 trade, the vehicle manufacturing and maintenance 6. IEEE COMPUTER SOCIETY business, and so on. IN 2022...... 120 7. SUMMARY AND NEXT STEPS.....124 8. AUTHORS...... 126 APPENDIX I. 23 Technologies Coverage 95 in IEEE Publications...... 134 CONTENTS 1. INTRODUCTION...... 5 2. SEAMLESS INTELLIGENCE These trends have increased LS rate of growth, SCENARIO...... 8 which exceeds most major sectors of the economy 3. 23 TECHNOLOGIES IN 2022...... 14 in developed countries. This growth has result- 3.1 Security Cross-Cutting Issues...... 14 ed in a fierce economic competition to motivate 3.2 The Open Intellectual industry (and government) to look for competitive Property Movement...... 17 advantages derived from a world-class academic 3.3 Sustainability...... 20 3.4 Massively Online Open Courses...... 25 research establishmentS. These stakeholders are 3.5 Quantum Computing ...... 28 increasing their demand for and expectation of 3.6 Device and Nanotechnology...... 31 “intelligence” in devices and systems, e.g., ubiq- 3.7 3D Integrated Circuits...... 33 3. uitous computing, communications, sensing, etc. CONTENTS3.8 Universal Memory...... 38 [Khargonekar] 3.9 Multicore...... 43 3.21 Life Sciences 3.10 Photonics...... 47 LS disciplines include bioengineering, biomedical 3.11 Networking and Interconnectivity... 52 3.21.1 Introduction engineering, healthcare technology, communica- 3.12 Software-Defined Networks...... 55 The life sciences (LS) industry uses modern biological tions, and computational technical domains such 3.13 High-Performance Computing techniques and supporting technologies with a goal as big data analytics, dependable and secure (HPC) ...... 63 computing, high-performance computing, informa- 3.14 Cloud Computing...... 68 to improve human and animal health; address threats tion technology, knowledge and data engineering, 3.15 The Internet of Things...... 74 to the environment; improve crop production; contain 3.16 Natural User Interfaces...... 77 emerging and existing diseases; and improve cur- machine learning, multimedia, parallel and distrib- 3.17 3D Printing ...... 81 rently used manufacturing technologies. LS industry uted processing systems, pattern analysis, security 3.18 Big Data and Analytics ...... 84 sectors include pharmaceuticals, biotechnology, and privacy, software engineering, and visualization 3.19 Machine Learning and and computer graphics. Intelligent Systems ...... 89 chemicals, medical devices, medical products and 3.20 Computer Vision and Pattern technology, and healthcare services. LS industry Recognition...... 92 employment has significant size and growth world- 3.21.2 State of the Art 3.21 Life Sciences...... 96 wide. AAAS reports job ads today are roughly spread There is a shift occurring in the character of LS. 3.22 Computational Biology and evenly across Europe, Asia, and US. In the US there The pharmaceutical, medical device, and biotech- Bioinformatics...... 102 were 1.61 million jobs in 2010, spanning over 70,000 nology sectors have led the way, while a persistent 3.23 Medical Robotics...... 108 individual companies [Battelle/BiO]. US healthcare focus on discovery, delivery, and continuous inno- 4. DRIVERS AND DISRUPTORS...... 112 spending grew 3.9 percent in 2011, reaching $2.7 vation remains a driver for growth. The subsectors 5. TECHNOLOGY COVERAGE IN trillion or $8,680 per person [CMS]. of engineering have started to bring efficiency, IEEE XPLORE AND BY IEEE SOCIETIES...... 115 Currently, LS is experiencing five mega-trends: effectiveness, and modern processes to advance the theories and research that bring LS to import- 6. IEEE COMPUTER SOCIETY increasing expectations and cost of healthcare, IN 2022...... 120 quality of life of aging populations, major challeng- ant and practical applications. [Thakor] 7. SUMMARY AND NEXT STEPS.....124 es in biology/medicine, and compliance challenges. For example, the grand challenge of affordable 8. AUTHORS...... 126 APPENDIX I. 23 Technologies Coverage 96 in IEEE Publications...... 134 CONTENTS 1. INTRODUCTION...... 5 2. SEAMLESS INTELLIGENCE and effective healthcare has spurred Singapore to could be vanquished—if we make full use of the SCENARIO...... 8 help physicians provide more complete diagnoses; tools of modern drug design and allow doctors the 3. 23 TECHNOLOGIES IN 2022...... 14 conduct regular treatments; follow evidence-based use of modern data gathering and analytical tools 3.1 Security Cross-Cutting Issues...... 14 practice; and conform to workflow treatments by when prescribing drugs to their patients. [Huber] 3.2 The Open Intellectual using enhanced, accessible big patent/healthcare Property Movement...... 17 LS is not just the interface between the disciplines data and analytics. Singapore has also started a 3.3 Sustainability...... 20 of engineering and biomedical sciences but also national medical records database that has one 3.4 Massively Online Open Courses...... 25 the convergent overlaps among bio-, nano-, and 3.5 Quantum Computing ...... 28 record per citizen for that person’s lifespan. This info-technologies. These interfaces are very ex- 3.6 Device and Nanotechnology...... 31 big data approach will expand to physicians and citing and fertile zones for highly original ideas, 3.7 3D Integrated Circuits...... 33 their healthcare centers, and this new environment experiments, and discoveries. LS requires quicker CONTENTS3.8 Universal Memory...... 38 will help shape how professional cultures can work translation—to bring discoveries into useful ap- 3.9 Multicore...... 43 better together. [Ying-I,] 3.10 Photonics...... 47 plications that help our patients, restore or assist 3.11 Networking and Interconnectivity... 52 human function, and address major needs in our 3.12 Software-Defined Networks...... 55 Convergence as the Fourth Revolution society. [Chuan] 3.13 High-Performance Computing The convergence of the life sciences, physical The public is another participant in this revolution (HPC) ...... 63 sciences, and engineering in advancing healthcare as people are downloading and running fold-it-to- 3.14 Cloud Computing...... 68 has four phases or revolutions: 3.15 The Internet of Things...... 74 solve puzzles for LS research. 3.16 Natural User Interfaces...... 77 • 1st golden age of biochemistry, 1900-1950 Framework standards are important. Classifying 3.17 3D Printing ...... 81 [Radda] and modeling LS information is daunting, and using 3.18 Big Data and Analytics ...... 84 • 2nd revolution: molecular and cellular biology a multiscale modeling approach could support the 3.19 Machine Learning and with Watson and Crick, 1950-2000 Intelligent Systems ...... 89 goal of building a complete virtual physiological • 3rd revolution: genomics, 2000-present 3.20 Computer Vision and Pattern human. Similar to the computer stack, the LS stack Recognition...... 92 • 4th revolution: integration of LS at the molec- starts at the nanometer scale and builds up to 3.21 Life Sciences...... 96 ular level with engineering, physical sciences, one-meter scale: quantum mechanics molecular 3.22 Computational Biology and and mathematics/computational science. This Bioinformatics...... 102 network proteins cell types organ tissue will increase understanding of how compo- 3.23 Medical Robotics...... 108 organ organ system, e.g., torso organism. nents collaborate to create complex biological [Hunter] 4. DRIVERS AND DISRUPTORS...... 112 systems and promote the flow of results into 5. TECHNOLOGY COVERAGE IN practice. [Sharp] IEEE XPLORE AND BY Teamwork IEEE SOCIETIES...... 115 The converging, synergistic power of the biochem- There is now a widespread recognition of the crit- 6. IEEE COMPUTER SOCIETY ical and digital revolutions enables us to read every ical importance of multidisciplinary team research IN 2022...... 120 letter of life’s code, create precisely targeted drugs, in government, industry, and academy. Real-world 7. SUMMARY AND NEXT STEPS.....124 and tailor their use to individual patients. Cancer, problems do not come in disciplinary-shaped boxes 8. AUTHORS...... 126 diabetes, Alzheimer’s, and countless other killers APPENDIX I. 23 Technologies Coverage 97 in IEEE Publications...... 134 CONTENTS 1. INTRODUCTION...... 5 2. SEAMLESS INTELLIGENCE [Jeffrey]. Large-space opportunities, such as LS, example is the 6,500 technologists at CERN work- SCENARIO...... 8 requiring interdisciplinary work have more risks ing on one problem (Higgs), a physics grand chal- 3. 23 TECHNOLOGIES IN 2022...... 14 than specific projects. [Jeffrey] lenge. [Radda] 3.1 Security Cross-Cutting Issues...... 14 3.2 The Open Intellectual Barriers to communication between disciplines To encourage graduates with multidisciplinary ex- Property Movement...... 17 as they have naturally grown into a “stovepipe” perience, universities need to 1) promote cross-dis- 3.3 Sustainability...... 20 reinforce the communication problem [Dewulf]. ciplinary interactions among their students, e.g., 3.4 Massively Online Open Courses...... 25 Cross-disciplinary project issues [Khargonekar] educational, sports, arts, and social facilities and 3.5 Quantum Computing ...... 28 include differences in terminology and methods, dormitories; 2) develop programs specifically 3.6 Device and Nanotechnology...... 31 setting priorities, effort needed to gain real un- focused on the interfaces of key disciplines; and 3) 3.7 3D Integrated Circuits...... 33 derstanding of the key technical and nontechnical encourage them to collaborate together in interna- CONTENTS3.8 Universal Memory...... 38 3.9 Multicore...... 43 issues, promotion and tenure, professional recog- tional research “collaboratories” working on inter- 3.10 Photonics...... 47 nition, publications in discipline-based journals, disciplinary research projects. [Chuan] 3.11 Networking and Interconnectivity... 52 intellectual property negotiations, dealing with gov- The US National Science Foundation is funding 3.12 Software-Defined Networks...... 55 ernment regulations, and potential loss of propri- new models for graduate education and training 3.13 High-Performance Computing etary information. (HPC) ...... 63 in an environment fostering collaborative research 3.14 Cloud Computing...... 68 An interdisciplinary shift in demand for talent within that transcends the traditional disciplinary bound- 3.15 The Internet of Things...... 74 the biotech industry is moving away from hiring aries and facilitates diversity in student participa- 3.16 Natural User Interfaces...... 77 narrowly focused specialists to individuals with tion and preparation. One of these NSF programs 3.17 3D Printing ...... 81 interdisciplinary academic training, highlighting is Integrative Graduate Education and Research 3.18 Big Data and Analytics ...... 84 the latest LS workforce trends. Hiring managers Traineeship (IGERT) [Ramasubramanian] 3.19 Machine Learning and Intelligent Systems ...... 89 and industry leaders are starting to profile their 3.20 Computer Vision and Pattern workforce-related capability needs, which include 3.21.3 Challenges Recognition...... 92 soft skills and the ability to work effectively across In 2008, the NSF identified 14 engineering grand 3.21 Life Sciences...... 96 disciplines. There is a clear shift in the industry’s challenges, and in 2013, the US National Academy 3.22 Computational Biology and demand for talent away from senior scientist po- Bioinformatics...... 102 of Engineering revisited them. In both studies, four sitions that tend to be more highly specialized and 3.23 Medical Robotics...... 108 LS-related grand challenges remain: narrowly focused to a talent pool consisting of indi- 4. DRIVERS AND DISRUPTORS...... 112 viduals who have interdisciplinary academic train- • advance healthcare informatics, 5. TECHNOLOGY COVERAGE IN ing and the ability to work broadly across multiple • engineer better medicines, IEEE XPLORE AND BY • manage the nitrogen cycle, and IEEE SOCIETIES...... 115 areas and in project teams where not everyone has to be an expert in everything. [Nugent] • reverse engineer the brain. 6. IEEE COMPUTER SOCIETY IN 2022...... 120 Ultimately, scientists and engineers must learn Some proposals transforming health and wellness 7. SUMMARY AND NEXT STEPS.....124 how to work in teams. An outstanding teamwork include genomics-enabled personalized medicine, 8. AUTHORS...... 126 APPENDIX I. 23 Technologies Coverage 98 in IEEE Publications...... 134 CONTENTS 1. INTRODUCTION...... 5 2. SEAMLESS INTELLIGENCE which would replace the creation of generic pro- 3.21.4 Where We Think It Will Go SCENARIO...... 8 prietary medicines. [Kun] In a related case study, As the pace of intertwined discovery and invention 3. 23 TECHNOLOGIES IN 2022...... 14 patients with one type of leukemia received a increases, we are on parallel paths of evolution and 3.1 Security Cross-Cutting Issues...... 14 one-time experimental therapy several years ago inspiration, where our computer scientists can both 3.2 The Open Intellectual and some remain cancer-free today. At least six Property Movement...... 17 learn and provide insights. Automation in all walks research groups have treated more than 120 pa- 3.3 Sustainability...... 20 of life will be the most disruptive technology in 3.4 Massively Online Open Courses...... 25 tients with many types of blood and bone marrow coming decades. For healthcare, this means 3.5 Quantum Computing ...... 28 cancers, with stunning results. [Marchione] • instant, expert diagnostic advice; 3.6 Device and Nanotechnology...... 31 For quick transfer of medical device development • personal preventative health advice; 3.7 3D Integrated Circuits...... 33 to the patient, proposals have been drafted to use CONTENTS3.8 Universal Memory...... 38 • enhanced bedside care; and 3.9 Multicore...... 43 modeling (virtual prototyping) as a tool for regula- • big data analysis of clinical trials and unstruc- 3.10 Photonics...... 47 tory approval. [Schiestl] Cardiovascular diseases tured research data. 3.11 Networking and Interconnectivity... 52 are the major cause of death, and the cardiovas- Big data in health and medicine will pull together 3.12 Software-Defined Networks...... 55 cular health informatics used in wearable medical 3.13 High-Performance Computing devices technologies and unobtrusive measure- databases with patients’ outcomes, leading to a (HPC) ...... 63 ments connect through a body sensor network. translation of research results directly into medical 3.14 Cloud Computing...... 68 Requirements attributes include advances in practice without delay. [Wah] 3.15 The Internet of Things...... 74 3.16 Natural User Interfaces...... 77 miniaturization, and intelligent, network, digital, and The role of scholarly Societies is to provide guid- 3.17 3D Printing ...... 81 standardization. [Zhang] ance not only on technical feasibility but on social 3.18 Big Data and Analytics ...... 84 Healthcare once consisted solely of killing germs, and psychological impact. Our challenge is to 3.19 Machine Learning and but tomorrow’s regimens will be guided and adjust- optimize deployment of willingly tolerated, naturally Intelligent Systems ...... 89 intelligent computers for healthcare and clinical 3.20 Computer Vision and Pattern ed using relevant biomarkers specific to individual Recognition...... 92 patients. 21st-century medicine is hampered by a research. [Finkel] 3.21 Life Sciences...... 96 regulatory regime built for the science of the 20th 3.22 Computational Biology and century. The search for cancer’s silver bullet, some- 3.21.5 Potential Disruptions Bioinformatics...... 102 thing that meets the FDA gauntlet, is still going on, A New Biology for the 21st Century report from the 3.23 Medical Robotics...... 108 but there has been limited success in reducing per National Academy of Sciences (NAS), National 4. DRIVERS AND DISRUPTORS...... 112 capita deaths from cancer since 1950 [Bashir]. Fur- Academy of Engineering (NAE), Institute of Medi- 5. TECHNOLOGY COVERAGE IN thermore, new medical devices are going to Europe cine (IOM), and National Research Council (NRC) IEEE XPLORE AND BY IEEE SOCIETIES...... 115 for regulatory approval because it takes half the announces biology is at an inflection point, poised time of obtaining FDA approval. [PWC] The FDA 6. IEEE COMPUTER SOCIETY to help solve major societal problems related IN 2022...... 120 still operates according to the requirements of the to food, environment, energy, and health using 7. SUMMARY AND NEXT STEPS.....124 age of mass drugs and must be reformed. [Huber] a cross-discipline integration of LS research by 8. AUTHORS...... 126 APPENDIX I. 23 Technologies Coverage 99 in IEEE Publications...... 134 CONTENTS 1. INTRODUCTION...... 5 2. SEAMLESS INTELLIGENCE physical, computational, Earth scientists, and engi- [CMS] CMS, “National Health Expenditures Data, His- SCENARIO...... 8 neers. [Sharp et al] torical,” 2012. 3. 23 TECHNOLOGIES IN 2022...... 14 Despite the potential of recent advances, there [Dewulf] Dewulf, A., “A framing approach to cross-dis- 3.1 Security Cross-Cutting Issues...... 14 is still much to be done to move from identifying ciplinary research collaboration: experiences from a 3.2 The Open Intellectual large-scale research project on adaptive water manage- Property Movement...... 17 parts to defining complex biological systems. Fur- ment,” Ecology and Society, 2007. 3.3 Sustainability...... 20 thermore, the systems design, manipulation, and 3.4 Massively Online Open Courses...... 25 prediction needed for practical applications such [Finkel] Finkel, Alan, “Panel on International Coopera- 3.5 Quantum Computing ...... 28 as ecosystem repair or individualized medicine tion in Life Sciences,” IEEE Life Sciences Grand Chal- 3.6 Device and Nanotechnology...... 31 are still well beyond current capabilities. The “new lenges Conference (LSGCC 2013), 2013. 3.7 3D Integrated Circuits...... 33 biology” will provide a framework to connect bio- [Huber] Huber, Peter W., The Cure in the Code: How CONTENTS3.8 Universal Memory...... 38 20th Century Law Is Undermining 21st Century Medicine, 3.9 Multicore...... 43 logical research with advances in other branches of 3.10 Photonics...... 47 science and engineering. [Kamm]. 2013. 3.11 Networking and Interconnectivity... 52 [Hunter] Hunter, Peter, “Frontiers of Computational 3.12 Software-Defined Networks...... 55 3.21.6 Summary Medicine,” IEEE Life Sciences Grand Challenges Con- 3.13 High-Performance Computing ference (LSGCC 2013), 2013. (HPC) ...... 63 LS industry is experiencing a large growth in the 21st 3.14 Cloud Computing...... 68 century, surpassing most other sectors. Most of the [Jeffrey] Jeffrey, P., “Smoothing the Waters: Observa- 3.15 The Internet of Things...... 74 growth is in addressing new needs with new solutions on the Process of Cross-Disciplinary Research 3.16 Natural User Interfaces...... 77 tions. These solutions were created with the help of Collaboration,” Social Studies of Science, 33 (4):539– 3.17 3D Printing ...... 81 new computational technologies and the technolo- 562, 2003. 3.18 Big Data and Analytics ...... 84 gists who are comfortable and effective in cross-dis- [Kamm] Kamm, Roger D, “Engineering Microvascular 3.19 Machine Learning and Intelligent Systems ...... 89 ciplinary teams. There future team members will need Networks for Therapeutic and in vitro Applications,” 3.20 Computer Vision and Pattern cross-discipline education and training. IEEE Life Sciences Grand Challenges Conference (LS- Recognition...... 92 GCC 2012), 2012. 3.21 Life Sciences...... 96 3.21.7 References [Khargonekar] Khargonekar, Pramod P., “Issues and 3.22 Computational Biology and Perspectives on Cross-Disciplinary Research and the Bioinformatics...... 102 [Bashir] Bashir, Rashid, “Cell-based Systems, Bio-fab- Role of Industry,” IEEE CDC 2003. 3.23 Medical Robotics...... 108 rication, and Cellular Machines,” IEEE Life Sciences Grand Challenges Conference (LSGCC 2012), 2012. 4. DRIVERS AND DISRUPTORS...... 112 [Kun] Kun, Luis G., National Defense University, personal communication. 5. TECHNOLOGY COVERAGE IN [Battelle/BiO] Battelle/BiO, Battelle and Biotechnology IEEE XPLORE AND BY industry Organization, State Bioscience Industry Devel- [Marchione] Marchione, Marilynn, “Gene therapy scores IEEE SOCIETIES...... 115 opment, 2012. big wins against blood cancers,” The Associated Press, 6. IEEE COMPUTER SOCIETY [Chuan] Chuan, Tan Chorh, “Life Sciences at the 2013. IN 2022...... 120 National University of Singapore,” IEEE Life Sciences [Nugent] Nugent, Kathy L. & Kulkarni, Avi, Nature Bio- 7. SUMMARY AND NEXT STEPS.....124 Grand Challenges Conference (LSGCC 2013), 2013. technology, September 2013. 8. AUTHORS...... 126 APPENDIX I. 23 Technologies Coverage 100 in IEEE Publications...... 134 CONTENTS 1. INTRODUCTION...... 5 2. SEAMLESS INTELLIGENCE [PWC] PWC 2011, “Medical Technology Innovation SCENARIO...... 8 Scorecard: US Falling Behind in the Race for Global 3. 23 TECHNOLOGIES IN 2022...... 14 Leadership,” 2011. 3.1 Security Cross-Cutting Issues...... 14 [Radda] Radda, George, “Biomedical Knowledge in the 3.2 The Open Intellectual Service of Man: Social Responsibility of the Scientist,” Property Movement...... 17 IEEE Life Sciences Grand Challenges Conference (LS- 3.3 Sustainability...... 20 GCC 2013), 2013. 3.4 Massively Online Open Courses...... 25 3.5 Quantum Computing ...... 28 [Ramasubramanian] Ramasubramanian, Melur K., “Next 3.6 Device and Nanotechnology...... 31 Generation Convergence- Driven Research and Inter- 3.7 3D Integrated Circuits...... 33 disciplinary Workforce and Development Efforts,” IEEE CONTENTS3.8 Universal Memory...... 38 Life Sciences Grand Challenges Conference (LSGCC 3.9 Multicore...... 43 2012), 2012. 3.10 Photonics...... 47 3.11 Networking and Interconnectivity... 52 [Schiestl], Schiestl, Randy, “Translation - From Bench 3.12 Software-Defined Networks...... 55 to Bedside, Medical Device Development,” IEEE Life 3.13 High-Performance Computing Sciences Grand Challenges Conference (LSGCC 2012), (HPC) ...... 63 2012. 3.14 Cloud Computing...... 68 [Sharp] Sharp, Phillip A. et al., “New Biology for the 21st 3.15 The Internet of Things...... 74 Century,” IEEE Life Sciences Grand Challenges Confer- 3.16 Natural User Interfaces...... 77 ence (LSGCC 2012), 2012. 3.17 3D Printing ...... 81 3.18 Big Data and Analytics ...... 84 [Thakor] Thakor, Nitish, “Frontiers of Human Brain,” IEEE 3.19 Machine Learning and Life Sciences Grand Challenges Conference (LSGCC Intelligent Systems ...... 89 2012), 2012. 3.20 Computer Vision and Pattern Recognition...... 92 [Wah] Wah, Benjamin, “Chinese University of Hong 3.21 Life Sciences...... 96 Kong, Life Sciences Big Data,” IEEE Life Sciences Grand 3.22 Computational Biology and Challenges Conference (LSGCC 2013), 2013. Bioinformatics...... 102 [Ying-I] Ying-I, Yong, “Affordability and Effective Health- 3.23 Medical Robotics...... 108 care in Singapore,” IEEE Life Sciences Grand Challeng- 4. DRIVERS AND DISRUPTORS...... 112 es Conference (LSGCC 2013), 2013. 5. TECHNOLOGY COVERAGE IN [Zhang] Zhang, Y.T., “Cardiovascular Health Informat- IEEE XPLORE AND BY IEEE SOCIETIES...... 115 ics: Wearable Technologies and Unobtrusive Measure- ments,” IEEE Life Sciences Grand Challenges Confer- 6. IEEE COMPUTER SOCIETY ence (LSGCC 2013), 2013. IN 2022...... 120 7. SUMMARY AND NEXT STEPS.....124 8. AUTHORS...... 126 APPENDIX I. 23 Technologies Coverage 101 in IEEE Publications...... 134 CONTENTS 1. INTRODUCTION...... 5 2. SEAMLESS INTELLIGENCE comprise the field of genomic bioinformatics. This SCENARIO...... 8 includes the initial acquisition of raw sequencing 3. 23 TECHNOLOGIES IN 2022...... 14 data, the interpretation and assembly of such data 3.1 Security Cross-Cutting Issues...... 14 into partial and complete genomes, the analysis of 3.2 The Open Intellectual sequenced genomes for statistical correlations in- Property Movement...... 17 dicative of diseases and other traits, and the mining 3.3 Sustainability...... 20 3.4 Massively Online Open Courses...... 25 of genomes for overrepresented motifs and other 3.5 Quantum Computing ...... 28 sequence features [1] . 3.6 Device and Nanotechnology...... 31 3.7 3D Integrated Circuits...... 33 3.22 Computational Biology Structural Bioinformatics CONTENTS3.8 Universal Memory...... 38 and Bioinformatics The analysis, modeling, and simulation of biological 3.9 Multicore...... 43 3.10 Photonics...... 47 macromolecules—namely, proteins, DNA (Deoxy- 3.11 Networking and Interconnectivity... 52 3.22.1 Background riboNucleic Acid), and RNA (RiboNucleic Acid)— 3.12 Software-Defined Networks...... 55 The importance of computation in the acquisition, comprise structural bioinformatics. The holy grail of 3.13 High-Performance Computing analysis, and modeling of biological systems has the field has been, for several decades, the predic- (HPC) ...... 63 been steadily increasing for the past several de- tion of the three-dimensional structure of proteins 3.14 Cloud Computing...... 68 from their amino acid sequence [2]. A similar chal- 3.15 The Internet of Things...... 74 cades. Contemporary bioinformatics and compu- lenge remains open for RNA molecules [3]. Beyond 3.16 Natural User Interfaces...... 77 tational biology, twin fields divided roughly along 3.17 3D Printing ...... 81 the lines of data acquisition and analysis for the structure prediction, structural bioinformatics is 3.18 Big Data and Analytics ...... 84 former and phenomenological modeling for the lat- concerned with the analysis and simulation of bio- 3.19 Machine Learning and ter, comprise a strikingly wide range of topics and molecules to predict their interactions with other Intelligent Systems ...... 89 disciplines. Owing to the intrinsic breadth of biolog- biomolecules and to infer useful physico-chemical 3.20 Computer Vision and Pattern properties [4]. Recognition...... 92 ical phenomena, which ranges from the molecular 3.21 Life Sciences...... 96 to the cellular to the organismal and ecological, 3.22 Computational Biology and computational biologists and bioinformaticists Systems Modeling Bioinformatics...... 102 must grapple with a diverse set of problems and The modeling and simulation of a set of biologi- 3.23 Medical Robotics...... 108 devise an equally diverse set of tools to solve them. cal parts is the domain of systems biology. What 4. DRIVERS AND DISRUPTORS...... 112 As a result, a number of distinct subdisciplines have constitutes an appropriate set for study can range 5. TECHNOLOGY COVERAGE IN come to define the field, coarsely demarcated by from a small subsystem of a biological organism, IEEE XPLORE AND BY the scale and type of phenomena they address. such as a single signaling pathway [5], to an en- IEEE SOCIETIES...... 115 tire biological cell with its complete metabolic 6. IEEE COMPUTER SOCIETY Genomic Bioinformatics and transcriptional networks [6], [7]. A plethora IN 2022...... 120 The acquisition and analysis of genomic data of modeling and simulation techniques are typi- 7. SUMMARY AND NEXT STEPS.....124 cally employed, depending on the complexity of 8. AUTHORS...... 126 APPENDIX I. 23 Technologies Coverage 102 in IEEE Publications...... 134 CONTENTS 1. INTRODUCTION...... 5 2. SEAMLESS INTELLIGENCE the underlying phenomena and the availability of has also created an unprecedented demand for SCENARIO...... 8 experimental data. new computing tools and infrastructure capable of 3. 23 TECHNOLOGIES IN 2022...... 14 analyzing enormous amounts of data. The trajecto- 3.1 Security Cross-Cutting Issues...... 14 Phylogenetics and Evolutionary Modeling ry of genomics is a classic example of a disruptive 3.2 The Open Intellectual technology, particularly on the computational side. Property Movement...... 17 The phenomena of the three aforementioned fields, To underscore the point, the cost of computation 3.3 Sustainability...... 20 the submolecular, molecular, and supramolecular, 3.4 Massively Online Open Courses...... 25 can all be studied in light of evolution. Evolutionary in the overall sequencing pipeline has historically 3.5 Quantum Computing ...... 28 genomics concerns the use and comparison of been fractional and inconsequential. As of 2010, 3.6 Device and Nanotechnology...... 31 multiple genomes to infer functional regions that the costliest aspect of the sequencing pipeline is 3.7 3D Integrated Circuits...... 33 are more likely to be conserved over evolution- the computational analysis required to turn raw CONTENTS3.8 Universal Memory...... 38 ary timescales. The use of evolutionary analysis data into completed genomes [12]. This presents 3.9 Multicore...... 43 of structures similarly helps identify functional a tremendous challenge to bioinformaticists and 3.10 Photonics...... 47 computer scientists to develop new algorithms and 3.11 Networking and Interconnectivity... 52 hotspots on biomolecules and informs the predic- computational infrastructures capable of keeping 3.12 Software-Defined Networks...... 55 tion of protein structure [8]. Finally, the analysis of 3.13 High-Performance Computing biological pathway evolution elucidates how the up with the unrelenting growth in genomic data (HPC) ...... 63 rewiring of cellular circuitry leads to new behaviors. predicted for the next several years. 3.14 Cloud Computing...... 68 3.15 The Internet of Things...... 74 3.22.3 Challenges 3.16 Natural User Interfaces...... 77 3.22.2 Current State of the Field 3.17 3D Printing ...... 81 While bioinformatics and computational biology The explosive growth in the availability of genomic 3.18 Big Data and Analytics ...... 84 constitute a broad field, genomic bioinformatics data, in particular when compared to other bio- 3.19 Machine Learning and currently occupies an oversized role within the informatics fields that have not benefited from Intelligent Systems ...... 89 similar data growth, has resulted in a high fraction 3.20 Computer Vision and Pattern field. This has been driven by significant chang- Recognition...... 92 es in both supply and demand over the past few of the bioinformatics effort being focused on solv- 3.21 Life Sciences...... 96 years. On the supply side, progress in sequencing ing sequencing problems. The overarching focus of 3.22 Computational Biology and technology resulted in explosive growth in the this effort has been the acquisition and assembly of Bioinformatics...... 102 availability of genomic sequences, with the rate of genomic data, but not necessarily its interpretation, 3.23 Medical Robotics...... 108 increase outpacing Moore’s law for over a decade as captured by the classic Cell article titled, “Se- 4. DRIVERS AND DISRUPTORS...... 112 now [9], [10]. In 2000, the first human genome quence First, Ask Questions Later” [13]. 5. TECHNOLOGY COVERAGE IN draft was completed at a cost of $3 billion after While this approach was appropriate during the IEEE XPLORE AND BY a 10-year effort. Today, an entire human genome IEEE SOCIETIES...... 115 initial stages of the genomic revolution, our abil- can be sequenced in less than a week and for less ity to analyze genomic data now lags our ability 6. IEEE COMPUTER SOCIETY IN 2022...... 120 than $10,000 [11]. This abundance of sequence to acquire it. One area where this is clear is ge- information, while a great scientific opportunity, nome-wide association studies, or GWAS. In such 7. SUMMARY AND NEXT STEPS.....124 8. AUTHORS...... 126 APPENDIX I. 23 Technologies Coverage 103 in IEEE Publications...... 134 CONTENTS 1. INTRODUCTION...... 5 2. SEAMLESS INTELLIGENCE studies, a large number of patient genomes are and if the exponential trajectory is maintained, SCENARIO...... 8 sequenced, and individual genomic loci are tested a 100- to 300-fold increase in the number of se- 3. 23 TECHNOLOGIES IN 2022...... 14 for statistical correlations with diseases. Despite quenced genomes by the end of the decade is 3.1 Security Cross-Cutting Issues...... 14 the initial high expectations for such studies, the possible. Such increases will provide a qualitative 3.2 The Open Intellectual current consensus is that most GWAS studies have improvement in available statistical power. Property Movement...... 17 been unsuccessful, because the typical strength 3.3 Sustainability...... 20 Improved statistical methodologies. Statistical of most disease correlations found has been very 3.4 Massively Online Open Courses...... 25 inference methods designed specifically to tackle 3.5 Quantum Computing ...... 28 weak [14]. So serious is the problem that it has genomic bioinformatics will become increasingly 3.6 Device and Nanotechnology...... 31 acquired its own name, “missing heritability,” which more common and will exploit the unique structure 3.7 3D Integrated Circuits...... 33 refers to the many diseases that are known to be of genomic data to infer subtle correlations, partic- CONTENTS3.8 Universal Memory...... 38 heritable but whose precise genetic causes have ularly ones in which a disease is dependent on the 3.9 Multicore...... 43 escaped elucidation [15]. 3.10 Photonics...... 47 state of multiple mutations. The causes of the so-called missing heritability are 3.11 Networking and Interconnectivity... 52 Convergence of genomic, structural, and sys- 3.12 Software-Defined Networks...... 55 myriad, including lack of sufficient data to provide tems approaches. Perhaps most importantly, the 3.13 High-Performance Computing the statistical power necessary to find very weak currently separate fields of genomic, structural, (HPC) ...... 63 correlations. But equally important are the statis- 3.14 Cloud Computing...... 68 and systems bioinformatics will converge. The tical and computational techniques used to mine 3.15 The Internet of Things...... 74 underlying driving force behind this shift is the genomic data, which were conceived in an era 3.16 Natural User Interfaces...... 77 complex mapping function between genotypes and 3.17 3D Printing ...... 81 when hundreds, instead of trillions, of data points phenotypes. Even with improvements in statistical 3.18 Big Data and Analytics ...... 84 were the norm. Furthermore, such methods typical- methodologies and increases in data sizes, if every 3.19 Machine Learning and ly assume a simple, even linear, mapping between Intelligent Systems ...... 89 human genome were sequenced, the scientific inputs (genomes) and outputs (phenotypes), when community would obtain around 1010 genomes. In 3.20 Computer Vision and Pattern in reality the functions mapping human genomes Recognition...... 92 contrast, the mutational landscape of the human 3.21 Life Sciences...... 96 to disease phenotypes are likely to be extremely genome is around 43,000,000,000 in size. Brute-force 3.22 Computational Biology and complex. statistics and data acquisition will be insufficient Bioinformatics...... 102 to decode the human genotype-phenotype func- 3.23 Medical Robotics...... 108 3.22.4 Where We Think It Will Go tion. Instead, the interpretation of genomic data 4. DRIVERS AND DISRUPTORS...... 112 The coming decade will see a shift in focus from will need to proceed in a stepwise fashion, with 5. TECHNOLOGY COVERAGE IN genome acquisition to genome interpretation. the initial focus on understanding the molecular IEEE XPLORE AND BY IEEE SOCIETIES...... 115 This will likely be precipitated by three important consequences of genomic changes. Doing so will developments. require an understanding of how sequence deter- 6. IEEE COMPUTER SOCIETY mines structure, elevating structural bioinformatics IN 2022...... 120 Qualitative increase in data quantity. Advances to a central role in a disruptive manner. The types 7. SUMMARY AND NEXT STEPS.....124 in sequencing technology continue to be made, 8. AUTHORS...... 126 APPENDIX I. 23 Technologies Coverage 104 in IEEE Publications...... 134 CONTENTS 1. INTRODUCTION...... 5 2. SEAMLESS INTELLIGENCE of analyses done within structural bioinformatics order, they will push the computational and data SCENARIO...... 8 will be different from today’s, as the emphasis storage requirements far beyond today’s limits, 3. 23 TECHNOLOGIES IN 2022...... 14 shifts from coarse-grained prediction of de novo potentially by several orders of magnitude, to the 3.1 Security Cross-Cutting Issues...... 14 structures to the precise prediction of mutational point where computation, instead of experiment, 3.2 The Open Intellectual effects on structure. The end result of this shift could become the major bottleneck. More impor- Property Movement...... 17 will be the convergence of genomic and structural tantly, these shifts will also require new types of 3.3 Sustainability...... 20 3.4 Massively Online Open Courses...... 25 bioinformatics. computation, which in the long term may prove to 3.5 Quantum Computing ...... 28 As the ability to interpret genomic data molecularly be a more substantial disruption. 3.6 Device and Nanotechnology...... 31 improves, the next step will be to interpret genom- On the data analysis side, machine learning meth- 3.7 3D Integrated Circuits...... 33 ic data in terms of systems-level phenotypes, at ods, including deep architectures that have recent- CONTENTS3.8 Universal Memory...... 38 3.9 Multicore...... 43 least on the pathway and cellular level. To do so ly seen a resurgence [18], will play an increasingly 3.10 Photonics...... 47 will require that genotypes are first mapped onto important role. Such methods have shown great 3.11 Networking and Interconnectivity... 52 structural phenotypes, which are then mapped potential for scalability when run on GPUs [19], 3.12 Software-Defined Networks...... 55 onto systems phenotype, in a bottom-up approach. which will likely further cement the role of GPUs in 3.13 High-Performance Computing In a similar vein to the first shift, understanding the bioinformatics. On the structural simulation side, (HPC) ...... 63 effects of structural changes on system behavior long time scale molecular dynamics simulations 3.14 Cloud Computing...... 68 will necessitate a move away from the study of will likely play an increasingly important role, and 3.15 The Internet of Things...... 74 3.16 Natural User Interfaces...... 77 individual biomolecules to the study of complexes specialized hardware, such as the Anton comput- 3.17 3D Printing ...... 81 of molecules and their interactions. This is currently er [20], have shown exceptional effectiveness at 3.18 Big Data and Analytics ...... 84 the domain of systems biology, but it is done in a tackling such problems [21]. 3.19 Machine Learning and top-down fashion in which high-level experimental Intelligent Systems ...... 89 The broader impact of these changes will first data is used to fit observed systems-level phenom- be felt in the basic life sciences, where the con- 3.20 Computer Vision and Pattern ena, instead of a bottom-up approach in which Recognition...... 92 vergence of disparate bioinformatics fields will 3.21 Life Sciences...... 96 known molecular interactions are simulated to help elucidate the mechanistic basis of biological 3.22 Computational Biology and obtain, in an emergent manner, the observed sys- pathways. In the longer term, this newfound un- Bioinformatics...... 102 tems-level behavior. Achieving this will result in the derstanding will be translated into new treatment 3.23 Medical Robotics...... 108 convergence of structural and systems bioinfor- strategies and therapeutic targets for human 4. DRIVERS AND DISRUPTORS...... 112 matics, where systems-scale structural simulations diseases. Two examples help illustrate the potential 5. TECHNOLOGY COVERAGE IN play a central role. Such a shift is already underway, impact of these shifts. IEEE XPLORE AND BY although on a limited scale [16], [17]. IEEE SOCIETIES...... 115 Cancer Modeling 6. IEEE COMPUTER SOCIETY 3.22.5 Potential Disruptions IN 2022...... 120 Many types of cancers are caused by somat- All the above shifts will prove disruptive. To a first 7. SUMMARY AND NEXT STEPS.....124 ic mutations, i.e., mutations acquired during the 8. AUTHORS...... 126 APPENDIX I. 23 Technologies Coverage 105 in IEEE Publications...... 134 CONTENTS 1. INTRODUCTION...... 5 2. SEAMLESS INTELLIGENCE lifetime of an individual, which disrupt important therapeutic approaches will take on an increas- SCENARIO...... 8 signaling pathways in human cells. Currently, many ingly polypharmacological bent, meaning they will 3. 23 TECHNOLOGIES IN 2022...... 14 large-scale projects are underway to identify the by design target multiple molecules because the 3.1 Security Cross-Cutting Issues...... 14 specific mutations responsible for different types of disease state is induced by multiple molecules. 3.2 The Open Intellectual cancers [22], [23]. These projects rely on acquiring Furthermore, even when a single molecule is tar- Property Movement...... 17 a large number of tumor genomes and searching geted, understanding the polypharmacology of a 3.3 Sustainability...... 20 3.4 Massively Online Open Courses...... 25 for overrepresented mutations that may be indic- drug is important, as some lack of specificity may 3.5 Quantum Computing ...... 28 ative of a causal role. Unfortunately, as described be more problematic than another. The integration 3.6 Device and Nanotechnology...... 31 earlier, finding such causal links is difficult, as of structural and systems approaches will play a 3.7 3D Integrated Circuits...... 33 many cancers are affected through a large number crucial role in making designed polypharmacology CONTENTS3.8 Universal Memory...... 38 of mutations acting in concert. Furthermore, the a reality. By enabling the analysis and simulation 3.9 Multicore...... 43 disruptions caused by these mutations often affect of a drug’s molecular interaction with all proteins 3.10 Photonics...... 47 multiple proteins in a signaling pathway, such that in a given pathway, its systems-level behavior can 3.11 Networking and Interconnectivity... 52 the integrative effect cannot be ascertained with- be predicted, and possibly designed. In addition, 3.12 Software-Defined Networks...... 55 3.13 High-Performance Computing out a systems-level model of how the signaling the information gained from a more sophisticated (HPC) ...... 63 pathway functions. The coming advances in struc- understanding of the basic science of disease will 3.14 Cloud Computing...... 68 tural and systems bioinformatics will make it pos- provide additional targets for drugs to act on. 3.15 The Internet of Things...... 74 sible to translate genomic data into molecular and 3.16 Natural User Interfaces...... 77 systems phenotypes, and to establish a causal link 3.22.6 Summary 3.17 3D Printing ...... 81 between genotype and disease that may ultimately 3.18 Big Data and Analytics ...... 84 The past decade has been an exciting time in bio- be disrupted therapeutically. 3.19 Machine Learning and informatics and the life sciences broadly, as fun- Intelligent Systems ...... 89 damental breakthroughs in technology have made 3.20 Computer Vision and Pattern Polypharmacology Recognition...... 92 it possible to amass unparalleled amounts of data. 3.21 Life Sciences...... 96 The development of therapeutic drugs is currently The core challenges of this and upcoming decades 3.22 Computational Biology and centered on finding a “target,” typically a protein will be the translation of such data into actionable Bioinformatics...... 102 believed to play a causal role in a disease and knowledge, one that can improve human health 3.23 Medical Robotics...... 108 whose activity is to be suppressed or enhanced. and shed light on the principal mysteries of life. 4. DRIVERS AND DISRUPTORS...... 112 Much of the effort in medicinal chemistry is in find- Much as mathematics, particularly group theory 5. TECHNOLOGY COVERAGE IN ing drugs with a “clean” profile, i.e., ones that only and topology, played a critical role in the devel- IEEE XPLORE AND BY affect their intended target while leaving all other opment of 20th century physics, computation and IEEE SOCIETIES...... 115 proteins unperturbed. In the current era of one machine learning are playing an analogous role in 6. IEEE COMPUTER SOCIETY molecule one disease, this approach makes sense. the development of 21st century biology. And much IN 2022...... 120 However, as our understanding of the complex as physics proved to be a constant source of dis- 7. SUMMARY AND NEXT STEPS.....124 interactions underlying disease states improves, ruptive developments in the past century, it is likely 8. AUTHORS...... 126 APPENDIX I. 23 Technologies Coverage 106 in IEEE Publications...... 134 CONTENTS 1. INTRODUCTION...... 5 2. SEAMLESS INTELLIGENCE that the intersection of computation and biology Sequencing Technology,” Nature, vol. 470, no. 7333, SCENARIO...... 8 will play a similarly disruptive role in this and up- 2011, pp. 198–203. 3. 23 TECHNOLOGIES IN 2022...... 14 coming decades. [11] K. Wetterstrand, “DNA Sequencing Costs: 3.1 Security Cross-Cutting Issues...... 14 Data from the NHGRI Genome Sequencing Program 3.2 The Open Intellectual (GSP),” DNA Sequencing Costs: Data from the NHGRI Property Movement...... 17 3.22.7 References Genome Sequencing Program (GSP); www.genome.gov/ 3.3 Sustainability...... 20 [1] D.W. Mount, Bioinformatics: Sequence and sequencingcosts. 3.4 Massively Online Open Courses...... 25 Genome Analysis, Cold Spring Harbor Laboratory Press, 3.5 Quantum Computing ...... 28 2004. [12] L.D. Stein, “The Case for Cloud Computing in 3.6 Device and Nanotechnology...... 31 [2] K.A. Dill and J.L. MacCallum, “The Protein-Fold- Genome Informatics,” Genome Biol., vol. 11, no. 5, 2010, 3.7 3D Integrated Circuits...... 33 ing Problem, 50 Years On,” Science, vol. 338, no. 6110, p. 207. CONTENTS3.8 Universal Memory...... 38 2012, pp. 1042–1046. [13] “Sequence First. Ask Questions Later,” Cell, vol. 3.9 Multicore...... 43 111, no. 1, 2002, pp. 13–16. 3.10 Photonics...... 47 [3] M.G. Seetin and D.H. Mathews, “RNA Structure 3.11 Networking and Interconnectivity... 52 Prediction: An Overview of Methods,” Methods Mol. Biol. [14] D.B. Goldstein, “Common Genetic Variation and 3.12 Software-Defined Networks...... 55 Clifton Nj, vol. 905, 2012, pp. 99–122. Human Traits,” N. Engl. J. Med., vol. 360, no. 17, 2009, pp. 3.13 High-Performance Computing 1696–1698. (HPC) ...... 63 [4] J. Gu and P.E. Bourne, Structural Bioinformatics, 3.14 Cloud Computing...... 68 Wiley-Blackwell, 2009. [15] T.A. Manolio et al., “Finding the Missing Herita- 3.15 The Internet of Things...... 74 [5] K. Sachs et al., “Causal Protein-Signaling Net- bility of Complex Diseases,” Nature, vol. 461, no. 7265, 3.16 Natural User Interfaces...... 77 works Derived from Multiparameter Single-Cell Data,” 2009, pp. 747–753. 3.17 3D Printing ...... 81 Science, vol. 308, no. 5721, 2005, pp. 523–529. [16] R.L. Chang et al., “Structural Systems Biology 3.18 Big Data and Analytics ...... 84 [6] J.R. Karr et al., “A Whole-Cell Computational Evaluation of Metabolic Thermotolerance in Escherichia 3.19 Machine Learning and coli,” Science, vol. 340, no. 6137, 2013, pp. 1220–1223. Intelligent Systems ...... 89 Model Predicts Phenotype from Genotype,” Cell, vol. 3.20 Computer Vision and Pattern 150, no. 2, 2012, pp. 389–401. [17] M. Duran-Frigola, R. Mosca, and P. Aloy, “Struc- Recognition...... 92 tural Systems Pharmacology: The Role of 3D Structures 3.21 Life Sciences...... 96 [7] J. Gunawardena, “Silicon Dreams of Cells into Symbols,” Nat. Biotechnol., vol. 30, no. 9, 2012, pp. in Next-Generation Drug Development,” Chem. Biol., vol. 3.22 Computational Biology and 20, no. 5, 2013, pp. 674–684. Bioinformatics...... 102 838–840. 3.23 Medical Robotics...... 108 [8] D.S. Marks et al., “Protein 3D Structure Comput- [18] S. Bengio et al., “Guest Editors’ Introduction: 4. DRIVERS AND DISRUPTORS...... 112 ed from Evolutionary Sequence Variation,” Plos One, vol. Special Section on Learning Deep Architectures,” IEEE Trans. Pattern Anal. Mach. Intell., vol. 35, no. 8, 2013, pp. 5. TECHNOLOGY COVERAGE IN 6, no. 12, 2011, p. e28766. 1795–1797. IEEE XPLORE AND BY [9] N.D. DeWitt, M.P. Yaffe, and A. Trounson, “Build- IEEE SOCIETIES...... 115 ing Stem-Cell Genomics in California and Beyond,” Nat. [19] A. Coates et al., “Deep Learning with COTS HPC 6. IEEE COMPUTER SOCIETY Biotechnol., vol. 30, no. 1, 2012, pp. 20–25. Systems,” Proc. 30th Int’l Conf. Machine Learning, 2013, IN 2022...... 120 pp. 1337–1345. [10] E.R. Mardis, “A Decade’s Perspective on DNA 7. SUMMARY AND NEXT STEPS.....124 [20] D.E. Shaw et al., “Anton, a Special-purpose 8. AUTHORS...... 126 APPENDIX I. 23 Technologies Coverage 107 in IEEE Publications...... 134 CONTENTS 1. INTRODUCTION...... 5 2. SEAMLESS INTELLIGENCE Machine for Molecular Dynamics Simulation,” Comm. SCENARIO...... 8 ACM, vol. 51, no. 7, 2008, pp. 91–97. 3. 23 TECHNOLOGIES IN 2022...... 14 [21] R.O. Dror et al., “Structural Basis for Modulation 3.1 Security Cross-Cutting Issues...... 14 of a G-Protein-Coupled Receptor by Allosteric Drugs,” 3.2 The Open Intellectual Nature, 2013. Property Movement...... 17 3.3 Sustainability...... 20 [22] Cancer Genome Atlas Research Network, 3.4 Massively Online Open Courses...... 25 “Comprehensive Genomic Characterization Defines Hu- 3.5 Quantum Computing ...... 28 man Glioblastoma Genes and Core Pathways,” Nature, 3.6 Device and Nanotechnology...... 31 vol. 455, no. 7216, 2008, pp. 1061–1068. 3.7 3D Integrated Circuits...... 33 3.23 MedicalRobotics [23] T.J. Hudson et al., “International Network of CONTENTS3.8 Universal Memory...... 38 Cancer Genome Projects,” Nature, vol. 464, no. 7291, 3.9 Multicore...... 43 3.23.1 Introduction 3.10 Photonics...... 47 2010, pp. 993–998. Advances in computer science, image processing, 3.11 Networking and Interconnectivity... 52 3.12 Software-Defined Networks...... 55 nanotechnology, materials science, and many other 3.13 High-Performance Computing fields have led to remarkable new applications for (HPC) ...... 63 robotics in medicine. These applications include 3.14 Cloud Computing...... 68 such diverse uses as autonomous delivery or 3.15 The Internet of Things...... 74 supplies in hospitals, telemedicine, robotic surgery, 3.16 Natural User Interfaces...... 77 and rehabilitative support. These and future ap- 3.17 3D Printing ...... 81 3.18 Big Data and Analytics ...... 84 plications suggest transformational advances in 3.19 Machine Learning and lower mortality rates, increased access to modern Intelligent Systems ...... 89 medical care, and vastly improved quality of life for 3.20 Computer Vision and Pattern millions of people worldwide. Cost, training, com- Recognition...... 92 plexity, and ethical concerns present significant 3.21 Life Sciences...... 96 issues to be overcome. 3.22 Computational Biology and Bioinformatics...... 102 3.23 Medical Robotics...... 108 3.23.2 The State of the Art 4. DRIVERS AND DISRUPTORS...... 112 A wide range of applications of robots in medicine 5. TECHNOLOGY COVERAGE IN can provide a variety of transformational benefits. IEEE XPLORE AND BY The following entries present just a small subset of IEEE SOCIETIES...... 115 current and future applications. 6. IEEE COMPUTER SOCIETY IN 2022...... 120 Advanced prostheses replace the functionality of 7. SUMMARY AND NEXT STEPS.....124 missing limbs and provide sophisticated and in- 8. AUTHORS...... 126 tuitive controls for the wearer. For example, Dean APPENDIX I. 23 Technologies Coverage 108 in IEEE Publications...... 134 CONTENTS 1. INTRODUCTION...... 5 2. SEAMLESS INTELLIGENCE Kamen’s robotic arm directly translates signals These systems are becoming quite common. The SCENARIO...... 8 from the user’s nerves and is so versatile that it can da Vinci Surgical System, for example, one of the 3. 23 TECHNOLOGIES IN 2022...... 14 handle very small and delicate objects and large most well-known computer-aided surgical systems, 3.1 Security Cross-Cutting Issues...... 14 heavy objects without manual adjustment, as in is installed in more than 2,700 operating environ- 3.2 The Open Intellectual many conventional prosthetic arms [Guizzo]. ments around the world. Property Movement...... 17 3.3 Sustainability...... 20 Hospital delivery robots are essentially autonomous Telemedicine uses remotely controlled robots to 3.4 Massively Online Open Courses...... 25 vehicles that transport such items as supplies, enable doctors to interact directly with patients 3.5 Quantum Computing ...... 28 instruments, specimens, and waste. The robots from anywhere in the world, so long as the nec- 3.6 Device and Nanotechnology...... 31 typically use route planning for navigation, laser essary communications bandwidth is available 3.7 3D Integrated Circuits...... 33 ranging for collision avoidance, and signaling to [Ackerman]. Telemedicine can be used to deliver CONTENTS3.8 Universal Memory...... 38 3.9 Multicore...... 43 open doors or use elevators [McNickle]. preventative healthcare and basic or specialized 3.10 Photonics...... 47 Microbots refer to any type of free-ranging “small” medical support for people in remote villages or 3.11 Networking and Interconnectivity... 52 robot that can execute specific tasks within the in the wake of disasters, combat troops in theater, 3.12 Software-Defined Networks...... 55 body. Sample applications include retinal surgery, travelers on cruiseships or aircraft, and even to 3.13 High-Performance Computing arterial plaque removal, or diagnosis of a diseases explorers and scientists in extreme environments (HPC) ...... 63 including space. Advances in virtual reality will lead 3.14 Cloud Computing...... 68 or injuries [McNickle]. to truly immersive telepresence experiences for 3.15 The Internet of Things...... 74 Rehabilitation exoskeletons are wearable robots are 3.16 Natural User Interfaces...... 77 surgeons. 3.17 3D Printing ...... 81 used to enhance the capabilities of weakened or 3.18 Big Data and Analytics ...... 84 paralyzed limbs. For example, Toyota’s Walk Train- 3.23.3 Challenges ing Assist robot helps a patient learn to walk again. 3.19 Machine Learning and High cost is a serious threat to widespread adop- Intelligent Systems ...... 89 A related application involves specialized assistive tion of any medical robot technology. The widely 3.20 Computer Vision and Pattern devices that provide a paralyzed user more inde- Recognition...... 92 deployed da Vinci system, for example, costs more pendence, such as a small robotic arm that can be 3.21 Life Sciences...... 96 than US$2 million and has higher maintenance controlled with a joystick or other control mecha- 3.22 Computational Biology and costs than conventional set ups [Marcus et al]. Bioinformatics...... 102 nism so that a patient can feed himself [McNickle]. These financial burdens are likely prohibitive in 3.23 Medical Robotics...... 108 Robotic surgery involves robotic or computer-aid- most developing countries. 4. DRIVERS AND DISRUPTORS...... 112 ed surgical systems that offer the advantages of The potential risks of hardware or software failures 5. TECHNOLOGY COVERAGE IN reduced invasiveness, lower hand vibration, and in the systems could lead to significant injury or IEEE XPLORE AND BY highly accurate positioning, all of which lead to re- IEEE SOCIETIES...... 115 death. Moreover, the “clinical efficacy and safety of markably reduced physical burdens on the patient, 6. IEEE COMPUTER SOCIETY robot-assisted surgery over [conventional] proce- a general enhancement to quality of life, and de- IN 2022...... 120 dures has not yet been ascertained” [Marcus et al]. creased irradiation exposure for medical personnel. 7. SUMMARY AND NEXT STEPS.....124 While advances in systems engineering and related 8. AUTHORS...... 126 APPENDIX I. 23 Technologies Coverage 109 in IEEE Publications...... 134 CONTENTS 1. INTRODUCTION...... 5 2. SEAMLESS INTELLIGENCE disciplines have produced many techniques for luxuries, or only in exclusive settings without signif- SCENARIO...... 8 producing high-reliability safety-critical systems, no icant government support. 3. 23 TECHNOLOGIES IN 2022...... 14 one can guarantee total system safety, particularly 3.1 Security Cross-Cutting Issues...... 14 in such highly complex systems. A related ethical 3.23.5 Potential Disruptions 3.2 The Open Intellectual and legal issue is who is responsible if something Property Movement...... 17 If widely available, medical robotics could dramat- goes wrong – the technician, nurse, physician, 3.3 Sustainability...... 20 ically decrease mortality rates and improve the engineers, or systems integrators? Legal entangle- 3.4 Massively Online Open Courses...... 25 quality of life for millions, worldwide. In particular 3.5 Quantum Computing ...... 28 ments may thwart innovation and deployment of many of these technologies can extend the reach 3.6 Device and Nanotechnology...... 31 new technologies. of advanced medical care into remote regions and 3.7 3D Integrated Circuits...... 33 All of these systems also require significant train- radically change healthcare delivery systems in CONTENTS3.8 Universal Memory...... 38 3.9 Multicore...... 43 ing for physicians, nurses, and technicians, and, for developed and underdeveloped countries. 3.10 Photonics...... 47 rehabilitative or supportive technologies, for the In the long term, these technologies could reduce 3.11 Networking and Interconnectivity... 52 patient as well. Medical professionals and patients healthcare costs because of the leveraging effects, 3.12 Software-Defined Networks...... 55 may have significant concerns about the use of improved preventive treatments, and increased 3.13 High-Performance Computing robot-assisted technologies. (HPC) ...... 63 health awareness. 3.14 Cloud Computing...... 68 Other issues include compatibility in high-radiation However, the promise of medical robotics could 3.15 The Internet of Things...... 74 environments, such as in the presence of magnetic displace resources from more imminent needs to 3.16 Natural User Interfaces...... 77 resonance imaging devices, and the need for signif- these long-range, speculative technologies. 3.17 3D Printing ...... 81 icant dedicated space in hospitals for most of these 3.18 Big Data and Analytics ...... 84 robot technologies. 3.19 Machine Learning and 3.23.6 References Intelligent Systems ...... 89 For telemedicine, particularly in rural areas, under- E. Ackerman, “iRobot and InTouch Health An- 3.20 Computer Vision and Pattern developed countries, and extreme environments, Recognition...... 92 nounce RP-VITA Telemedicine Robot,” IEEE Spec- higher speed, secure, and stable connectivity are 3.21 Life Sciences...... 96 trum Online, July 2012; http://spectrum.ieee.org/ 3.22 Computational Biology and problematic. automaton/robotics/medical-robots/irobot-and-in- Bioinformatics...... 102 touch-health-announce-rpvita-telemedicine-robot 3.23 Medical Robotics...... 108 3.23.4 Where We Think It Will Go (nice images available here) 4. DRIVERS AND DISRUPTORS...... 112 While significant life-changing innovations will M. McNickle, “10 Medical Robots That Could 5. TECHNOLOGY COVERAGE IN continue at a rapid pace, adoptions will be very Change Healthcare,” Information Week Online, IEEE XPLORE AND BY slow, particularly in underdeveloped countries due IEEE SOCIETIES...... 115 December 2012; http://www.informationweek.com/ to the aforementioned shortcomings, particularly mobile/10-medical-robots-that-could-change-health- 6. IEEE COMPUTER SOCIETY cost. Government agency approvals will also slow IN 2022...... 120 care/d/d-id/1107696 deployment of promising technologies. Many of H. Marcus et al., “Surgical Robotics Through a Keyhole: 7. SUMMARY AND NEXT STEPS.....124 these technologies will exist purely as curiosities, 8. AUTHORS...... 126 APPENDIX I. 23 Technologies Coverage 110 in IEEE Publications...... 134 CONTENTS 1. INTRODUCTION...... 5 2. SEAMLESS INTELLIGENCE SCENARIO...... 8 3. 23 TECHNOLOGIES IN 2022...... 14 From Today’s Translational Barriers to Tomorrow’s “Dis- 3.1 Security Cross-Cutting Issues...... 14 appearing” Robots,” IEEE Trans. Biomedical Engineering , 3.2 The Open Intellectual vol.60, no.3, March 2013, pp.674-681. Property Movement...... 17 P. Haigron, L. Luo, and J-L. Coatrieux, “Issues in Im- 3.3 Sustainability...... 20 age-Guided Therapy [A Look at...],” IEEE Engineering 3.4 Massively Online Open Courses...... 25 in Medicine and Biology, vol. 28, no. 4, 2009, pp. 96-98. 3.5 Quantum Computing ...... 28 (nice images available here) 3.6 Device and Nanotechnology...... 31 3.7 3D Integrated Circuits...... 33 E. Guizzo, “Dean Kamen’s “Luke Arm” Prosthesis Re- CONTENTS3.8 Universal Memory...... 38 ceives FDA Approval,” IEEE Spectrum Online, May 13, 3.9 Multicore...... 43 2014; http://spectrum.ieee.org/automaton/biomedical/ 3.10 Photonics...... 47 bionics/dean-kamen-luke-arm-prosthesis-receives-fda- 3.11 Networking and Interconnectivity... 52 approval (nice images available here) 3.12 Software-Defined Networks...... 55 3.13 High-Performance Computing (HPC) ...... 63 3.14 Cloud Computing...... 68 3.15 The Internet of Things...... 74 3.16 Natural User Interfaces...... 77 3.17 3D Printing ...... 81 3.18 Big Data and Analytics ...... 84 3.19 Machine Learning and Intelligent Systems ...... 89 3.20 Computer Vision and Pattern Recognition...... 92 3.21 Life Sciences...... 96 3.22 Computational Biology and Bioinformatics...... 102 3.23 Medical Robotics...... 108 4. DRIVERS AND DISRUPTORS...... 112 5. TECHNOLOGY COVERAGE IN IEEE XPLORE AND BY IEEE SOCIETIES...... 115 6. IEEE COMPUTER SOCIETY IN 2022...... 120 7. SUMMARY AND NEXT STEPS.....124 8. AUTHORS...... 126 APPENDIX I. 23 Technologies Coverage 111 in IEEE Publications...... 134 CONTENTS 1. INTRODUCTION...... 5 2. SEAMLESS INTELLIGENCE SCENARIO...... 8 4. Drivers and Disruptors 3. 23 TECHNOLOGIES IN 2022...... 14 4. DRIVERS AND DISRUPTORS...... 112 5. TECHNOLOGY COVERAGE IN IEEE XPLORE AND BY To verify the premises and conclusions that the • Quickening pace of knowledge transfer (e.g., IEEE SOCIETIES...... 115 IEEE CS 2022 team made, we surveyed a few instantaneous global communication) 6.CONTENTS IEEE COMPUTER SOCIETY thousand IEEE members. Questionnaires went • Long-term availability of certain energy sources IN 2022...... 120 out after we selected the technologies and wrote • Alternative distribution chains (such as manu- 7. SUMMARY AND NEXT STEPS.....124 initial drafts. facturers selling directly to consumers) 8. AUTHORS...... 126 We posted two classes of questions, asking those • Use of technology for medical procedures • Wireless/broadband connectivity APPENDIX I. 23 Technologies Coverage who responded to rank driver and disruptor tech- in IEEE Publications...... 134 nologies. We offered the following items to be Disruptors ranked: • Crowd-sourcing/open-sourcing of hardware development Drivers • Changes in educational structure/design (e.g., • Increases in average life expectancy MOOCs) • Increasing ratio of retirees to workers • Virtual/alternative currencies (such as Bitcoin) • Public concern over control over access/ • Smartphone use as a device for payment amount of personal information • Cloud computing • Desire for sustainable energy sources • Use of robots as a source of labor • Reduction in availability of grants and philan- • Nonvolatile memory influencing big data acces- thropic resources sibility and portability • Widening economic inequality worldwide • Quantum/nondeterministic computing • Reduced job security in a global market econo- • Use of 3D printing my • Green computing Questionnaires • Climate change • New user interfaces (e.g., Siri, Kinect instead of asked respondents • Global terrorism traditional keyboards) to rank driver • Use of big data and analytics • We received the following answers, represent- • Reduction in cost of data collection and reten- and disruptor ed in two figures. tion (for use in analytics) technologies.

112 CONTENTS 1. INTRODUCTION...... 5 Major Driver 2. SEAMLESS INTELLIGENCE SCENARIO...... 8 50.0% 3. 23 TECHNOLOGIES IN 2022...... 14 45.0% 40.0% 4. DRIVERS AND DISRUPTORS...... 112 35.0% 5. TECHNOLOGY COVERAGE IN 30.0% IEEE XPLORE AND BY 25.0% IEEE SOCIETIES...... 115 20.0% 6.CONTENTS IEEE COMPUTER SOCIETY 15.0% IN 2022...... 120 10.0% 7. SUMMARY AND NEXT STEPS.....124 5.0% 0.0% Major Driver 8. AUTHORS...... 126 APPENDIX I. 23 Technologies Coverage in IEEE Publications...... 134

Figure 15. Comparison of major drivers. Major Disruptor

Results aligned well with30.0% independent report long-term availability of energy resources; and Questionnaires findings. For example,25.0% the highest ranked drivers quickening pace of knowledge transfer. All of these were the use of technology20.0% for medical procedures, drivers are discussed in the report. asked respondents followed by wireless/broadband15.0% connectivity and Similarly, for major disruptors, use of robots as la- to rank driver desire for sustainable energy sources. Also high- 10.0% bor and 3D printing led the votes, followed by cloud and disruptor ly ranked were the use of big data and analytics; 5.0% computing, MOOCs, and new user interfaces. technologies. 0.0% 113 Major Driver 50.0% 45.0% 40.0% 35.0% 30.0% 25.0% 20.0% 15.0% 10.0% 5.0% 0.0% Major Driver

CONTENTS 1. INTRODUCTION...... 5 Major Disruptor 2. SEAMLESS INTELLIGENCE 30.0% SCENARIO...... 8 25.0% 3. 23 TECHNOLOGIES IN 2022...... 14 20.0% 4. DRIVERS AND DISRUPTORS...... 112 15.0% 5. TECHNOLOGY COVERAGE IN 10.0% IEEE XPLORE AND BY 5.0% IEEE SOCIETIES...... 115 0.0% 6.CONTENTS IEEE COMPUTER SOCIETY IN 2022...... 120 7. SUMMARY AND NEXT STEPS.....124 8. AUTHORS...... 126 APPENDIX I. 23 Technologies Coverage in IEEE Publications...... 134

Figure 16. Comparison of major disruptors.

Questionnaires asked respondents to rank driver and disruptor technologies.

114 CONTENTS 1. INTRODUCTION...... 5 2. SEAMLESS INTELLIGENCE SCENARIO...... 8 5. Technology Coverage 3. 23 TECHNOLOGIES IN 2022...... 14 4. DRIVERS AND DISRUPTORS...... 112 in IEEE Xplore and 5. TECHNOLOGY COVERAGE IN IEEE XPLORE AND BY IEEE SOCIETIES...... 115 CONTENTS5.1 Introduction...... 115 by IEEE Societies 5.2 Comparison...... 116 5.3 Summary of Quantitative Analysis Findings...... 119 6. IEEE COMPUTER SOCIETY 5.1 Introduction together includes sharing investments, branding, IN 2022...... 120 These 23 technologies were analyzed for how IEEE publicity, leadership, and stewardship for future 7. SUMMARY AND NEXT STEPS.....124 Computer Society volunteer leaders can recruit investments. The IEEE Computer Society continues 8. AUTHORS...... 126 existing experts and organizers and then invest in to attracted many authors to our periodicals, conferences, and standards, thus the Society will use this APPENDIX I. 23 Technologies Coverage these technologies for growth. As technology con- in IEEE Publications...... 134 vergence is inevitable, the Society will also partner base as an important stake to grow these 23 future with other major technical Societies to grow these technologies. As we get organized around these areas. As IEEE encourages thought leadership in technologies to invest in them, these authors will be key technologies, funding might be available for attracted to publish again. these investments. The motivation for this analysis included the need IEEE periodicals, conferences, and standards are key to identify potential Society partners to best col- knowledge creation centers. IEEE-published articles laborate and to better understand overlaps for have high quality and show the current state of the future consolidation. This was reinforced by the art. These articles are widely available to support Communications Society’s 2012 ComSoc 2020 report, which stated that its current technology Our future additional research, attract funding, and attract other authors to build out the technical roadmap to drive trends were going to require partnering with other technologies require the technology evolution. These knowledge centers IEEE Societies, such as the Computer Society. partnering with are franchised by Societies, so Societies need to work Within our society, Special Technical Commu- other Societies. together on emerging technology trends. Working nities (STCs) are formed to address emerging

115 CONTENTS technologies such as these 23. This analysis helps we tested them in Xplore’s advanced search for STC leaders reach out to other Societies’ techni- IEEE periodicals articles’ metadata between 2000 1. INTRODUCTION...... 5 cal communities for collaboration on conference and 2013 to avoid popular terms that included ar- 2. SEAMLESS INTELLIGENCE tracks, special issues in our periodicals, or draft ticles outside the technical areas. These keywords SCENARIO...... 8 standards to focus on an emerging technology. were arranged into a Boolean search expression to 3. 23 TECHNOLOGIES IN 2022...... 14 select the most relevant articles within the technol- 4. DRIVERS AND DISRUPTORS...... 112 5.2 Comparison ogy area. A large sample (up to 2,000) of articles’ metadata was downloaded to assign the Society 5. TECHNOLOGY COVERAGE IN Each of these 23 technologies was analyzed for its sponsoring the periodical. We tabularized the Soci- IEEE XPLORE AND BY coverage of IEEE periodical articles published from IEEE SOCIETIES...... 115 ety’s articles into a pie chart to highlight the top 10 2000 to 2013. These periodicals are technically and CONTENTS5.1 Introduction...... 115 Societies plus “other” collected Societies with low- financially sponsored by Societies, whose stew- er article counts. Below are the four pie charts that 5.2 Comparison...... 116 ardship is for growth in quality reputation, author our survey identified as “drivers” (See Footnote12 5.3 Summary of Quantitative prestige, author submissions, relevance, and sub- for the description of acronyms for societies): Analysis Findings...... 119 scribers and readers. Mapping the numbers of pe- 6. IEEE COMPUTER SOCIETY riodical articles to the sponsoring Societies reveals The appendix provides additional and enlarged IN 2022...... 120 each Society stake or coverage in our 23 technol- chart views. Note that two of the four show the 7. SUMMARY AND NEXT STEPS.....124 ogies. This coverage is shown in each technology Computer Society with the largest coverage: the 8. AUTHORS...... 126 pie chart, where portions show the major IEEE first indicates that many potential partners could APPENDIX I. 23 Technologies Coverage technical Societies contribution according to their in IEEE Publications...... 134 share of related periodical articles. The final chart 12 AE=Aerospace & Electronic Systems AP=Antennas & Propagation shows the all the IEEE technical Societies for all 23 BIO=Biometrics Council CAS=Circuits & Systems CE=Consumer Electron- technologies along with a companion word cloud. ics CEDA=Council on Electronic Design Automation CI=Computational [Note that neither conferences nor standards were Intelligence COMM=Communications CPMT=Components, Packaging, included in this analysis as the true sponsoring & Manufacturing Technology CSS=Control Systems EC=Electromagnetic influence is difficult to assign to a Society.] Compatibility Ed=Education ED=Electron Devices EMB=Engineering in Medicine & Biology GRS=Geoscience & Remote Sensing IE=Industrial The objective measure of a Society’s “coverage” in Electronics IM=Instrumentation & Measurement IT=Information Theory a technical area was measured by the number of ITS=Intelligent Transportation Systems MAG=Magnetics MTT=Micro- IEEE periodical articles published. As every Society wave Theory & Techniques NANO=Nanotechnology Council NPS=Nuclear Our future (co-)sponsors periodicals, the Society was given & Plasma Sciences OCEAN=Oceanic Engineering PE=Power & Energy credit by the number of articles discovered when technologies require PHO=Photonics RA=Robotics & Automation SEN=Sensors Council searching with technical area’s keywords. These SMC=Systems, Man, & Cybernetics SP=Signal Processing SSC=Sol- partnering with keywords frame the technical area and were pro- id-State Circuits SSIT=Social Implications of Technology SUPERC=Council other Societies. vided by our contributing technology area experts; on Superconductivity TM=Technology Management Council UFFC=Ul- trasonics, Ferroelectrics, and Frequency Control VT=Vehicular Technology

116 IEEE CS 2022 , Report DRAFT APPENDIX I. 23 Technologies Coverage in IEEE Publications 1/26/2014 5:18 IEEE CS 2022 , Report DRAFT APPENDIX I. 23 Technologies Coverage in IEEE Publications 1/26/2014 5:18 PM PM

1 Security Cross-cutting Issues 18 Big Data Analytics Sponsoring Societies Sponsoring Societies CONTENTS AE CE 1% 2% VT TM 1. INTRODUCTION...... 5 BIO PE 1% 2% 2% 2% IT COMM 2. SEAMLESS INTELLIGENCE 3% IEEE 5% SCENARIO...... 8 4% 3. 23 TECHNOLOGIES IN 2022...... 14 OTHER 5% 4. DRIVERS AND DISRUPTORS...... 112 Compute 5. TECHNOLOGY COVERAGE IN PE r IEEE XPLORE AND BY 9% 43% IEEE SOCIETIES...... 115 CONTENTS5.1 Introduction...... 115

5.2 Comparison...... 116 SP 5.3 Summary of Quantitative 15% Analysis Findings...... 119 Computer 6. IEEE COMPUTER SOCIETY IEEE CS 2022 , Report DRAFT APPENDIX I. COMM 23 Technologies Coverage in IEEE Publications 1/26/2014 5:18 IEEE CS 2022 , Report DRAFT APPENDIX I. 23 Technologies 91% Coverage in IEEE Publications 1/26/2014 5:18 PM 15% PM IN 2022...... 120

AE=Aerospace & Electronic Systems AP=Antennas & Propagation BIO=Biometrics Council umer CAS=Circuits & Systems CE=Cons AE=Aerospace & Electronic Systems AP=Antennas & Propagation BIO=Biometrics Council umer CAS=Circuits & Systems CE=Cons Electronics CEDA=Council on Electronic Design Automation CI=Computational =Communications Intelligence COMM Electronics CEDA=Council on Electronic omation Design Aut CI=Computational Intelligence COMM=Communications 7. SUMMARY AND NEXT STEPS.....124 CPMT=Components, Packaging, & Manufacturing Technology CSS=Control Systems EC=Electromagnetic Compatibility Ed=Education CPMT=Components, 11 Networking Packaging, & Manufacturing & Interconnectivity Technology CSS=Control Systems EC=Electromagnetic tion Compatibility Ed=Educa 23 ED=Electron Robotics Devices EMB=Engineering in Medicine in Medicine & Biology GRS=Geoscience ndustrial & Remote Sensing IE=I Electronics ED=Electron Devices EMB=Engineering ology in Medicine & Bi GRS=Geoscience & Remote Sensing IE=Industrial Electronics IM=Instrumentation & Measurement IT=Information Theory ITS=Intelligent Transportation Systems MAG=Magnetics IM=Instrumentation & Measurement IT=Information Theory ITS=Intelligent Transportation Systems MAG=Magnetics 8. AUTHORS...... 126 MTT=Microwave Theory & Techniques NANO=Nanotechnology Council NPS=Nuclear nic & Plasma Sciences OCEAN=Ocea Engineering Sponsoring Societies Sponsoring Societies MTT=Microwave Theory & Techniques NANO=Nanotechnology l Counci NPS=Nuclear & Plasma Sciences OCEAN=Oceanic Engineering PE=Power & Energy PHO=Photonics RA=Robotics & Automation SEN=Sensors Council SMC=Systems, Man, & Cybernetics Educ PE=Power & Energy PHO=Photonics RA=Robotics & Automation SEN=Sensors Council SMC=Systems, Man, & Cybernetics SP=Signal Processing -‐ SSC=Solid State Circuits Ocean SSIT=Social Implications ITS of Technology SUPERC=Council on Superconductivity CE SP=Signal Processing -‐ SSC=Solid State Circuits SSIT=Social Implications of Technology SUPERC=Council on Superconductivity APPENDIX I. 23 Technologies Coverage TM=Technology Management Council SEN 1% UFFC=Ultrasonics, 1% 1% Ferroelectrics, and Frequency Control VT=Vehicular Technology 1% TM=Technology Management Council UFFC=Ultrasonics, Ferroelectrics, and Frequency Control VT=Vehicular Technology Figure 21 . The 2% breakdown of security cross -‐cutting issues by sponsoring societies. in IEEE Publications...... 134 Figure 38 . The breakdown of big data analytics by sponsoring societies. IM

CI 2% OTHER 133 COMM 3% 12% IEEE IEEE 150 17% 4% 2% Computer NPS 4% 4% CAS EMB 3% 5% RA SSC PHO 48% 5% 15% SMC 8% MTT 6%

Our future CEDA IE 5% CPMT technologies require 10% 10% Computer partnering with CSS 9% ED 10% 12% other Societies. AE=Aerospace & Electronic Systems AP=Antennas & Propagation BIO=Biometrics Council CAS=Circuits & Systems CE=Consumer AE=Aerospace & Electronic Systems AP=Antennas & Propagation BIO=Biometrics Council umer CAS=Circuits & Systems CE=Cons Electronics CEDA=Council on Electronic Design Automation CI=Computational Intelligence COMM=Communications Electronics CEDA=Council on Electronic Design Automation CI=Computational Intelligence COMM=Communications CPMT=Components, Packaging, & Manufacturing Technology CSS=Control Systems EC=Electromagnetic Compatibility Ed=Education CPMT=Components, Packaging, & Manufacturing Technology CSS=Control Systems EC=Electromagnetic Compatibility Ed=Education ED=Electron Devices Figure EMB=Engineering 17. Coverage in Medicine of & some Biology of the GRS=Geoscience top drivers & Remote Sensing IE=Industrial Electronics in IEEEED=Electro n Libraries Devices by EMB=Engineering individual in Medicine societies. & Biology GRS=Geoscience & Remote Sensing IE=Industrial Electronics IM=Instrumentation & Measurement IT=Information Theory ITS=Intelligent Transportation Systems MAG=Magnetics IM=Instrumentation & Measurement IT=Information Theory ITS=Intelligent Transportation Systems MAG=Magnetics MTT=Microwave Theory & Techniques NANO=Nanotechnology ncil Cou NPS=Nuclear & Plasma Sciences OCEAN=Oceanic Engineering MTT=Microwave Theory & Techniques NANO=Nanotechnology Council NPS=Nuclear & Plasma Sciences OCEAN=Oceanic Engineering PE=Power & Energy PHO=Photonics RA=Robotics & Automation SEN=Sensors Council SMC=Systems, Man, & Cybernetics PE=Power & Energy PHO=Photonics RA=Robotics & Automation SEN=Sensors Council SMC=Systems, Man, & Cybernetics SP=Signal Processing -‐ SSC=Solid State Circuits l SSIT=Socia Implications of Technology SUPERC=Council on Superconductivity SP=Signal Processing SSC=Solid-‐State Circuits SSIT=Social Implications of Technology SUPERC=Council on Superconductivity TM=Technology Management Council UFFC=Ultrasonics, Ferroelectrics, and Frequency Control VT=Vehicular Technology TM=Technology Management Council UFFC=Ultrasonics, Ferroelectrics, and Frequency Control 117 VT=Vehicular Technology Figure 43 . The breakdown of robotics in medicine by sponsoring societies. Figure 31 . The breakdown of networking and interconnectivity by sponsoring societies.

155 143 IEEE CS 2022 , Report DRAFT APPENDIX I. 23 Technologies Coverage in IEEE Publications 1/26/2014 IEEE 5:18 CS 2022 , Report DRAFT APPENDIX I. 23 Technologies Coverage in IEEE Publications 1/26/2014 5:18 PM PM

23 Robotics in Medicine 4 MOOC Sponsoring Societies Sponsoring Societies CONTENTS Ocean Educ ITS SEN 1% 1% CE 1% 1% 2% IEEE! 1. INTRODUCTION...... 5 CSS! IT! 2%! IM 3%! 3%! 2. SEAMLESS INTELLIGENCE CI 2% EMBS! CI! SCENARIO...... 8 3% CAS!4%! 20%! IEEE 3%! 4% 3. 23 TECHNOLOGIES IN 2022...... 14 COMM! Computer 4%! 4. DRIVERS AND DISRUPTORS...... 112 4% Other! 5. TECHNOLOGY COVERAGE IN EMB 4%! 5% IEEE XPLORE AND BY RA IE! 48% 5%! IEEE SOCIETIES...... 115 Computer! SMC CONTENTS5.1 Introduction...... 115 18%! 8% SMC! 5.2 Comparison...... 116 8%! 5.3 Summary of Quantitative IE Analysis Findings...... 119 10% SP! 10%! Ed! 6. IEEE COMPUTER SOCIETY CSS IEEE CS 2022 , Report DRAFT APPENDIX I. 23 Technologies Coverage in IEEE Publications 1/26/2014 5:18 PM 16%! IN 2022...... 120 IEEE CS 2022 , Report DRAFT APPENDIX 10% I. 23 Technologies Coverage in IEEE Publications 1/26/2014 5:18 PM AE=Aerospace & Electronic Systems AP=Antennas & Propagation ouncil BIO=Biometrics C CAS=Circuits & Systems CE=Consumer AE=Aerospace & Electronic Systems AP=Antennas & Propagation BIO=Biometrics Council CAS=Circuits & Systems CE=Consumer Electronics CEDA=Council on Electronic Design Automation CI=Computational Intelligence COMM=Communications 7. SUMMARY AND NEXT STEPS.....124 Electronics CEDA=Council on Electronic Design Automation CI=Computational Intelligence COMM=Communications CPMT=Components, Packaging, & Manufacturing Technology CSS=Control romagnetic Systems EC=Elect Compatibility Ed=Education CPMT=Components, Packaging, & Manufacturing Technology CSS=Control Systems EC=Electromagnetic Compatibility Ed=Education ED=Electron 16 Devices Natural EMB=Engineering in User Medicine & Biology Interfaces GRS=Geoscience & Remote Sensing IE=Industrial Electronics ED=Electron Devices EMB=Engineering in Medicine & Biology GRS=Geoscience & Remote Sensing IE=Industrial Electronics IM=Instrumentation & Measurement IT=Information Theory ortation ITS=Intelligent Transp Systems MAG=Magnetics 8. AUTHORS...... 126 IM=Instrumentation & Measurement IT=Information Theory ITS=Intelligent Transportation Systems MAG=Magnetics 17 3D Printing MTT=Microwave Theory & Techniques NANO=Nanotechnology Council NPS=Nuclear & Plasma g Sciences OCEAN=Oceanic Engineerin MTT=Microwave Theory & Techniques NANO=Nanotechnology ncil Cou NPS=Nuclear & Plasma Sciences OCEAN=Oceanic Engineering PE=Power Sponsoring & Energy PHO=Photonics Socie8es RA=Robotics & Automation SEN=Sensors Council SMC=Systems, Man, & Cybernetics PE=Power & Energy PHO=Photonics RA=Robotics & Automation SEN=Sensors Council SMC=Systems, Man, & Cybernetics SP=Signal Processing -‐ SSC=Solid State Circuits SSIT=Social Implications of Technology SUPERC=Council on Superconductivity APPENDIX I. 23 Technologies Coverage SP=Signal Sponsoring Processing -‐ SSC=Solid State Circuits Societies l SSIT=Socia Implications of Technology SUPERC=Council on Superconductivity UFFC RA TM=Technology Management Council UFFC=Ultrasonics, SP Ferroelectrics, and Frequency Control VT=Vehicular Technology TM=Technology Management Council UFFC=Ultrasonics, Ferroelectrics, and Frequency Control VT=Vehicular Technology 1% 0% in IEEE Publications...... 134 Figure 43 . The breakdown of robotics in medicine by sponsoring societies. Figure 24 . The breakdown 1% of MOOC by Other sponsoring societies.

PHO NPS 1% OTHER 5% 2% 155 IM 5% 2% 136 IEEE IEEE CPMT 2% 5% CE 26% 4% UFFC 6% SMC 8% EMBS 6% Computer 52% MTT 9% AP EMBS Our future 16% 27% technologies require EC 10% with Computer partnering 12% other Societies. AE=Aerospace & Electronic Systems AP=Antennas & Propagation BIO=Biometrics Council umer CAS=Circuits & Systems CE=Cons AE=Aerospace & Electronic Systems AP=Antennas & Propagation rcuits BIO=Biometrics Council CAS=Ci & Systems CE=Consumer Electronics CEDA=Council on Electronic Design Automation CI=Computational Intelligence COMM=Communications Electronics CEDA=Council on Electronic Design Automation CI=Computational Intelligence COMM=Communications CPMT=Components, Packaging, & Manufacturing Technology CSS=Control Systems EC=Electromagnetic Compatibility Ed=Education CPMT=Components, Packaging, & Manufacturing Technology CSS=Control Systems EC=Electromagnetic Compatibility Ed=Education ED=Electron Devices Figure 18. EMB=Engineering Coverage in Medicine of & some Biology of the GRS=Geoscience top disruptors & Remote Sensing IE=Industrial Electronics in IEEE Libraries by individual societies ED=Electron Devices EMB=Engineering in Medicine & Biology GRS=Geoscience & Remote Sensing IE=Industrial Electronics IM=Instrumentation & Measurement IT=Information Theory ITS=Intelligent Transportation Systems MAG=Magnetics IM=Instrumentation & Measurement IT=Information Theory ITS=Intelligent Transportation Systems MAG=Magnetics MTT=Microwave Theory & Techniques NANO=Nanotechnology Council NPS=Nuclear & Plasma Sciences OCEAN=Oceanic Engineering MTT=Microwave Theory & Techniques NANO=Nanotechnology Council NPS=Nuclear & Plasma g Sciences OCEAN=Oceanic Engineerin PE=Power & Energy PHO=Photonics RA=Robotics & Automation SEN=Sensors Council SMC=Systems, Man, & Cybernetics SP=Signal Processing SSC=Solid-‐State Circuits SSIT=Social Implications of Technology SUPERC=Council on Superconductivity PE=Power & Energy PHO=Photonics RA=Robotics & Automation SEN=Sensors Council SMC=Systems, Man, & Cybernetics TM=Technology Management Council UFFC=Ultrasonics, Ferroelectrics, and Frequency Control VT=Vehicular Technology SP=Signal Processing -‐ SSC=Solid State Circuits SSIT=Social Implications of Technology SUPERC=Council 118 on Superconductivity TM=Technology Management Council UFFC=Ultrasonics, Ferroelectrics, and Frequency ular Control VT=Vehic Technology

Figure 37 . The breakdown of 3D printing by sponsoring societies. Figure 36 . The breakdown of natural user interfaces by sponsoring societies.

149 148 CONTENTS be organized, whereas the second hints at going 5.3 Summary of Quantitative alone. In the other two, the Computer Society has Analysis Findings 1. INTRODUCTION...... 5 a minority of the coverage and desire to achieve As mentioned above, our future technologies higher growth the first graph showcases the large 2. SEAMLESS INTELLIGENCE require partnering with other Societies with more SCENARIO...... 8 leadership coverage and could organize the many assets, authors, experts, and organizers. 3. 23 TECHNOLOGIES IN 2022...... 14 contributing Societies to work together, the second 4. DRIVERS AND DISRUPTORS...... 112 shows many equal players that may lack leadership While our Society is known for its software engi- 5. TECHNOLOGY COVERAGE IN to accomplish together larger goals. neering heritage, there are now many hardware-focused Societies using our technologies, and they IEEE XPLORE AND BY The survey also identified four major disruptors IEEE SOCIETIES...... 115 are ripe for future collaboration. CONTENTS whose pie charts are shown in Figure 18. 5.1 Introduction...... 115 IEEE has encouraged Societies in the stewardship Again, we see two Societies with the most coverage 5.2 Comparison...... 116 of their own technical field of interest “silo” for that are also able to provide the leadership to orga- 5.3 Summary of Quantitative excellence and growth, but our analysis shows the nize this technology domain. The other two show Analysis Findings...... 119 Societies need to partner on many emerging/fu- wide coverage with no obvious organizing leader. 6. IEEE COMPUTER SOCIETY ture technologies. Of our chosen 23 technologies, IN 2022...... 120 Because this quantitative analysis looks at the the Computer Society has the top share in only 7. SUMMARY AND NEXT STEPS.....124 Societies’ current technical assets in Xplore as the 11, or 50 Percent; our rivals include EMB (Engi- 8. AUTHORS...... 126 foundation for future technologies, there is the risk neering in Medicine and Biology), RA (Robotics APPENDIX I. 23 Technologies Coverage of disruptive forces that could kill some current and Automation), NANO (Nanotechnology), PHO in IEEE Publications...... 134 technology trends, revive dead-end “solutions,” or (Photonics), COMM (Communications), CI (Com- create new marriages/mergers of technologies. putational Intelligence), ED (Electronics Devices), This document and its analysis is the Computer CEDA (Council on Electronic Design Automation), Society’s perspective, but it could compliment and CPMT (Components, Packaging, and Manu- the Communication Society’s perspective in its facturing Technology). It is easy to partner when ComSoc 2020 report. At the higher IEEE Technical you are in the top position, but it is more difficult Activities level, the Future Technologies Committee to reach out to those Societies with more “wealth” has its own perspective. The Standards Associa- and “talent” as a contributing partner. Thus, IEEE tion enlisted several Societies in its future view of Societies need incentives to reach out of their field smart grid, for example, the IEEE Smart Grid Vision to work with other Societies in moving IEEE to Our future for Computing: 2030 and Beyond. At the highest thought leadership. For example, while the Com- technologies require level, the IEEE Board of Governors is also defining puter Society trails in a distant second place to the partnering with our future world of technologies. All IEEE operating EMB Society in life sciences, we will have to work hard on this partnership to jointly reap the benefits other Societies. units need to share and plan for the future of future growth.

119 CONTENTS 1. INTRODUCTION...... 5 2. SEAMLESS INTELLIGENCE SCENARIO...... 8 6. IEEE Computer 3. 23 TECHNOLOGIES IN 2022...... 14 4. DRIVERS AND DISRUPTORS...... 112 Society in 2022 5. TECHNOLOGY COVERAGE IN IEEE XPLORE AND BY IEEE SOCIETIES...... 115 6.CONTENTS IEEE COMPUTER SOCIETY IN 2022...... 120 The previous sections of this document have connected with other members virtually or at face- 7. SUMMARY AND NEXT STEPS.....124 focused on the “what” and the “why.” What to-face meetings. 8. AUTHORS...... 126 technologies will be important? Why will they be Early use of technology by next-generation profes- important (drivers and disruptors)? APPENDIX I. 23 Technologies Coverage sionals will drive the average age of our members in IEEE Publications...... 134 Left to be addressed are the questions of “who” down 10 years or more. The invent-to-publish cycle and “how.” will be much shorter, with almost instant access • Who will take up the work of tomorrow? This to materials, ability to collaborate, and physical/ requires a focus on the development of the pro- virtual meetings. Crowd-sourced peer review will fession and professionals. The IEEE Computer be the norm, but more important are new stan- Society must foster a highly skilled workforce. dards that will drive high quality: communities will • How can the CS fulfill its mission to benefit tend to code that “lives,” and community members humanity? This requires a focus on the impact will enforce professional codes of practice and of technology on society. The CS must take collaborate to develop “building codes” for secure responsibility for the beneficial implementation infrastructure (as the CS is pursuing through its of technology. cybersecurity initiative). The value of these new products will be known The CS will support “seamless intelligence” for our immediately: users will rate their benefits. While members. Our members will all be global—truly technology will always have its “cool factor,” tradi- CS must take global—and truly connected. Truly global means tional engineering principles and rigor will not be responsibility almost anyone on the globe who has interest can compromised. What will be different is how knowl- instantly become a member and participate in for the beneficial edge is transferred. You will learn by doing and the Society’s special technical communities, get implementation of learn from doers. access to all of its products and services, and be technology.

120 CONTENTS 1. INTRODUCTION...... 5 This new culture will give rise to more interactive We must also make sure our Society changes at 2. SEAMLESS INTELLIGENCE events like hackathons, gaming conventions, and pace with technology, but no slower. While this SCENARIO...... 8 meetups. Traditional academic meetings will be document does not go into detail about the internal 3. 23 TECHNOLOGIES IN 2022...... 14 joined by more practitioner conferences; the two workings of IEEE or the CS, some things become 4. DRIVERS AND DISRUPTORS...... 112 will complement each other. apparent from a simple strengths, weaknesses, opportunities, and threats (SWOT) analysis. 5. TECHNOLOGY COVERAGE IN Recent developments in Internet security and pri- IEEE XPLORE AND BY vacy have eroded the universal view that technolo- CONTENTSIEEE SOCIETIES...... 115 gy advancement is always for the good. “Seamless 6. IEEE COMPUTER SOCIETY intelligence” may not sound so positive to every- IN 2022...... 120 one. The CS must be societally mindful of preserv- 7. SUMMARY AND NEXT STEPS.....124 ing privacy as well as overall use of technology. We 8. AUTHORS...... 126 must make sure that societal changes are at pace APPENDIX I. 23 Technologies Coverage with technology, but not slower. in IEEE Publications...... 134 More generally, all Societies should realize that technology is just one tile in a complex puzzle, and we can understand technology implications not only by looking at technology itself but also by looking at the whole puzzle. In a way, technology is about to be commoditized, and we need to appre- ciate that its value is in the economic, social, and cultural domain. These aspects are so intertwined with technology that we can no longer claim that technology is neutral. We need to consider in our publications and conferences voices from other sectors and progress the whole puzzle (Saracco, personal communication). CS must take responsibility for the beneficial implementation of technology.

121 Table 1. IEEE SWOT.

Strengths Weaknesses Brand denotes quality Not nimble: IEEE has a hierarchical structure, CONTENTS whereas flat structures are known to be more 1. INTRODUCTION...... 5 nimble and less bureaucratic 2. SEAMLESS INTELLIGENCE Scope encompasses all technologies No presence in workforce development, which will SCENARIO...... 8 (multidisciplinary) be widely needed 3. 23 TECHNOLOGIES IN 2022...... 14 Global reach puts it in a position to develop technolo- IEEE’s delivery channels are not keeping pace 4. DRIVERS AND DISRUPTORS...... 112 gies that cross borders 5. TECHNOLOGY COVERAGE IN Neutral organization, which is an advantage in tech- Very little professionally written content in CS IEEE XPLORE AND BY IEEE SOCIETIES...... 115 nical fields that are highly proprietaryt publications, making it difficult to quickly reflect technology trends 6.CONTENTS IEEE COMPUTER SOCIETY IN 2022...... 120 Opportunities Threats 7. SUMMARY AND NEXT STEPS.....124 Need for highly skilled technical workforce is rising Organizations in emerging regions may be able to 8. AUTHORS...... 126 offer similar or better content, communications, APPENDIX I. 23 Technologies Coverage services, communities, and chapter models that in IEEE Publications...... 134 reflect local languages and norms; the same organizations can offer more universal products and services via the Internet at lower prices CS can lead IEEE in how to do open access Open access will slow IEEE’s financial engine and disrupt its business model: intellectual property will not be granted to the IEEE at little to no cost IEEE can expand its influence in technology applica- Self-credentialing makes membership in a Society tions, such as healthcare obsolete (stack overflow); peer review bodies are also self-organizing (research gate) Exciting time to reframe what it means to be a tech- Google Scholar-like services undermine IEEE nology professional, at all levels collections The need for technical education will grow; IEEE can Distrust of technology could grow, putting IEEE in develop professional standards of practice, such as an unfavorable light CS must take software engineering; IEEE can offer certifications responsibility that require membership for the beneficial Asia is a huge opportunity at the moment: IEEE can Other professional associations and information uncover what’s next services companies are moving faster than IEEE implementation of right now technology.

122 Table 2. Breakdown of who and how will be benefiting from IEEE CS.

CONTENTS Who How Multidisciplinary professionals Deliver highest quality content, but in small units 1. INTRODUCTION...... 5 2. SEAMLESS INTELLIGENCE Global citizens Digest, synthesize, summarize, and repackage content SCENARIO...... 8 Constantly become “instant experts” Offer skills development and training 3. 23 TECHNOLOGIES IN 2022...... 14 Volunteers who can devote small bits of time Focus on “high touch” but small face-to-face meetings 4. DRIVERS AND DISRUPTORS...... 112 intermittently on the latest topics 5. TECHNOLOGY COVERAGE IN IEEE XPLORE AND BY A SWOT analysis can be used for scenario plan- IEEE SOCIETIES...... 115 CONTENTS ning. The CS used scenario planning in 2004 when 6. IEEE COMPUTER SOCIETY 1 IN 2022...... 120 it was developing Strategic Plan 5, to imagine what the year 2020 might look like. Much of the 2004 7. SUMMARY AND NEXT STEPS.....124 scenario includes elements that are still current. 8. AUTHORS...... 126 This table summarizes the SP5 predictions as to APPENDIX I. 23 Technologies Coverage who the CS would be serving in 2020, and how. in IEEE Publications...... 134 If these 2020 future states are still relevant, how is the CS doing? SWOT shows that we have developed some strength in these areas, but most of these challenges remain. The second part of the 2022 report, CS Strategic Plan 8, will address how the Computer Society must be organized to have an impact in the year 2022, in light of the technologies outlined in this report. The opportunities for IEEE Computer Society in 2022 are endless, and the future is exciting.

CS must take responsibility for the beneficial implementation of technology.

123 CONTENTS 1. INTRODUCTION...... 5 2. SEAMLESS INTELLIGENCE SCENARIO...... 8 7. Summary 3. 23 TECHNOLOGIES IN 2022...... 14 4. DRIVERS AND DISRUPTORS...... 112 and Next Steps 5. TECHNOLOGY COVERAGE IN IEEE XPLORE AND BY IEEE SOCIETIES...... 115 6.CONTENTS IEEE COMPUTER SOCIETY IN 2022...... 120 In this report, several technologists evaluated seamless intelligence scenario has deep roots in 7. SUMMARY AND NEXT STEPS.....124 23 technology areas that have the potential to technology advancement that did not exist in the 8. AUTHORS...... 126 disrupt the world we perceive today. Some of near past. In particular, by 2022, computing devices APPENDIX I. 23 Technologies Coverage them are already known and being adopted today, will vary from nano- to mega-scale, and wireless/ in IEEE Publications...... 134 such as multicore, high-performance computing, wired networks will enable access to integrated cloud computing, and software-defined networks. services. Virtual connectivity will enable integration Others are only being explored at this time, such of relevant computing resources to provide users as 3D printing, nonvolatile memories, and quan- with integrated and seamless services. The re- tum computing. These 23 technology areas cover sulting ecosystem will offer seamless, continuous, a spectrum of policies (open intellectual property uninterrupted services that enhance automation, movement, massively online open courses), tech- productivity, collaboration, and access to intelli- nologies (15 areas), market categories (compu- gence and knowledge through emerging HCI. tational biology and bioinformatics, life sciences, However, the benefit of technology is what we and robotics in medical care), and some vertically make of it. Societies will face challenges in realiz- applied areas (sustainability and security cross- ing technologies that benefit humanity instead of cutting issues). destroying and intruding on the human rights of These 23 areas have resulted out of brainstorm- privacy and freedom of access to information. How ing by technologists. They have been confirmed will these advancements will help humanity will The benefit of through subsequent questionnaires and also depend on the pace of the policies and regulations technology is what compared through digital libraries exploration. All that accompany the technologies’ evolution. Like areas have been tied into a single scenario that we we make of it. many times in the past, technology is an enabler. call “seamless intelligence.” While similar to past It is up to the human race to leverage it in the best pervasive and ubiquitous computing scenarios, the possible way to advance human society.

124 CONTENTS 1. INTRODUCTION...... 5 This report is made freely available, but it was grad- 2. SEAMLESS INTELLIGENCE ually distributed within the IEEE Computer Society SCENARIO...... 8 to get the best feedback from our readership. Over 3. 23 TECHNOLOGIES IN 2022...... 14 the course of the following year, it will be used in 4. DRIVERS AND DISRUPTORS...... 112 preparation of the IEEE Computer Society’s strategic plan. While IEEE CS 2022 is more technology 5. TECHNOLOGY COVERAGE IN IEEE XPLORE AND BY focused, the strategic plan will be more IEEE Com- CONTENTSIEEE SOCIETIES...... 115 puter Society focused. 6. IEEE COMPUTER SOCIETY Ultimately, this was a rewarding exercise. It was IN 2022...... 120 very interesting to lead and participate in technol- 7. SUMMARY AND NEXT STEPS.....124 ogy discussions about the future and encouraging 8. AUTHORS...... 126 to see how many of the technologists converged APPENDIX I. 23 Technologies Coverage on a single scenario as well as how many similar in IEEE Publications...... 134 concerns about privacy and security transpired through all other discussions. We all have learned a lot from this process, and we hope that the readers of this document will learn something, too.

Core Authors: Arif Merchant, Danny Lange, Dejan Milojicic, Eitan Frachtenburg, Hasan Alkhatib, Hironori Kasahara, Karsten Schwan, Paolo Faraboschi, and Phil Laplante. The benefit of technology is what we make of it.

125 CONTENTS 1. INTRODUCTION...... 5 2. SEAMLESS INTELLIGENCE SCENARIO...... 8 8. Authors 3. 23 TECHNOLOGIES IN 2022...... 14 4. DRIVERS AND DISRUPTORS...... 112 5. TECHNOLOGY COVERAGE IN IEEE XPLORE AND BY This document was a team effort, spearheaded in networking and distributed computing. In 1998, IEEE SOCIETIES...... 115 by a core team of authors who formulated the he founded IP Dynamics, a venture-backed start- 6.CONTENTS IEEE COMPUTER SOCIETY overall text and process. This team, organized by up, where he was CEO until 2007. IP Dynamics IN 2022...... 120 Dejan Milojicic, met twice in face-to-face meetings developed the industry’s first solution for pol- 7. SUMMARY AND NEXT STEPS.....124 and had a few phone conferences. In addition, oth- icy-based software-defined connectivity from 8. AUTHORS...... 126 er people contributed to various parts of the docu- anywhere to anywhere, regardless of location and ment; the rest of this section lists all contributors. underlying physical networks. In 2007, Alkhatib 8.1 The Core Team of Authors...... 126 joined Microsoft as General Manager of Enterprise 8.2 Major Contributors of Networking, and then became Chief Architect of Individual Sections...... 130 8.1 The Core Team of Authors Networking and Network Security for Microsoft’s 8.3 Acknowledgements...... 133 The core team of authors included Hasan Alkhatib, Paolo Faraboschi, Eitan Frachtenburg, Hironori Ka- cloud computing platform, Windows Azure, in APPENDIX I. 23 Technologies Coverage 2008. He’s been President of SSN Services, a con- in IEEE Publications...... 134 sahara, Danny Lange, Phil Laplante, Arif Merchant, sulting firm specializing in network virtualization, Dejan Milojicic, and Karsten Schwan. cloud computing, and innovations in higher edu- Hasan Alkhatib, Entrepreneur and President of cation, since 2011. Alkhatib has published over 50 SSN Services, LLC papers and holds 26 patents and 37 other pending patent applications on networking, virtualization, security, and cloud computing. He has chaired five IEEE/CS conferences and was keynote speaker at five others. He has served as guest editor for IEEE Micro and chaired TCMM in 1991-1992. Alkhatib holds a PhD in Electrical and Computer Engineer- ing from UC Santa Barbara. Alkhatib is a Silicon Valley entrepreneur and vet- Hasan Alkhatib wrote the Seamless Intelligence eran. He was a Computer Engineering Professor at Scenario and Cloud Computing sections. Santa Clara University from 1981-1998, specializing

126 CONTENTS 1. INTRODUCTION...... 5 Paolo Faraboschi, HP Labs, Spain technologies, parallel algorithms, and computer 2. SEAMLESS INTELLIGENCE architecture. Prior to Facebook, Frachtenberg was SCENARIO...... 8 an Applied Researcher at Microsoft/Powerset 3. 23 TECHNOLOGIES IN 2022...... 14 (working on Semantic Web search), and before 4. DRIVERS AND DISRUPTORS...... 112 that, a Postdoctoral Fellow at Los Alamos National Laboratory (working on supercomputer operating 5. TECHNOLOGY COVERAGE IN IEEE XPLORE AND BY systems). He obtained his PhD in Computer Sci- CONTENTSIEEE SOCIETIES...... 115 ence from Hebrew University. 6. IEEE COMPUTER SOCIETY Frachtenberg wrote the sections on Big Data IN 2022...... 120 Paolo Faraboschi is a Distinguished Technologist and Analytics, and Open Intellectual Property 7. SUMMARY AND NEXT STEPS.....124 at HP Labs, working on energy-efficient servers. (together with Phil Laplante and encompassing 8. AUTHORS...... 126 From 2004 to 2009, he led a group on system-level crowd-sourcing). 8.1 The Core Team of Authors...... 126 simulation. From 1995 to 2003, at HPL Cambridge (Mass.), he was the Principal Architect of the Lx/ Hironori Kasahara, Waseda University, Japan 8.2 Major Contributors of Individual Sections...... 130 ST200 family of VLIW embedded cores. Faraboschi is an active member of the computer architec- 8.3 Acknowledgements...... 133 ture community: he has served as guest editor of APPENDIX I. 23 Technologies Coverage in IEEE Publications...... 134 IEEE Micro’s TopPicks 2012, and Program Chair for CF2012, HiPEAC2010, MICRO2008 and MI- CRO2001. He has authored 23 patents, the book Embedded Computing: A VLIW Approach, and over 50 papers. Faraboschi holds a PhD in EECS (1993) Hironori Kasahara has been an IEEE CS BoG from University of Genoa, Italy. member since 2009 and a Chair of the Multicore STC since 2012. In 1985, he received a PhD in EE Paolo Faraboschi contributed the sections on from Waseda University, Tokyo, where he has been Universal Memory, 3D Integrated Circuits, and a professor of computer science since 1997, and Photonics. a Director of the Advanced Multicore Research Institute. He was a visiting scholar at UC Berkeley Eitan Frachtenberg, Facebook and the University of Illinois at Urbana-Cham- Eitan Frachtenberg is a Research Scientist at paign’s Center for Supercomputing R&D. Kasahara Facebook, analyzing social behavior on large-scale received the IFAC World Congress Young Author datasets. His research interests include data min- Prize, and IPSJ Sakai Memorial Special Research ing, performance evaluation and optimization, Web Award. He has led Japanese national projects on

127 CONTENTS 1. INTRODUCTION...... 5 parallelizing compilers, multicore, and green com- has numerous patents to his credit, has presented 2. SEAMLESS INTELLIGENCE puting systems. http://www.kasahara.cs.waseda. his work at leading conferences, and published SCENARIO...... 8 ac.jp/kasahara.html.en articles in many journals. 3. 23 TECHNOLOGIES IN 2022...... 14 Hironori Kasahara wrote the Multicore section of Danny Lange contributed the Machine Learning 4. DRIVERS AND DISRUPTORS...... 112 the report. and Intelligent Systems, Natural User Interfaces, 5. TECHNOLOGY COVERAGE IN Danny Lange, Microsoft and Quantum Computing sections. IEEE XPLORE AND BY IEEE SOCIETIES...... 115 Phil Laplante, Pennsylvania State University 6.CONTENTS IEEE COMPUTER SOCIETY IN 2022...... 120 7. SUMMARY AND NEXT STEPS.....124 8. AUTHORS...... 126 8.1 The Core Team of Authors...... 126 8.2 Major Contributors of Individual Sections...... 130 8.3 Acknowledgements...... 133 Danny B. Lange is Manager of Elastic Machine APPENDIX I. 23 Technologies Coverage Learning at Amazon.com. Prior to Amazon, he in IEEE Publications...... 134 was Principal Development Manager at Microsoft, Phil Laplante is Professor of Software Engineering where he was leading the product team for large- at The Pennsylvania State University. He received scale machine learning. Previously, he was the Bing his BS, M.Eng., and PhD from Stevens Institute of Software Architect responsible for mobile search. Technology and an MBA from the University of Lange was the co-founder of Vocomo Software, Colorado. Laplante is a Fellow of IEEE and SPIE a speech technology company, and as CTO of and has won several international awards for his General Magic, built the architecture for its OnStar teaching, research, and service. He has worked in voice response service. Prior to joining General avionics, CAD, and software testing systems and Magic, he was Computer Scientist at IBM Tokyo has published 27 books and more than 200 schol- Research. Lange has made significant contribu- arly papers. Laplante’s research interests are in tions in the areas of distributed computing, big data software testing, requirements engineering, and analytics, cloud computing, mobile agents, speech software quality and management. recognition, program visualization, and hypertext. Phil Laplante wrote the section on Open Intellectu- He holds an MS and a PhD in Computer Science al Property (together with Eitan Frachtenbery) and from the Technical University of Denmark. Lange the section on MOOCs.

128 CONTENTS 1. INTRODUCTION...... 5 Arif Merchant, Google He was a founding editor in chief of IEEE Com- 2. SEAMLESS INTELLIGENCE putingNow and has been on many conference SCENARIO...... 8 program committees and journal editorial boards. 3. 23 TECHNOLOGIES IN 2022...... 14 Milojicic worked at the OSF Research Institute, 4. DRIVERS AND DISRUPTORS...... 112 Cambridge, MA [1994-1998] and Institute “Mihajlo Pupin,” Belgrade, Serbia [1983-1991]. He received 5. TECHNOLOGY COVERAGE IN IEEE XPLORE AND BY his PhD from University of Kaiserslautern, Germa- IEEE SOCIETIES...... 115 ny (1993) and MSc/BSc from Belgrade University, 6.CONTENTS IEEE COMPUTER SOCIETY Serbia (1983/86). Milojicic is an IEEE Fellow, ACM IN 2022...... 120 Distinguished Engineer, and USENIX member. He 7. SUMMARY AND NEXT STEPS.....124 has published over 130 papers and 2 books; he has Arif Merchant is a Research Scientist with the Stor- 8. AUTHORS...... 126 12 patents and 25 patent applications. His areas age Analytics group at Google, where he studies 8.1 The Core Team of Authors...... 126 of expertise include systems software, distributed interactions between components of the storage computing, mobile computing, and services. 8.2 Major Contributors of stack. Prior to this, he was with HP Labs, where he Individual Sections...... 130 worked on storage QoS, distributed storage sys- Dejan Milojicic wrote the sections on High-Per- 8.3 Acknowledgements...... 133 tems, and stochastic models of storage. Merchant formance Computing and Sustainability. He also APPENDIX I. 23 Technologies Coverage holds a B.Tech. from IIT Bombay and a PhD in Com- initiated and organized this effort and contributed in IEEE Publications...... 134 puter Science from Stanford University. to the remaining general sections. Karsten Schwan, GaTech Arif Merchant contributed the 3D Printing section.

Dejan Milojicic, HP Labs, Palo Alto

Karsten Schwan is a Regents’ Professor in the College of Computing at the Georgia Institute of Technology, where he is also a Director of the Cen- ter for Experimental Research in Computer Sys- Dejan Milojicic is a senior researcher at HP Labs tems (CERCS), with co-directors from both GT’s and the IEEE Computer Society 2014 President.

129 CONTENTS 1. INTRODUCTION...... 5 College of Computing and School of Electrical and Mohammed AlQuraishi, Harvard Medical School 2. SEAMLESS INTELLIGENCE Computer Engineering. His MS and PhD are from SCENARIO...... 8 Carnegie-Mellon University; his PhD concerned 3. 23 TECHNOLOGIES IN 2022...... 14 high-performance computing, addressing operat- 4. DRIVERS AND DISRUPTORS...... 112 ing and programming systems support for the Cm* multiprocessor, after which he conducted exten- 5. TECHNOLOGY COVERAGE IN IEEE XPLORE AND BY sive research in real-time and distributed systems. IEEE SOCIETIES...... 115 His current work ranges from topics in operating Mohammed AlQuraishi is a Systems Biology Fellow 6.CONTENTS IEEE COMPUTER SOCIETY systems to middleware to parallel and distributed IN 2022...... 120 systems, focusing on information-intensive dis- at Harvard Medical School. Prior to joining Harvard, 7. SUMMARY AND NEXT STEPS.....124 tributed applications in the enterprise domain and he completed his PhD in Genetics from Stanford University under the supervision of Harley McAd- 8. AUTHORS...... 126 in the high-performance domain. www.cc.gatech. edu/~schwan ams and Lucy Shapiro. His research interests lie at 8.1 The Core Team of Authors...... 126 the intersection of systems and structural biology. 8.2 Major Contributors of Karsten Schwan contributed the sections on De- AlQuraishi aims to obtain a systems-level under- Individual Sections...... 130 vice and Nanotechnology, Internet of Things, and standing of biological processes through a molecu- 8.3 Acknowledgements...... 133 Networking and Interconnectivity. lar-level understanding of biological structures and APPENDIX I. 23 Technologies Coverage their interactions, and to that end, he is developing in IEEE Publications...... 134 8.2 Major Contributors computational methods for predicting the bind- of Individual Sections ing partners and quantitative binding affinities of biological molecules from their atomic structure. In addition to the core team, a few individual con- His work combines recent advances in machine tributed to substantial parts of the document. learning and information theory with concepts from These valuable contributors include Mohammed statistical mechanics and biophysics. AlQaraishi, Angela Burgess, Hiroyasu Iwata, Rick McGeer, and John Walz. Mohammed AlQuraishi wrote the section on Bioinformatics.

130 CONTENTS 1. INTRODUCTION...... 5 Angela Burgess, IEEE Computer Society David Forsyth, University of Illinois 2. SEAMLESS INTELLIGENCE at Urbana-Champaign SCENARIO...... 8 3. 23 TECHNOLOGIES IN 2022...... 14 4. DRIVERS AND DISRUPTORS...... 112 5. TECHNOLOGY COVERAGE IN IEEE XPLORE AND BY IEEE SOCIETIES...... 115 6.CONTENTS IEEE COMPUTER SOCIETY Angela R. Burgess is the Executive Director of IN 2022...... 120 the IEEE Computer Society, the world’s leading David Forsyth holds a B.Sc. and an M.Sc. from the 7. SUMMARY AND NEXT STEPS.....124 membership organization for computing profes- University of the Witwatersrand, Johannesburg, 8. AUTHORS...... 126 sionals. The IEEE Computer Society is the largest and an M.A. and a D.Phil from Oxford University. 8.1 The Core Team of Authors...... 126 technical organization within the IEEE, which has He has held academic appointments in computer 8.2 Major Contributors of more than 400,000 members worldwide. As head science at the University of Iowa, the University of Individual Sections...... 130 of staff, Burgess oversees the Computer Society’s California, Berkeley, and at the University of Illinois, 8.3 Acknowledgements...... 133 Digital Library, 17 journals, 12 technical magazines, Urbana-Champaign. He has published papers on APPENDIX I. 23 Technologies Coverage and 300 conference proceedings annually, along most areas in computer vision, on computer graph- in IEEE Publications...... 134 webinars, podcasts, courseware, bodies of knowl- ics, on machine learning, and on human-computer edge, and certifications. She has more than 25 interaction. He is co-author of a leading textbook years’ experience with the IEEE Computer Society on computer vision. He is a Fellow of the IEEE and has been the Executive Director since 2007. and of the ACM, and is currently serving a term as Burgess received a BS in Journalism and Interna- Editor-in-Chief of IEEE Transactions on Pattern tional Studies from Iowa State University and an Analysis and Machine Intelligence. Executive MBA from the Peter F. Drucker School of David Forsyth wrote the section on Computer Management, Claremont Graduate University. Vision and Pattern Analysis. Angela Burgess wrote the section on SWOT analysis and also contributed to the Section on 2022 Technologies Coverage in IEEE.

131 CONTENTS 1. INTRODUCTION...... 5 Hiroyasu Iwata, Waseda University Professor in the Computer Science Department at 2. SEAMLESS INTELLIGENCE the University of British Columbia, before returning SCENARIO...... 8 to UC Berkeley as a Research Engineer in 1991. In 3. 23 TECHNOLOGIES IN 2022...... 14 1993, he co-founded Cadence Berkeley Laborato- 4. DRIVERS AND DISRUPTORS...... 112 ries, the research arm of Cadence Design Systems, and five years later, he co-founded Softface, Inc., 5. TECHNOLOGY COVERAGE IN IEEE XPLORE AND BY where he remained as Chief Scientist until 2003, IEEE SOCIETIES...... 115 when he joined Hewlett-Packard Laboratories, 6.CONTENTS IEEE COMPUTER SOCIETY Hiroyasu Iwata received a BS, MS, and PhD in me- leaving in 2014 as a Distinguished Technologist. IN 2022...... 120 chanical engineering from Waseda University, Tokyo, McGeer co-founded the PlanetLab consortium in 7. SUMMARY AND NEXT STEPS.....124 Japan. He was a Research Associate and an Assistant 2003, and currently serves on the Steering Com- 8. AUTHORS...... 126 Professor at Waseda University from 2001 to 2004, mittee. In 2013, he joined US Ignite on a part-time, 8.1 The Core Team of Authors...... 126 and 2005, has been an Assistant Professor at the volunteer basis as Chief Scientist. He is currently a Principal Investigator with the Communications 8.2 Major Contributors of Institute for Biomedical Engineering, Consolidated Individual Sections...... 130 Research Institute for Advanced Science and Medical and Design Group, a research arm of SAP America. McGeer is the author of over 100 papers and one 8.3 Acknowledgements...... 133 Care, Waseda University. He is also a member of the book in the fields of CaD, circuit theory, program- APPENDIX I. 23 Technologies Coverage Humanoid Robotics Institute and the WABOT HOUSE in IEEE Publications...... 134 Laboratory of Waseda University. ming languages, distributed systems, networking, and information system design. His research inter- Hiroyasu Iwata wrote the section on Robotics in ests include logic synthesis, timing analysis, formal Medical Care. verification, circuit simulation, programming lan- Rick McGeer, Communications and Design Group, guages, networking, wide-area distributed systems, SAP America and cloud systems. He has acted as a Principal Investigator on three DARPA and three GENI projects over the past two decades. He is also an Adjunct Professor of Computer Science at the Uni- versity of Victoria, Victoria, BC, Canada. Rick McGeer wrote the section on Software-De- fined Networks.

Rick McGeer received his PhD in Computer Sci- ence from UC Berkeley. He was an Assistant

132 CONTENTS 1. INTRODUCTION...... 5 John Walz, Retired from Lucent/AT&T 8.3 Acknowledgements 2. SEAMLESS INTELLIGENCE Greg Astfalk, HP Labs, for contributions to the Uni- SCENARIO...... 8 versal Memory section. 3. 23 TECHNOLOGIES IN 2022...... 14 Evan Butterfield, IEEE Computer Society, for secur- 4. DRIVERS AND DISRUPTORS...... 112 ing a number of images for final production of the 5. TECHNOLOGY COVERAGE IN document IEEE XPLORE AND BY IEEE SOCIETIES...... 115 Elena Gerstman, for facilitating the first 2022 Report 6.CONTENTS IEEE COMPUTER SOCIETY Former IEEE Computer Society President John W. core team meeting. IN 2022...... 120 Walz has been elected 2014 IEEE Division VIII Dele- Moray McLaren, HP Labs, for contributions to the gate-Elect/Director-Elect. Walz retired from Lu- 7. SUMMARY AND NEXT STEPS.....124 Photonics section. 8. AUTHORS...... 126 cent Technologies/AT&T with more than 20 years John Reimer, IEEE Computer Society, for conducting 8.1 The Core Team of Authors...... 126 of management/coaching experience, covering searches and producing pie charts for appendix. 8.2 Major Contributors of positions in hardware and software engineering, Individual Sections...... 130 quality planning and auditing, standards implemen- Jenny Stout, IEEE Computer Society, for copyediting 8.3 Acknowledgements...... 133 tation, and strategic planning. He has coauthored this document. three books covering the use of IEEE software APPENDIX I. 23 Technologies Coverage Michael Werhman, for conducting the survey on in IEEE Publications...... 134 engineering standards to support CMMI, ISO 9001, and Lean Six Sigma. Walz has held leadership drivers and disruptors. positions in national and international industry and Chandrakant Patel for contributing Figures 2 and 13. professional organizations, including US Techni- Robert Stack contributing black and white drawings cal Advisory Group on Quality Management ISO on pages 5, 9, 10, 11, 13, 16, 28, 52, 67, 73, 82, and 87. 9001 and Risk Management ISO 31000; American Society for Quality (ASQ) Electronics and Commu- nications Division and its Sarbanes-Oxley Forum; the Quality Excellence for Suppliers of Telecom- munications Forum; and the Information Integrity Coalition. John Walz wrote the section on Life Sciences and lead writing the section on 2022 Technologies Coverage in IEEE.

133 CONTENTS 1. INTRODUCTION...... 5 2. SEAMLESS INTELLIGENCE SCENARIO...... 8 APPENDIX I. 23 3. 23 TECHNOLOGIES IN 2022...... 14 4. DRIVERS AND DISRUPTORS...... 112 Technologies Coverage 5. TECHNOLOGY COVERAGE IN IEEE XPLORE AND BY IEEE SOCIETIES...... 115 6.CONTENTS IEEE COMPUTER SOCIETY in IEEE Publications IN 2022...... 120 7. SUMMARY AND NEXT STEPS.....124 8. AUTHORS...... 126 APPENDIX I. 23 Technologies Coverage Many IEEE Societies and councils are publish- Also included are summary charts of the amount of in IEEE Publications...... 134 ing content about each of the 23 technologies. this content published in each technology area. For These potential partners are identified on the sub- an example of how to utilize this data, life sciences sequent pages of this Appendix. accounts for 27 percent of the content assessed The percentages represent one measurement of for this exercise, and the Computer Society ac- the degree of involvement each Society/council counts for 12 percent of that content. Or, taken as (S/C) has in each technology. It is calculated as the a whole, the Computer Society accounts for 25 share of relevant content among the top-publishing percent of the 23 technologies’ content identified. S/Cs of that material in Xplore (the lowest-publish- The keywords chosen to identify the relevant ing S/Cs for any one technology are aggregated content were developed by the 2022 technologies’ into the “Other” category for clarity). subject matter experts, and are as follows:

134 Table 3. Search keywords summary.

Total # Xplore Technology & indexing terms Boolean search query articles CONTENTS 1. Security Cross-Cutting ((Privacy OR Security OR Intrusion) OR Intrusion OR “Security legislation”) OR (((cyber) OR 12,389 Issues cybersecurity) OR cyber-security) OR “cyber security” 1. INTRODUCTION...... 5 2. Open Intellectual (((“Crowd sourcing”) OR “Open IP”) OR Open AND “Intellectual Property”) OR “Open 1,416 Property Movement standards” 2. SEAMLESS INTELLIGENCE 3. Sustainability ((((“Energy usage”) AND computing)) OR (“Sustainability”) OR (“Green computing”) OR (“Car- 882 SCENARIO...... 8 bon footprint”) OR (“Earth friendly”)) OR (Green ICT)) OR (Sustainable Computing) 3. 23 TECHNOLOGIES IN 2022...... 14 4. Massively Online Open ((((“Open Courses”) OR (“Massively Online”) NOT “Games”) OR “Massively” AND “Courses”) 458 Courses OR “Online learning”) OR “Automated grading” 4. DRIVERS AND DISRUPTORS...... 112 5. Quantum Computing “Quantum Computing”) OR (“Quantum” AND “mechanical phenomena”) OR (“Quantum prop- 2,823 5. TECHNOLOGY COVERAGE IN erties”) OR (“Quantum annealing”) OR (“factorization algorithm” OR “Shor”) OR (“Qubit” IEEE XPLORE AND BY 6. Device and (((“Microelectromechanical systems”) OR “Nano-technology” OR “Nanotechnology” OR “Nano 7,546 IEEE SOCIETIES...... 115 Nano-technology technology”) OR “Microelectromechanical systems”) OR “Micro machine” OR “Micro ma- CONTENTS chines” OR “Micromachines” OR “Micromachine” OR “Micro-machine” OR “Micro-machines” 6. IEEE COMPUTER SOCIETY 7. 3D Integrated Circuits ((((((“2.5D chip” OR “2.5-D chip” OR “2.5D chips” OR “2.5-D chips”))) OR (“3D chip” OR “3-D 1,759 IN 2022...... 120 chip” OR “3D chips” OR “3-D chips”)) OR “System on a Chip”) OR “System in a Package”) 7. SUMMARY AND NEXT STEPS.....124 8. Universal Memory ((((“Non-volatile memory”) OR Memristor) OR “Spin Transfer Torque” RAM) OR “Phase Change 460 Memory”) OR “Universal Memory” 8. AUTHORS...... 126 9. Multicore (((((“Multicore”) OR “Multiprocessor”) OR GPU) OR (“Accelerators”) AND “processor”)) OR 2,276 APPENDIX I. 23 Technologies Coverage GPGPU) OR “Manycore” in IEEE Publications...... 134 10. Photonics ((“Photonics interconnect”) AND “Silicon photonics”) OR VCSEL OR “Vertical Cavity Surface 1,313 Emitting Laser”) 11. Networking and (((“Interconnects”) OR (((“Inter-connectivity”) OR “Interconnectivity”) OR “Inter connectivity”) 2,939 Inter-connectivity AND Networking) OR (“Ethernet”) AND “internet”) OR “Ethernet” AND Networking 12. Software Defined (((((((“Software Defined Networks”) OR “Software defined networking”) OR “Index 496 Networks Terms”:SDN) OR OpenFlow) OR “Software radio”) OR “Active networking”) OR “Virtual Local Area Networks”) OR VLAN 13. High Performance ((((((“High Performance Computing” OR HPC)) OR Supercomputers) OR “Message Passing 1,068 Computing Interface”) OR GPGPU) OR “Compute-intensive”) OR Petascale) OR Exascale 14. Cloud Computing ((((((((Cloud Computing) OR “Grid computing”) OR “Cluster computing”) OR Virtualization) OR 4,252 “-as-a-Service”) OR IaaS) OR PaaS) OR SaaS) OR “Pay as you go” 15. Internet of Things (((((“Internet of Things”) OR “Smart homes”) OR Ubiquity) OR Pervasiveness) OR Interconnec- 442 tivity) OR “Smart dust” 16. Natural User Interfaces (“Natural User Interfaces” OR “NUI”) OR ((“gesture recognition”) OR (“Speech and gesture 3,581 recognition”)) OR (“Graphical user interface” OR “NUI”) OR (“Human Computer Interface” OR “HCI”) OR (“Multi sensor input” OR “Multiple sensor input”) OR (“Augmented reality”) 17. 3D Printing ((((3-D) OR 3D) AND Printing) OR “Additive manufacturing”) OR “Selective laser sintering” 215 18. Big Data and Analytics ((“Big data”) OR “Massive Data”) AND Analytics 42 19. Machine Learning and (((((“Artificial intelligence”) OR “Machine Intelligence”) OR “Intelligent systems”) OR “Machine 13,199 Intelligent Systems Learning”) OR “Supervised learning”) OR “Reinforcement learning” 20. Life Sciences (((((((Bioinformatic) OR Biology) OR Biomedical) OR Biometrics) OR (Health) OR “Health care”) 28,510 OR Healthcare) OR “Life Sciences”) OR Medical) OR Medicine 21. Computational Biology ((((“Computational Biology”) OR Bioinformatics) OR “Structural bioinformatics”) OR Phyloge- 2,145 and Bioinformatics netics and evolutionary modeling) OR Phylogenetics 22. Robotics (Robotics) OR Robot 8,817

135 CONTENTS 1. INTRODUCTION...... 5 Xplore Boolean searches were performed for In the case of co-sponsoring partner S/Cs, each 2. SEAMLESS INTELLIGENCE each technology’s keyword set using the following S/C received an equal pro rata share of the aggre- SCENARIO...... 8 parameters: gate article count for its articles. For example, if a 3. 23 TECHNOLOGIES IN 2022...... 14 Search for: metadata only co-sponsoring partnership of three S/Cs had pub- 4. DRIVERS AND DISRUPTORS...... 112 lished 24 articles of content relevant to the tech- Publisher: IEEE only [inclusive of its S/Cs] nology, each of the three S/Cs would be assigned 5. TECHNOLOGY COVERAGE IN IEEE XPLORE AND BY Content type: Journals and Magazines a share of eight articles. An exception was made IEEE SOCIETIES...... 115 when the count was less than five articles for that Publication years: 2000 through 2013 [for cur- 6.CONTENTS IEEE COMPUTER SOCIETY co-sponsoring group of S/C, to avoid negligible IN 2022...... 120 rency as well as replication ability by excluding per-S/C counts. the newest 2014 content which is added daily] 7. SUMMARY AND NEXT STEPS.....124 [*Note: Xplore is currently limited to 2,000 records 8. AUTHORS...... 126 Various combinations of the search terms were for its export ability. A supplemental process was APPENDIX I. 23 Technologies Coverage assessed to determine the keyword set best iden- used for technologies’ searches resulting in more in IEEE Publications...... 134 tifying the subject matter of interest. The search than 2,000 articles. The periodicals expected to process for each technology is replicable and have a significant number of articles published in documentation is available upon request detailing: the technology were identified until all periodicals the search terms of interest, the number of search with significant involvement were ranked. The hits for the various combinations of keywords article count estimates were then assigned to the tried, the final Boolean aggregate search term sponsoring S/Cs of those titles.] chosen (see above table), and the Xplore URL for the search results. The following pages consist of: The Xplore search results were exported* for fur- • Summary by technology ther analysis. The sponsoring S/Cs for each arti- • Summary by S/C cle’s periodical were identified, except in the case • Sponsoring S/C for each technology, in detail of S/Cs with only one or two articles published • Tabulation of article counts by technology and (they were also assigned to the “Other” category). S/C The number of articles published by each sponsoring S/C, across all of the titles with relevant content, was then tabulated. Finally, if an S/C had a total number of articles too low to be shown on a technology pie chart, its total was also added to the “Other” category.

136 IEEE CS 2022 , Report DRAFT APPENDIX I. 23 Technologies Coverage in IEEE Publications 1/26/2014 5:18 PM

2022 Technologies by CONTENTS Periodicals Articles 1. INTRODUCTION...... 5 2. SEAMLESS INTELLIGENCE SCENARIO...... 8 Open IP, 2% 3. 23 TECHNOLOGIES IN 2022...... 14 3D ICs, 2% Multicore, 3% 4. DRIVERS AND DISRUPTORS...... 112 Quantum 5. TECHNOLOGY COVERAGE IN CCB, 2% Comp, 3% IEEE XPLORE AND BY Ntwk & IEEE SOCIETIES...... 115 Interconnets, Photonics, 1% 6.CONTENTS IEEE COMPUTER SOCIETY 3% Life IN 2022...... 120 NUI, 4% HPC, 1% Sciences, 7. SUMMARY AND NEXT STEPS.....124 27% CC, 5% 8. AUTHORS...... 126 APPENDIX I. 23 Technologies Coverage in IEEE Publications...... 134 Device_Nan o, 8% ML & Intel Sys, 15%

Robotics, Security, 10% 14%

AE=Aerospace AE=Aerospace & Electronic & Electronic Systems Systems s AP=Antenna & AP=Antennas Propagation & Propagation BIO=Biometrics BIO=Biometrics Council Council umer CAS=Circuits & Systems CE=Cons CAS=Circuits & Systems CE=Consumer AE=Aerospace & ElectronicElectronics Electronics Systems AP=Antennas CEDA=Council CEDA=Council & Propagation on on Electronic Electronic ion Design AutomatBIO=Biometrics Design CI=Computational Council Automation CAS=Circuits Intelligence CI=Computational & Systems COMM=Communications CE=Consumer Intelligence COMM=Communications Electronics CPMT=Components, CPMT=Components, Packaging, Packaging, & & Manufacturing Manufacturing Technology Technology CSS=Control Systems CSS=Control Systems EC=Electromagnetic EC=Electromagnetic Compatibility Compatibility Ed=Education Ed=Education CEDA=Council on ElectronicED=Electron ED=Electron Design Automation Devices Devices CI=Computational EMB=Engineering EMB=Engineering Intelligence in y Medicine & Biolog in Medicine COMM=Communications GRS=Geoscience & Biology & Remote CPMT=Components, GRS=Geoscience Sensing & IE=Industrial Electronics Remote Sensing IE=Industrial Electronics Packaging, & Manufac- turing Technology CSS=ControlIM=Instrumentation IM=Instrumentation Systems & EC=Electromagnetic & Measurement Measurement Compatibility IT=Information IT=Information Theory Ed=Education Theory ITS=Intelligent ED=Electron ITS=Intelligent Transportation Devices Transportation EMB=Engineering Systems Systems MAG=Magnetics MAG=Magnetics in Medicine & MTT=Microwave MTT=Microwave Theory Theory & & Techniques Techniques NANO=Nanotechnology NANO=Nanotechnology Council NPS=Nuclear & Council Plasma Sciences NPS=Nuclear OCEAN=Oceanic & Plasma Engineering g Sciences OCEAN=Oceanic Engineerin Biology GRS=GeosciencePE=Power &PE=Power Remote & Sensing & Energy Energy IE=Industrial PHO=Photonics PHO=Photonics Electronics RA=Robotics & Automation IM=Instrumentation RA=Robotics SEN=Sensors & & Automation Measurement Council IT=Information SEN=Sensors SMC=Systems, Council SMC=Systems, Man, & Cybernetics Theory Man, & Cybernetics ITS=Intelligent Transportation Systems SP=Signal MAG=MagneticsSP=Signal Processing Processing MTT=Microwave -‐ SSC=Solid State -‐ SSC=Solid State Circuits Circuits Theory & Techniques SSIT=Social cations Impli SSIT=Social of NANO=Nanotechnology Technology Implications of Technology Council SUPERC=Council NPS=Nuclear on Superconductivity SUPERC=Council on Superconductivity & Plasma Sciences TM=Technology TM=Technology Management Management Council Council UFFC=Ultrasonics, UFFC=Ultrasonics, Ferroelectrics, Ferroelectrics, and Frequency and Control Frequency Control VT=Vehicular Technology VT=Vehicular Technology OCEAN=Oceanic Engineering PE=Power & Energy PHO=Photonics RA=Robotics & Automation SEN=Sensors Council SMC=Systems, Man, & Cyber- netics SP=Signal Processing SSC=Solid-State CircuitsFigure 19 SSIT=Social. The breakdown Implications of 22 technologies of Technology SUPERC=Council by periodical articles. on Superconductivity TM=Technol- ogy Management Council UFFC=Ultrasonics, Ferroelectrics, and Frequency Control VT=Vehicular Technology

Figure 19. The breakdown of 22 technologies131 by periodical articles.

137 IEEE CS 2022 , Report DRAFT APPENDIX I. 23 Technologies Coverage in IEEE Publications 1/26/2014 5:18 PM

2022 Technologies by CONTENTS Sponsoring Societies 1. INTRODUCTION...... 5 2. SEAMLESS INTELLIGENCE IEEE, 3% SCENARIO...... 8 PHO, 3% 3. 23 TECHNOLOGIES IN 2022...... 14 CI, 4% 4. DRIVERS AND DISRUPTORS...... 112 NPS, 4% 5. TECHNOLOGY COVERAGE IN IEEE XPLORE AND BY IEEE SOCIETIES...... 115 Computer, COMM, 5% 6.CONTENTS IEEE COMPUTER SOCIETY 28% IN 2022...... 120 OTHER, 7. SUMMARY AND NEXT STEPS.....124 8. AUTHORS...... 126 6% APPENDIX I. 23 Technologies Coverage in IEEE Publications...... 134 SP, 7%

EMB, NANO, 16% 11%

RA, 13%

AE=Aerospace & Electronic Systems AP=Antennas & Propagation BIO=Biometrics Council CAS=Circuits & Systems CE=Consumer Electronics

CEDA=Council on Electronic Design Automation CI=Computational Intelligence COMM=Communications CPMT=Components, Packaging, & Manufac- turing Technology CSS=Control Systems EC=Electromagnetic Compatibility Ed=Education ED=Electron Devices EMB=Engineering in Medicine & Biology GRS=Geoscience & Remote Sensing IE=Industrial Electronics IM=Instrumentation & Measurement IT=Information Theory ITS=Intelligent Transportation Systems MAG=Magnetics MTT=Microwave Theory & Techniques NANO=Nanotechnology Council NPS=Nuclear & Plasma Sciences OCEAN=Oceanic Engineering PE=Power & Energy PHO=Photonics RA=Robotics & Automation SEN=Sensors Council SMC=Systems, Man, & Cyber- netics SP=Signal Processing SSC=Solid-StateFigure Circuits20 . The SSIT=Social breakdown Implications of 22 technologies of Technology by sponsoring SUPERC=Council societies. on Superconductivity TM=Technol- ogy Management Council UFFC=Ultrasonics, Ferroelectrics, and Frequency Control VT=Vehicular Technology

Figure 20. The breakdown of 22 technologies by sponsoring societies. 132 138 IEEE CS 2022 , Report DRAFT APPENDIX I. 23 Technologies Coverage in IEEE Publications 1/26/2014 5:18 PM

1 Security Cross-cutting Issues Sponsoring Societies CONTENTS AE CE 1% 1. INTRODUCTION...... 5 2% VT BIO 2. SEAMLESS INTELLIGENCE 1% SCENARIO...... 8 2% IT 3. 23 TECHNOLOGIES IN 2022...... 14 3% IEEE 4. DRIVERS AND DISRUPTORS...... 112 4% 5. TECHNOLOGY COVERAGE IN IEEE XPLORE AND BY OTHER IEEE SOCIETIES...... 115 5% 6.CONTENTS IEEE COMPUTER SOCIETY IN 2022...... 120 Compute 7. SUMMARY AND NEXT STEPS.....124 PE r 8. AUTHORS...... 126 9% APPENDIX I. 23 Technologies Coverage 43% in IEEE Publications...... 134

SP 15%

COMM 15%

CEDA=Council on ElectronicCPMT=Components, Design Automation Packaging, CI=Computational & Manufacturing Intelligence Technology COMM=Communications CSS=Control Systems CPMT=Components, EC=Electromagnetic Packaging, Compatibility Ed=Education & Manufac- ED=Electron Devices EMB=Engineering in Medicine & Biology GRS=Geoscience ndustrial & Remote Sensing IE=I Electronics turing Technology CSS=Control IM=Instrumentation Systems EC=Electromagnetic & Measurement Compatibility IT=Information Ed=Education Theory ED=Electron ITS=Intelligent Devices EMB=Engineering Transportation Systems MAG=Magnetics in Medicine & Biology GRS=Geoscience MTT=Microwave & Remote Sensing Theory IE=Industrial & Techniques Electronics IM=Instrumentation NANO=Nanotechnology & Council Measurement IT=Information NPS=Nuclear nic & Plasma Sciences OCEAN=Ocea Engineering Theory ITS=Intelligent PE=Power & Energy PHO=Photonics RA=Robotics & Automation SEN=Sensors Council SMC=Systems, Man, & Cybernetics Transportation Systems MAG=MagneticsSP=Signal Processing MTT=Microwave -‐ SSC=Solid State Circuits Theory & Techniques SSIT=Social NANO=Nanotechnology Implications of Technology Council SUPERC=Council on Superconductivity NPS=Nuclear & Plasma Sciences OCEAN=Oceanic EngineeringTM=Technology PE=Power & Management Energy PHO=Photonics Council RA=Robotics UFFC=Ultrasonics, & Automation Ferroelectrics, SEN=Sensors and Frequency Council Control SMC=Systems, VT=Vehicular Technology Man, & Cyber- netics SP=Signal Processing SSC=Solid-StateFigure Circuits21 . The SSIT=Social breakdown Implications of security cross of Technology-‐cutting issues SUPERC=Council by sponsoring on societies. Superconductivity TM=Technol-

ogy Management Council UFFC=Ultrasonics, Ferroelectrics, and Frequency Control VT=Vehicular Technology

Figure 21. The breakdown of security cross-cutting133 issues by sponsoring societies. 139 IEEE CS 2022 , Report DRAFT APPENDIX I. 23 Technologies Coverage in IEEE Publications 1/26/2014 5:18 PM

2 Open Intellectual Property CONTENTS Sponsoring Societies 1. INTRODUCTION...... 5 NPS GRS SP 2. SEAMLESS INTELLIGENCE 2% 2% CE 2% SCENARIO...... 8 CSS 3% 3. 23 TECHNOLOGIES IN 2022...... 14 IEEE3% 4. DRIVERS AND DISRUPTORS...... 112 3% PE 5. TECHNOLOGY COVERAGE IN IEEE XPLORE AND BY 3% IEEE SOCIETIES...... 115 MTT OTHER 6.CONTENTS IEEE COMPUTER SOCIETY 3% 36% IN 2022...... 120 IM 7. SUMMARY AND NEXT STEPS.....124 3% 8. AUTHORS...... 126 APPENDIX I. 23 Technologies Coverage in IEEE Publications...... 134

COMM 19%

Computer 21%

AE=Aerospace & Electronic Systems AP=Antennas & Propagation BIO=Biometrics Council CAS=Circuits & Systems CE=Consumer AE=Aerospace & Electronic Electronics Systems AP=Antennas CEDA=Council & Propagation on Electronic BIO=Biometrics Design Automation Council CAS=Circuits CI=Computational & Systems CE=Consumer Intelligence COMM=Communications Electronics CEDA=Council on Electronic CPMT=Compon Design Automationents, Packaging, CI=Computational & Manufacturing Intelligence Technology COMM=Communications CSS=Control Systems CPMT=Components, EC=Electromagnetic Packaging, Compatibility Ed=Education & Manufac- ED=Electron Devices EMB=Engineering in Medicine & Biology GRS=Geoscience & Remote Sensing IE=Industrial Electronics turing Technology CSS=ControlIM=Instrumentation Systems & EC=Electromagnetic Measurement Compatibility IT=Information Ed=Education Theory ED=Electron ITS=Intelligent Devices Transportation EMB=Engineering Systems MAG=Magnetics in Medicine & Biology GRS=GeoscienceMTT=Microwave & Remote Sensing Theory IE=Industrial & Techniques Electronics IM=Instrumentation NANO=Nanotechnology Council & Measurement NPS=Nuclear IT=Information & Theory Plasma g Sciences OCEAN=Oceanic Engineerin ITS=Intelligent PE=Power & Energy PHO=Photonics RA=Robotics & Automation SEN=Sensors Council SMC=Systems, Man, & Cybernetics Transportation Systems SP=Signal MAG=Magnetics Processing MTT=Microwave -‐ SSC=Solid State Circuits Theory & Techniques SSIT=Social NANO=Nanotechnology Implications of Technology Council NPS=Nuclear SUPERC=Council on Superconductivity & Plasma Sciences OCEAN=Oceanic EngineeringTM=Techno PE=Powerlogy Management & Energy PHO=Photonics Council UFFC=Ultrasonics, RA=Robotics & Ferroelectrics, Automation SEN=Sensors and Frequency Council Control VT=Vehicular Technology SMC=Systems, Man, & Cyber- netics SP=Signal Processing SSC=Solid-StateFigure Circuits22 . SSIT=Social The breakdown Implications of open intellectual of Technology property SUPERC=Councilby sponsoring on Superconductivity societies. TM=Technol- ogy Management Council UFFC=Ultrasonics, Ferroelectrics, and Frequency Control VT=Vehicular Technology

Figure 22. The breakdown of open intellectual property by sponsoring societies. 134 140 IEEE CS 2022 , Report DRAFT APPENDIX I. 23 Technologies Coverage in IEEE Publications 1/26/2014 5:18 PM

3 Sustainability CONTENTS Sponsoring Socie8es 1. INTRODUCTION...... 5 SSIT IEEE 2. SEAMLESS INTELLIGENCE 3% SCENARIO...... 8 NPS 3% 4% 3. 23 TECHNOLOGIES IN 2022...... 14 MTT 4. DRIVERS AND DISRUPTORS...... 112 4%

5. TECHNOLOGY COVERAGE IN GRS IEEE XPLORE AND BY 5% Other IEEE SOCIETIES...... 115 35% 6.CONTENTS IEEE COMPUTER SOCIETY IN 2022...... 120 PE 7. SUMMARY AND NEXT STEPS.....124 7% 8. AUTHORS...... 126 APPENDIX I. 23 Technologies Coverage in IEEE Publications...... 134 COMM 7%

AP 9% Computer 23%

AE=Aerospace & Electronic AE=Aerospace Systems & AP=Antennas Electronic & Systems AP=Antennas Propagation & BIO=Biometrics Propagation Council BIO=Biometrics CAS=Circuits Council & Systems CE=Consumer CAS=Circuits & Systems CE=Consumer Electronics Electronics CEDA=Council on Electronic Design Automation CI=Computational Intelligence COMM=Communications CEDA=Council on Electronic CPMT=Components, Design Automation Packaging, CI=Computational & Manufacturing Technology Intelligence COMM=Communications CSS=Control Systems CPMT=Components, EC=Electromagnetic Packaging, Compatibility Ed=Education & Manufac- ED=Electron Devices EMB=Engineering in Medicine & Biology GRS=Geoscience & Remote Sensing IE=Industrial Electronics turing Technology CSS=ControlIM=Instrumentation Systems & EC=Electromagnetic Measurement =Information IT Compatibility Theory Ed=Education ITS=Intelligent ED=Electron Transportation Devices Systems EMB=Engineering MAG=Magnetics in Medicine & Biology GRS=GeoscienceMTT=Microwave & Remote Sensing Theory IE=Industrial & Techniques Electronics IM=Instrumentation NANO=Nanotechnology Council & Measurement NPS=Nuclear IT=Information & Theory Plasma g Sciences OCEAN=Oceanic Engineerin ITS=Intelligent PE=Power & Energy PHO=Photonics RA=Robotics & Automation SEN=Sensors Council SMC=Systems, Man, & Cybernetics Transportation Systems SP=Signal MAG=Magnetics Processing MTT=Microwave -‐ SSC=Solid State Circuits Theory & Techniques SSIT=Social NANO=Nanotechnology Implications of Technology Council NPS=Nuclear SUPERC=Council on Superconductivity & Plasma Sciences OCEAN=Oceanic EngineeringTM=Technology PE=Power Management & Energy Council PHO=Photonics UFFC=Ultrasonics, RA=Robotics Ferroelectrics, & and Automation Frequency SEN=Sensors Control Council VT=Vehicular Technology SMC=Systems, Man, & Cyber- Figure 23. The breakdown of sustainability by sponsoring societies. netics SP=Signal Processing SSC=Solid-State Circuits SSIT=Social Implications of Technology SUPERC=Council on Superconductivity TM=Technol- ogy Management Council UFFC=Ultrasonics, Ferroelectrics, and Frequency Control VT=Vehicular Technology

Figure 23. The breakdown of sustainability135 by sponsoring societies. 141 IEEE CS 2022 , Report DRAFT APPENDIX I. 23 Technologies Coverage in IEEE Publications 1/26/2014 5:18 PM

4 MOOC CONTENTS Sponsoring Societies 1. INTRODUCTION...... 5 2. SEAMLESS INTELLIGENCE IEEE! CSS! IT! SCENARIO...... 8 2%! 3%! 3%! 3. 23 TECHNOLOGIES IN 2022...... 14 EMBS! CI! 4. DRIVERS AND DISRUPTORS...... 112 CAS!4%! 20%! 5. TECHNOLOGY COVERAGE IN 3%! IEEE XPLORE AND BY IEEE SOCIETIES...... 115 COMM! 6.CONTENTS IEEE COMPUTER SOCIETY 4%! IN 2022...... 120 Other! 7. SUMMARY AND NEXT STEPS.....124 4%! 8. AUTHORS...... 126 IE! APPENDIX I. 23 Technologies Coverage 5%! in IEEE Publications...... 134 Computer! 18%! SMC! 8%!

SP! 10%! Ed! 16%!

AE=Aerospace & ElectronicAE=Aerospace Systems AP=Antennas & Electronic & Propagation Systems BIO=Biometrics AP=Antennas Council & Propagation ouncil BIO=Biometrics CCAS=Circuits & CAS=Circuits Systems CE=Consumer & Systems CE=Consumer Electronics Electronics CEDA=Council on Electronic Design Automation CI=Computational Intelligence COMM=Communications CEDA=Council on Electronic CPMT=Components, Design Automation Packaging, CI=Computational & Manufacturing Intelligence Technology COMM=Communications CSS=Control Systems EC=Electromagnetic CPMT=Components, Compatibility Packaging, Ed=Education & Manufac- turing Technology CSS=Control ED=Electron Systems Devices EC=Electromagnetic EMB=Engineering Compatibility in Medicine Ed=Education & Biology ED=Electron Devices GRS=Geoscience EMB=Engineering & Remote Sensing IE=Industrial Electronics in Medicine & IM=Instrumentation & Measurement IT=Information Theory ortation ITS=Intelligent Transp Systems MAG=Magnetics Biology GRS=GeoscienceMTT=Microwave & Remote Sensing Theory IE=Industrial & Techniques Electronics IM=Instrumentation NANO=Nanotechnology & Council Measurement IT=Information NPS=Nuclear & Theory Plasma g Sciences OCEAN=Oceanic Engineerin ITS=Intelligent Transportation Systems MAG=MagneticsPE=Power & Energy MTT=Microwave PHO=Photonics Theory & Techniques RA=Robotics NANO=Nanotechnology & Automation SEN=Sensors Council SMC=Systems, Council Man, NPS=Nuclear & Cybernetics & Plasma Sciences SP=Signal Processing -‐ SSC=Solid State Circuits SSIT=Social Implications of Technology SUPERC=Council on Superconductivity OCEAN=Oceanic EngineeringTM=Technology PE=Power Management & Energy PHO=Photonics Council RA=Robotics UFFC=Ultrasonics, & Automation Ferroelectrics, and Frequency Control SEN=Sensors VT=Vehicular Council SMC=Systems, Technology Man, & Cyber-

netics SP=Signal Processing SSC=Solid-State Circuits SSIT=SocialFigure 24 . Implications The breakdown of Technology of MOOC by sponsoring SUPERC=Council societies. on Superconductivity TM=Technol- ogy Management Council UFFC=Ultrasonics, Ferroelectrics, and Frequency Control VT=Vehicular Technology

Figure 24. The breakdown of MOOC by sponsoring societies. 136 142 IEEE CS 2022 , Report DRAFT APPENDIX I. 23 Technologies Coverage in IEEE Publications 1/26/2014 5:18 PM

5 Quantum Compu8ng Sponsoring Socie8es CONTENTS Computer 1. INTRODUCTION...... 5 IEEE 2. SEAMLESS INTELLIGENCE 2% Other SCENARIO...... 8 3% 3. 23 TECHNOLOGIES IN 2022...... 14 MAG 6% 4. DRIVERS AND DISRUPTORS...... 112 5. TECHNOLOGY COVERAGE IN 4% IEEE XPLORE AND BY NPS IEEE SOCIETIES...... 115 4% 6.CONTENTS IEEE COMPUTER SOCIETY IN 2022...... 120 NANO 7. SUMMARY AND NEXT STEPS.....124 6% 8. AUTHORS...... 126 PHO APPENDIX I. 23 Technologies Coverage 51% in IEEE Publications...... 134 IT 7% ED 8% SUPERC 9%

AE=Aerospace & Electronic Systems AP=Antennas & Propagation BIO=Biometrics Council umer CAS=Circuits & Systems CE=Cons AE=Aerospace & Electronic Electro Systemsnics AP=Antennas CEDA=Council & Propagation on Electronic BIO=Biometrics Design Automation Council CAS=Circuits CI=Computational & Systems CE=Consumer Intelligence COMM=Communications Electronics CEDA=Council on Electronic CPMT=Components, Design Automation Packaging, CI=Computational & Manufacturing Intelligence Technology COMM=Communications CSS=Control Systems CPMT=Components, EC=Electromagnetic Compatibility Ed=Education Packaging, & Manufac- ED=Electron Devices EMB=Engineering in Medicine & Biology GRS=Geoscience & Remote Sensing IE=Industrial Electronics turing Technology CSS=ControlIM=Instrumentation Systems & EC=Electromagnetic Measurement Compatibility IT=Information Ed=Education Theory ED=Electron ITS=Intelligent Devices Transportation EMB=Engineering Systems MAG=Magnetics in Medicine & Biology GRS=GeoscienceMTT=Microwave & Remote Sensing Theory & IE=Industrial Techniques Electronics NANO=Nanotechnology IM=Instrumentation Council & Measurement NPS=Nuclear IT=Information & Plasma Theory Sciences OCEAN=Oceanic Engineering ITS=Intelligent PE=Power & Energy PHO=Photonics RA=Robotics & Automation SEN=Sensors Council SMC=Systems, Man, & Cybernetics Transportation Systems SP=Signal MAG=Magnetics Processing SSC=Solid MTT=Microwave-‐State Circuits Theory & Techniques SSIT=Social NANO=Nanotechnology Implications of Technology Council NPS=Nuclear SUPERC=Council on Superconductivity & Plasma Sciences OCEAN=Oceanic EngineeringTM=Technology PE=Power Management & Energy PHO=Photonics Council UFFC=Ultrasonics, RA=Robotics & Ferroelectrics, Automation SEN=Sensors and Frequency Council Control VT=Vehicular Technology SMC=Systems, Man, & Cyber- netics SP=Signal Processing SSC=Solid-State CircuitsFigure 25 SSIT=Social. The breakdown Implications of of Technology quantum computing by SUPERC=Council sponsoring on societies. Superconductivity TM=Technol-

ogy Management Council UFFC=Ultrasonics, Ferroelectrics, and Frequency Control VT=Vehicular Technology

Figure 25. The breakdown of quantum137 computing by sponsoring societies. 143 IEEE CS 2022 , Report DRAFT APPENDIX I. 23 Technologies Coverage in IEEE Publications 1/26/2014 5:18 PM

6 Device and Nano-tech. CONTENTS Sponsoring Societies

1. INTRODUCTION...... 5 Computer 1% 2. SEAMLESS INTELLIGENCE IM SCENARIO...... 8 1% OTHER 3. 23 TECHNOLOGIES IN 2022...... 14 10% IEEE 4. DRIVERS AND DISRUPTORS...... 112 2% CAS 5. TECHNOLOGY COVERAGE IN 1% IEEE XPLORE AND BY MAG NANO IEEE SOCIETIES...... 115 3% 29% 6.CONTENTS IEEE COMPUTER SOCIETY MTT IN 2022...... 120 4% 7. SUMMARY AND NEXT STEPS.....124 8. AUTHORS...... 126 PHO 6% APPENDIX I. 23 Technologies Coverage in IEEE Publications...... 134 SEN 3%

RA ED 12% 16% IE 12%

AE=Aerospace & Electronic AE=Aerospace Systems AP=Antennas & Electronic & Propagation Systems BIO=Biometrics AP=Antennas Council & Propagation CAS=Circuits & Systems BIO=Biometrics CE=Consumer Council umer CAS=Circuits & Systems CE=Cons Electronics Electronics CEDA=Council on Electronic Automation Design CI=Computational Intelligence COMM=Communications CEDA=Council on Electronic CPMT=Components, Design Automation Packaging, CI=Computational & Manufacturing Intelligence Technology COMM=Communications CSS=Control Systems CPMT=Components, EC=Electromagnetic Packaging, Compatibility Ed=Education & Manufac- ED=Electron Devices EMB=Engineering in Medicine & Biology GRS=Geoscience & Remote Sensing IE=Industrial Electronics turing Technology CSS=ControlIM=Instrumentation Systems EC=Electromagnetic & Measurement Compatibility IT=Information Ed=Education Theory ED=Electron ITS=Intelligent Devices EMB=Engineering Transportation Systems MAG=Magnetics in Medicine & Biology GRS=Geoscience MTT=Microwave & Remote Sensing Theory IE=Industrial & Techniques Electronics IM=Instrumentation NANO=Nanotechnology ncil Cou & NPS=Nuclear Measurement & Plasma IT=Information Sciences Theory OCEAN=Oceanic Engineering ITS=Intelligent PE=Power & Energy PHO=Photonics RA=Robotics & Automation SEN=Sensors Council SMC=Systems, Man, & Cybernetics Transportation Systems MAG=MagneticsSP=Signal Processing MTT=Microwave -‐ SSC=Solid State Circuits Theory & Techniques l SSIT=Socia Implications NANO=Nanotechnology of Technology Council SUPERC=Council NPS=Nuclear on Superconductivity & Plasma Sciences OCEAN=Oceanic EngineeringTM=Technology PE=Power & Management Energy PHO=Photonics Council RA=Robotics UFFC=Ultrasonics, & Automation Ferroelectrics, SEN=Sensors and Frequency Council Control SMC=Systems, VT=Vehicular Technology Man, & Cyber- netics SP=Signal Processing SSC=Solid-StateFigure Circuits26 . SSIT=Social The breakdown Implications of device of d an Technology nano-‐technology SUPERC=Council by sponsoring on Superconductivity societies. TM=Technol-

ogy Management Council UFFC=Ultrasonics, Ferroelectrics, and Frequency Control VT=Vehicular Technology

Figure 26. The breakdown of device and nano-technology138 by sponsoring societies. 144 IEEE CS 2022 , Report DRAFT APPENDIX I. 23 Technologies Coverage in IEEE Publications 1/26/2014 5:18 PM

7 3D Integrated Circuits CONTENTS Sponsoring Socie8es 1. INTRODUCTION...... 5 2. SEAMLESS INTELLIGENCE CE IM 3% SCENARIO...... 8 IEEE 4% 3. 23 TECHNOLOGIES IN 2022...... 14 4% SSC MTT 4. DRIVERS AND DISRUPTORS...... 112 4% 20% 5. TECHNOLOGY COVERAGE IN IEEE XPLORE AND BY Other IEEE SOCIETIES...... 115 5% 6.CONTENTS IEEE COMPUTER SOCIETY IN 2022...... 120 7. SUMMARY AND NEXT STEPS.....124 ED 8. AUTHORS...... 126 8% APPENDIX I. 23 Technologies Coverage in IEEE Publications...... 134 Computer 19%

CAS 14%

CEDA 19%

AE=Aerospace & Electronic Systems AP=Antennas & Propagation BIO=Biometrics Council umer CAS=Circuits & Systems CE=Cons AE=Aerospace & Electronic SystemsElectronics AP=Antennas CEDA=Council & Propagation on Electronic BIO=Biometrics Design Council Automation CAS=Circuitstelligence CI=Computational In & Systems COMM=Communications CE=Consumer Electronics CPMT=Components, Packaging, & Manufacturing Technology CSS=Control Systems EC=Electromagnetic Compatibility Ed=Education CEDA=Council on Electronic DesignED=Electron Automation Devices CI=Computational EMB=Engineering Intelligence in Medicine COMM=Communications & Biology te GRS=Geoscience & Remo Sensing CPMT=Components, IE=Industrial Packaging, Electronics & Manufac- turing Technology CSS=ControlIM=Instrumentation Systems EC=Electromagnetic & Measurement Compatibility IT=Information Ed=Education Theory ED=Electron ITS=Intelligent Devices EMB=Engineering Transportation Systems MAG=Magnetics in Medicine & MTT=Microwave Theory & Techniques NANO=Nanotechnology Council nces NPS=Nuclear & Plasma Scie OCEAN=Oceanic Engineering Biology GRS=Geoscience & RemotePE=Power Sensing & Energy IE=Industrial PHO=Photonics Electronics IM=Instrumentation RA=Robotics & Measurement & Automation IT=Information SEN=Sensors Theory Council SMC=Systems, Man, & Cybernetics ITS=Intelligent Transportation Systems MAG=MagneticsSP=Signal Processing MTT=Microwave -‐ SSC=Solid State Theory Circuits & Techniques SSIT=Social NANO=Nanotechnology Implications of Technology SUPERC=Council Council on NPS=Nuclear Superconductivity & Plasma Sciences OCEAN=Oceanic Engineering TM=Technology PE=Power & Energy Management PHO=Photonics Council RA=Robotics UFFC=Ultrasonics, & Automation Ferroelectrics, SEN=Sensors and Frequency Council Control SMC=Systems, VT=Vehicular Technology Man, & Cyber- netics SP=Signal Processing SSC=Solid-State CircuitsFigure 27 SSIT=Social. The breakdown Implications3D of of integrated Technology circuits SUPERC=Councilby sponsoring on Superconductivity societies. TM=Technol- ogy Management Council UFFC=Ultrasonics, Ferroelectrics, and Frequency Control VT=Vehicular Technology

Figure 27. The breakdown of 3D integrated139 circuits by sponsoring societies.

145 IEEE CS 2022 , Report DRAFT APPENDIX I. 23 Technologies Coverage in IEEE Publications 1/26/2014 5:18 PM

8 Universal Memory CONTENTS Sponsoring Societies 1. INTRODUCTION...... 5 2. SEAMLESS INTELLIGENCE CEDA SCENARIO...... 8 NANO 3% 3. 23 TECHNOLOGIES IN 2022...... 14 5% IEEE 4. DRIVERS AND DISRUPTORS...... 112 5% 5. TECHNOLOGY COVERAGE IN CE ED IEEE XPLORE AND BY 27% IEEE SOCIETIES...... 115 5% 6.CONTENTS IEEE COMPUTER SOCIETY IN 2022...... 120 IT 7. SUMMARY AND NEXT STEPS.....124 6% 8. AUTHORS...... 126 APPENDIX I. 23 Technologies Coverage SSC in IEEE Publications...... 134 7%

MAG CAS 7% 10% NPS OTHER 7% Computer 10% 8%

turing Technology CSS=ControlED=Electron Systems Devices EC=Electromagnetic EMB=Engineering Compatibility in Medicine Ed=Education & Biology ED=Electron Devices GRS=Geoscience EMB=Engineering & cs Remote Sensing IE=Industrial Electroni in Medicine & IM=Instrumentation & Measurement IT=Information Theory ITS=Intelligent Transportation Systems MAG=Magnetics Biology GRS=GeoscienceMTT=Microwave & Remote Sensing Theory IE=Industrial & Techniques Electronics IM=Instrumentation NANO=Nanotechnology & Council Measurement IT=Information NPS=Nuclear & Theory Plasma g Sciences OCEAN=Oceanic Engineerin ITS=Intelligent Transportation Systems MAG=MagneticsPE=Power & Energy MTT=Microwave PHO=Photonics Theory & Techniques RA=Robotics NANO=Nanotechnology & Automation Council NPS=Nuclear SEN=Sensors Council SMC=Systems, Man, & Cybernetics & Plasma Sciences SP=Signal Processing -‐ SSC=Solid State Circuits SSIT=Social Implications of Technology SUPERC=Council on Superconductivity OCEAN=Oceanic EngineeringTM=Technology PE=Power Management & Energy PHO=Photonics Council RA=Robotics UFFC=Ultrasonics, & Automation Ferroelectrics, SEN=Sensors and Frequency Council Control VT=Vehicular TechnologySMC=Systems, Man, & Cyber- netics SP=Signal Processing SSC=Solid-State CircuitsFigure SSIT=Social28 . The Implications breakdown of Technology of universal SUPERC=Councilby memory sponsoring on societies. Superconductivity TM=Technol- ogy Management Council UFFC=Ultrasonics, Ferroelectrics, and Frequency Control VT=Vehicular Technology

Figure 28. The breakdown of universal140 memory by sponsoring societies.

146 IEEE CS 2022 , Report DRAFT APPENDIX I. 23 Technologies Coverage in IEEE Publications 1/26/2014 5:18 PM

9 Multicore

CONTENTS Sponsoring SocietiesGRS MAG 1% 1. INTRODUCTION...... 5 IEEE 2% AP 2. SEAMLESS INTELLIGENCE 2% CE NPS 1% 1% SCENARIO...... 8 2% 3. 23 TECHNOLOGIES IN 2022...... 14 COMM 4. DRIVERS AND DISRUPTORS...... 112 SP 3% 5. TECHNOLOGY COVERAGE IN 3% IEEE XPLORE AND BY PHO IEEE SOCIETIES...... 115 3% SSC 6.CONTENTS IEEE COMPUTER SOCIETY 3% IN 2022...... 120 7. SUMMARY AND NEXT STEPS.....124 CAS 8. AUTHORS...... 126 4% APPENDIX I. 23 Technologies Coverage in IEEE Publications...... 134 Computer CEDA 56% 9%

OTHER 10%

AE=Aerospace & ElectronicAE=Aerospace Systems & AP=Antennas Electronic Systems AP=Antennas & Propagation & Propagation BIO=Biometrics Council BIO=Biometrics CAS=Circuits Council & Systems CE=Consumer CAS=Circuits & Systems CE=Consumer Electronics Electronics CEDA=Council on Electronic Design Automation CI=Computational Intelligence COMM=Communications CEDA=Council on Electronic CPMT=Components, Design Automation Packaging, CI=Computational & Manufacturing Technology Intelligence CSS=Control COMM=Communications Systems CPMT=Components, EC=Electromagnetic Compatibility Packaging, Ed=Education & Manufac- turing Technology CSS=ControlED=Electron Systems Devices EC=Electromagnetic EMB=Engineering Compatibility in Medicine Ed=Education & Biology ED=Electron GRS=Geoscience Devices & EMB=Engineering Remote Sensing IE=Industrial Electronics in Medicine & IM=Instrumentation & Measurement IT=Information Theory ITS=Intelligent Transportation Systems MAG=Magnetics Biology GRS=GeoscienceMTT=Microwave & Remote Theory Sensing & IE=Industrial Techniques Electronics NANO=Nanotechnology IM=Instrumentation Council & Measurement NPS=Nuclear IT=Information & Plasma g Sciences OCEAN=Oceanic EngineerinTheory ITS=Intelligent Transportation SystemsPE=Power MAG=Magnetics & Energy MTT=Microwave PHO=Photonics RA=Robotics Theory & & Automation Techniques NANO=Nanotechnology SEN=Sensors Council Council NPS=Nuclear SMC=Systems, Man, & Cybernetics & Plasma Sciences SP=Signal Processing -‐ SSC=Solid State Circuits SSIT=Social Implications of Technology SUPERC=Council on Superconductivity OCEAN=Oceanic EngineeringTM=Technology PE=Power Management & Energy Council PHO=Photonics UFFC=Ultrasonics, Ferroelectrics, RA=Robotics and & Automation Frequency Control SEN=Sensors Council VT=Vehicular Technology SMC=Systems, Man, & Cyber-

netics SP=Signal Processing SSC=Solid-State CircuitsFigure 29 SSIT=Social. The breakdown Implicationsmuticore of of Technology by sponsoring SUPERC=Council societies. on Superconductivity TM=Technol- ogy Management Council UFFC=Ultrasonics, Ferroelectrics, and Frequency Control VT=Vehicular Technology

Figure 29. The breakdown of 141muticore by sponsoring societies. 147 IEEE CS 2022 , Report DRAFT APPENDIX I. 23 Technologies Coverage in IEEE Publications 1/26/2014 5:18 PM

10 Photonics

Sponsoring SocietiesIEEE CONTENTS CPMT SSC NPS 1% 1% 1% Computer 1% 1. INTRODUCTION...... 5 UFFC 0% 2% 2. SEAMLESS INTELLIGENCE AE SCENARIO...... 8 IM 2% 2% 3. 23 TECHNOLOGIES IN 2022...... 14 OTHER 4. DRIVERS AND DISRUPTORS...... 112 2% ED 5. TECHNOLOGY COVERAGE IN 3% IEEE XPLORE AND BY MTT IEEE SOCIETIES...... 115 3% 6.CONTENTS IEEE COMPUTER SOCIETY COMM IN 2022...... 120 3% 7. SUMMARY AND NEXT STEPS.....124 8. AUTHORS...... 126 APPENDIX I. 23 Technologies Coverage in IEEE Publications...... 134

PHO 79%

AE=Aerospace & Electronic Systems AP=Antennas & Propagation l BIO=Biometrics Counci CAS=Circuits & Systems CE=Consumer AE=Aerospace & Electronic Electronics Systems AP=Antennas CEDA=Council & Propagation on Electronic BIO=Biometrics Design Automation Council CAS=Circuits CI=Computational & Systems CE=Consumer Intelligence COMM=Communications Electronics CEDA=Council on Electronic CPMT=Components, Design Automation Packaging, CI=Computational & Manufacturing Intelligence Technology COMM=Communications CSS=Control Systems EC=Electromagnetic CPMT=Components, Compatibility Packaging, Ed=Education & Manufac- ED=Electron Devices EMB=Engineering in Medicine & Biology GRS=Geoscience & Remote Sensing IE=Industrial Electronics turing Technology CSS=ControlIM=Instrumentation Systems EC=Electromagnetic & Measurement Compatibility IT=Information Ed=Education Theory ED=Electron ITS=Intelligent Transportation Systems Devices EMB=Engineering MAG=Magnetics in Medicine & Biology GRS=GeoscienceMTT=Microwave & Remote Sensing Theory IE=Industrial & Techniques Electronics IM=Instrumentation NANO=Nanotechnology & Council Measurement IT=Information NPS=Nuclear & Theory Plasma g Sciences OCEAN=Oceanic Engineerin ITS=Intelligent PE=Power & Energy PHO=Photonics RA=Robotics & Automation SEN=Sensors Council SMC=Systems, Man, & Cybernetics Transportation Systems MAG=MagneticsSP=Signal Processing MTT=Microwave -‐ SSC=Solid State Circuits Theory & Techniques SSIT=Social NANO=Nanotechnology Implications of Technology Council NPS=Nuclear SUPERC=Council on Superconductivity & Plasma Sciences OCEAN=Oceanic EngineeringTM=Technology PE=Power Management & Energy PHO=Photonics Council RA=Robotics UFFC=Ultrasonics, & Automation Ferroelectrics, SEN=Sensors and Frequency Control VT=Vehicular Council Technology SMC=Systems, Man, & Cyber- netics SP=Signal Processing SSC=Solid-State Circuits Figure SSIT=Social 30. The Implications breakdown of of photonicsTechnology by SUPERC=Council sponsoring societies. on Superconductivity TM=Technol-

ogy Management Council UFFC=Ultrasonics, Ferroelectrics, and Frequency Control VT=Vehicular Technology

Figure 30. The breakdown of photonics142 by sponsoring societies. 148 IEEE CS 2022 , Report DRAFT APPENDIX I. 23 Technologies Coverage in IEEE Publications 1/26/2014 5:18 PM

11 Networking & Interconnectivity Sponsoring Societies CONTENTS 1. INTRODUCTION...... 5 2. SEAMLESS INTELLIGENCE SCENARIO...... 8 OTHER 3. 23 TECHNOLOGIES IN 2022...... 14 COMM 12% 4. DRIVERS AND DISRUPTORS...... 112 IEEE 17% 5. TECHNOLOGY COVERAGE IN 2% IEEE XPLORE AND BY NPS CONTENTSIEEE SOCIETIES...... 115 4% 6. IEEE COMPUTER SOCIETY CAS IN 2022...... 120 3% 7. SUMMARY AND NEXT STEPS.....124 SSC PHO 8. AUTHORS...... 126 5% APPENDIX I. 23 Technologies Coverage 15% in IEEE Publications...... 134 MTT 6%

CEDA 5% CPMT 10% Computer 9% ED 12%

AE=Aerospace & Electronic Systems AP=Antennas & Propagation BIO=Biometrics Council umer CAS=Circuits & Systems CE=Cons AE=Aerospace & Electronic Electro Systemsnics AP=Antennas CEDA=Council & Propagation on Electronic BIO=Biometrics Design Automation Council CAS=Circuits CI=Computational & Systems CE=Consumer Intelligence COMM=Communications Electronics CPMT=Components, Packaging, & Manufacturing Technology CSS=Control Systems EC=Electromagnetic Compatibility Ed=Education CEDA=Council on ElectronicED=Electro Design n Automation Devices CI=Computational EMB=Engineering Intelligence in Medicine COMM=Communications & Biology CPMT=Components, GRS=Geoscience & Remote Sensing IE=Industrial Electronics Packaging, & Manufac- turing Technology CSS=ControlIM=Instrumentation Systems EC=Electromagnetic & Measurement Compatibility IT=Information Ed=Education Theory ED=Electron ITS=Intelligent Devices EMB=Engineering Transportation Systems MAG=Magnetics in Medicine & MTT=Microwave Theory & Techniques NANO=Nanotechnology Council NPS=Nuclear & Plasma Sciences OCEAN=Oceanic Engineering Biology GRS=Geoscience PE=Power & Remote & Sensing Energy IE=Industrial PHO=Photonics Electronics IM=Instrumentation RA=Robotics & & Measurement Automation IT=Information SEN=Sensors Council SMC=Systems, Man, & Cybernetics Theory ITS=Intelligent Transportation Systems MAG=MagneticsSP=Signal Processing MTT=Microwave SSC=Solid-‐State Circuits Theory & Techniques SSIT=Social NANO=Nanotechnology Implications of Technology Council NPS=Nuclear SUPERC=Council on Superconductivity & Plasma Sciences TM=Technology Management Council UFFC=Ultrasonics, Ferroelectrics, and Frequency Control VT=Vehicular Technology OCEAN=Oceanic Engineering PE=Power & Energy PHO=Photonics RA=Robotics & Automation SEN=Sensors Council SMC=Systems, Man, & Cyber- Figure 31 . The breakdown of networking and interconnectivity by sponsoring societies. netics SP=Signal Processing SSC=Solid-State Circuits SSIT=Social Implications of Technology SUPERC=Council on Superconductivity TM=Technol- ogy Management Council UFFC=Ultrasonics, Ferroelectrics, and Frequency Control VT=Vehicular Technology

143 Figure 31. The breakdown of networking and interconnectivity by sponsoring societies.

149 IEEE CS 2022 , Report DRAFT APPENDIX I. 23 Technologies Coverage in IEEE Publications 1/26/2014 5:18 PM

12 Software Deﬁned Networks CONTENTS Sponsoring Societies 1. INTRODUCTION...... 5 2. SEAMLESS INTELLIGENCE IM SCENARIO...... 8 VT 4% 4% 3. 23 TECHNOLOGIES IN 2022...... 14 AP 4. DRIVERS AND DISRUPTORS...... 112 4% 5. TECHNOLOGY COVERAGE IN IEEE IEEE XPLORE AND BY 5% COMM IEEE SOCIETIES...... 115 32% 6.CONTENTS IEEE COMPUTER SOCIETY SSC IN 2022...... 120 6% 7. SUMMARY AND NEXT STEPS.....124 8. AUTHORS...... 126 APPENDIX I. 23 Technologies Coverage SP in IEEE Publications...... 134 7%

MTT 7% OTHER Computer 15% 8% CAS 8%

AE=Aerospace & Electronic Systems AP=Antennas & Propagation BIO=Biometrics Council umer CAS=Circuits & Systems CE=Cons AE=Aerospace & Electronic Electronics Systems AP=Antennas CEDA=Council ctronic on Ele & Propagation Design BIO=Biometrics Automation Council CI=Computational CAS=Circuits Intelligence & Systems CE=Consumer COMM=Communications Electronics CPMT=Components, Packaging, & Manufacturing Technology CSS=Control Systems EC=Electromagnetic Compatibility Ed=Education CEDA=Council on ElectronicED=Electron Design Automation Devices CI=Computational g EMB=Engineerin in Medicine Intelligence & Biology COMM=Communications GRS=Geoscience CPMT=Components, & Remote Sensing IE=Industrial Electronics Packaging, & Manufac- turing Technology CSS=ControlIM=Instrumentation Systems & EC=Electromagnetic Measurement Compatibility IT=Information Ed=Education Theory ED=Electron ITS=Intelligent Devices Transportation EMB=Engineering Systems MAG=Magnetics in Medicine & MTT=Microwave Theory & Techniques NANO=Nanotechnology Council NPS=Nuclear & Plasma Sciences OCEAN=Oceanic Engineering Biology GRS=GeosciencePE=Power & Remote & Sensing Energy IE=Industrial PHO=Photonics Electronics IM=Instrumentation RA=Robotics & & Automation Measurement IT=Information SEN=Sensors Council SMC=Systems, Man, & Cybernetics Theory ITS=Intelligent Transportation Systems SP=Signal MAG=Magnetics Processing MTT=Microwave -‐ SSC=Solid State Circuits Theory & Techniques SSIT=Social Implications NANO=Nanotechnology of Technology Council NPS=Nuclear SUPERC=Council on Superconductivity & Plasma Sciences OCEAN=Oceanic EngineeringTM=Technology PE=Power Management & Energy PHO=Photonics Council UFFC=Ultrasonics, RA=Robotics & Ferroelectrics, Automation SEN=Sensors and Frequency Council Control VT=Vehicular Technology SMC=Systems, Man, & Cyber- Figure 32 . The breakdown of software -‐defined networks by sponsoring societies. netics SP=Signal Processing SSC=Solid-State Circuits SSIT=Social Implications of Technology SUPERC=Council on Superconductivity TM=Technol- ogy Management Council UFFC=Ultrasonics, Ferroelectrics, and Frequency Control VT=Vehicular Technology

Figure 32. The breakdown of software-defined144 networks by sponsoring societies.

150 IEEE CS 2022 , Report DRAFT APPENDIX I. 23 Technologies Coverage in IEEE Publications 1/26/2014 5:18 PM

13 HPC Sponsoring Societies CONTENTS CAS CEDA 2% 1. INTRODUCTION...... 5 2% SP 2. SEAMLESS INTELLIGENCE PHO 2% SCENARIO...... 8 COMM 2% 3. 23 TECHNOLOGIES IN 2022...... 14 MAG 3% 4. DRIVERS AND DISRUPTORS...... 112 3% 5. TECHNOLOGY COVERAGE IN GRS IEEE XPLORE AND BY 3% IEEE SOCIETIES...... 115 AP 3% 6.CONTENTS IEEE COMPUTER SOCIETY IN 2022...... 120 NPS 3% 7. SUMMARY AND NEXT STEPS.....124 8. AUTHORS...... 126 Computer APPENDIX I. 23 Technologies Coverage IEEE 52% in IEEE Publications...... 134 8%

OTHER 17%

AE=Aerospace & Electronic Systems AP=Antennas & Propagation BIO=Biometrics Council umer CAS=Circuits & Systems CE=Cons AE=Aerospace & Electronic Electronics Systems AP=Antennas CEDA=Council & Propagation on Electronic BIO=Biometrics Design mputational Automation CI=Co Council Intelligence CAS=Circuits & Systems COMM=Communications CE=Consumer Electronics CEDA=Council on Electronic CPMT=Components, Design Automation Packaging, CI=Computational & Manufacturing Intelligence Technology COMM=Communications CSS=Control Systems CPMT=Components, EC=Electromagnetic Packaging, Compatibility Ed=Education & Manufac- ED=Electron Devices EMB=Engineering in Medicine science & Biology GRS=Geo & Remote Sensing IE=Industrial Electronics turing Technology CSS=ControlIM=Instrumentation Systems EC=Electromagnetic & Measurement Compatibility IT=Information Ed=Education Theory ED=Electron ITS=Intelligent Devices EMB=Engineering Transportation Systems MAG=Magnetics in Medicine & MTT=Microwave Theory & Techniques NANO=Nanotechnology Council NPS=Nuclear & Plasma Sciences OCEAN=Oceanic Engineering Biology GRS=Geoscience PE=Power & Remote & Sensing Energy IE=Industrial PHO=Photonics Electronics IM=Instrumentation RA=Robotics & & Measurement Automation IT=Information SEN=Sensors Council SMC=Systems, Man, & Cybernetics Theory ITS=Intelligent Transportation Systems MAG=MagneticsSP=Signal Processing MTT=Microwave -‐ SSC=Solid State Circuits Theory & Techniques SSIT=Social NANO=Nanotechnology Implications of Technology SUPERC=Council Council NPS=Nuclear on Superconductivity & Plasma Sciences OCEAN=Oceanic EngineeringTM=Technology PE=Power & Management Energy PHO=Photonics Council RA=Robotics UFFC=Ultrasonics, & Automation Ferroelectrics, SEN=Sensors and Frequency Council Control SMC=Systems, VT=Vehicular Technology Man, & Cyber- netics SP=Signal Processing SSC=Solid-State CircuitsFigure SSIT=Social33 . The Implications breakdown of HPC of Technology by sponsoring SUPERC=Council societies. on Superconductivity TM=Technol- ogy Management Council UFFC=Ultrasonics, Ferroelectrics, and Frequency Control VT=Vehicular Technology

Figure 33. The breakdown of HPC by sponsoring societies. 145 151 IEEE CS 2022 , Report DRAFT APPENDIX I. 23 Technologies Coverage in IEEE Publications 1/26/2014 5:18 PM

14 Cloud Computing Sponsoring Societies CONTENTS CEDA 1. INTRODUCTION...... 5 1% 2. SEAMLESS INTELLIGENCE SMC CI 2% Other SCENARIO...... 8 2% AP 3. 23 TECHNOLOGIES IN 2022...... 14 2% 6% 4. DRIVERS AND DISRUPTORS...... 112 MAG 2% 5. TECHNOLOGY COVERAGE IN SP IEEE XPLORE AND BY 3% CONTENTSIEEE SOCIETIES...... 115 IEEE 6. IEEE COMPUTER SOCIETY 3% IN 2022...... 120 GRS 7. SUMMARY AND NEXT STEPS.....124 3% 8. AUTHORS...... 126 NPS Computer APPENDIX I. 23 Technologies Coverage 3% 51% in IEEE Publications...... 134 EMB 5%

PE 8% COM 9%

AE=Aerospace & Electronic Systems AP=Antennas & Propagation BIO=Biometrics Council umer CAS=Circuits & Systems CE=Cons

AE=Aerospace & ElectronicElectronics Systems AP=Antennas CEDA=Council & Propagation on Electronic BIO=Biometrics Design Automation Council CAS=Circuits CI=Computational & Systems CE=Consumer Intelligence COMM=Communications Electronics

CEDA=Council on ElectronicCPMT= DesignComponents, Automation Packaging, CI=Computational & Manufacturing Intelligence Technology COMM=Communications CSS=Control Systems CPMT=Components, EC=Electromagnetic Packaging, Compatibility Ed=Education & Manufac- ED=Electron Devices EMB=Engineering in Medicine & Biology GRS=Geoscience & Remote Sensing IE=Industrial Electronics turing Technology CSS=ControlIM=Instrumentation Systems & EC=Electromagnetic Measurement Compatibility IT=Information Ed=Education Theory ED=Electron ITS=Intelligent Devices Transportation EMB=Engineering Systems MAG=Magnetics in Medicine & Biology GRS=GeoscienceMTT=Microwave & Remote Sensing Theory IE=Industrial & Techniques Electronics IM=Instrumentation NANO=Nanotechnology Council & Measurement IT=Information NPS=Nuclear & Theory Plasma g Sciences OCEAN=Oceanic Engineerin ITS=Intelligent PE=Power & Energy PHO=Photonics RA=Robotics & Automation SEN=Sensors Council SMC=Systems, Man, & Cybernetics Transportation Systems SP=Signal MAG=Magnetics Processing MTT=Microwave -‐ SSC=Solid State Circuits Theory & Techniques SSIT=Social NANO=Nanotechnology Implications of Technology Council NPS=Nuclear SUPERC=Council on Superconductivity & Plasma Sciences OCEAN=Oceanic EngineeringTM=Technology PE=Power Management & Energy PHO=Photonics Council UFFC=Ultrasonics, RA=Robotics & Ferroelectrics, Automation SEN=Sensors and Frequency Council Control VT=Vehicular Technology SMC=Systems, Man, & Cyber- netics SP=Signal Processing SSC=Solid-StateFigure Circuits34 . The SSIT=Social breakdown Implications of of Technology cloud computing by sponsoring SUPERC=Council societies. on Superconductivity TM=Technol- ogy Management Council UFFC=Ultrasonics, Ferroelectrics, and Frequency Control VT=Vehicular Technology

Figure 34. The breakdown of cloud computing146 by sponsoring societies.

152 IEEE CS 2022 , Report DRAFT APPENDIX I. 23 Technologies Coverage in IEEE Publications 1/26/2014 5:18 PM

15 IoT Sponsoring Societies CONTENTS PE 2% 1. INTRODUCTION...... 5 SP 2. SEAMLESS INTELLIGENCE EMBS 2% SCENARIO...... 8 AP 3% 3. 23 TECHNOLOGIES IN 2022...... 14 3% 4. DRIVERS AND DISRUPTORS...... 112 SMC 4% 5. TECHNOLOGY COVERAGE IN Computer IEEE XPLORE AND BY IEEE IEEE SOCIETIES...... 115 31% 6% 6.CONTENTS IEEE COMPUTER SOCIETY IN 2022...... 120 7. SUMMARY AND NEXT STEPS.....124 SEN 8. AUTHORS...... 126 6% APPENDIX I. 23 Technologies Coverage in IEEE Publications...... 134 CE 9%

OTHER COMM 14% 20%

AE=Aerospace & Electronic Systems AP=Antennas & Propagation BIO=Biometrics Council CAS=Circuits & Systems CE=Consumer Electronics CEDA=Council on Electronic Design Automation CI=Computational Intelligence COMM=Communications AE=Aerospace & Electronic CPMT=Components, Systems AP=Antennas Packaging, & Propagation & Manufacturing BIO=Biometrics Technology Council CSS=Control CAS=Circuits Systems & Systems EC=Electromagnetic CE=Consumer Compatibility Electronics Ed=Education CEDA=Council on Electronic ED=Electron Design Automation Devices CI=Computational EMB=Engineering Intelligence in Medicine COMM=Communications & Biology GRS=Geoscience CPMT=Components, & Remote Sensing IE=Industrial Electronics Packaging, & Manufac- IM=Instrumentation & Measurement IT=Information Theory ITS=Intelligent Transportation Systems MAG=Magnetics turing Technology CSS=ControlMTT=Microwave Systems Theory EC=Electromagnetic & Techniques Compatibility NANO=Nanotechnology Ed=Education Council ED=Electron Devices NPS=Nuclear EMB=Engineering & Plasma g Sciences OCEAN=Oceanic Engineerin in Medicine & Biology GRS=GeosciencePE=Power & Remote & Sensing Energy IE=Industrial PHO=Photonics Electronics RA=Robotics & Automation IM=Instrumentation SEN=Sensors & Measurement Council IT=Information SMC=Systems, Man, & Cybernetics Theory ITS=Intelligent SP=Signal Processing -‐ SSC=Solid State Circuits SSIT=Social Implications of Technology SUPERC=Council on Superconductivity Transportation Systems MAG=MagneticsTM=Technology Management MTT=Microwave Council Theory UFFC=Ultrasonics, & Techniques Ferroelectrics, NANO=Nanotechnology and Frequency Control Council NPS=Nuclear VT=Vehicular Technology & Plasma Sciences OCEAN=Oceanic Engineering PE=Power & Energy PHO=PhotonicsFigure 35. The RA=Robotics breakdown & of IoT Automation by sponsoring SEN=Sensors societies. Council SMC=Systems, Man, & Cyber- netics SP=Signal Processing SSC=Solid-State Circuits SSIT=Social Implications of Technology SUPERC=Council on Superconductivity TM=Technol-

ogy Management Council UFFC=Ultrasonics, Ferroelectrics, and Frequency Control VT=Vehicular Technology

Figure 35. The breakdown of IoT147 by sponsoring societies. 153 IEEE CS 2022 , Report DRAFT APPENDIX I. 23 Technologies Coverage in IEEE Publications 1/26/2014 5:18 PM

16 Natural User Interfaces CONTENTS Sponsoring Socie8es UFFC RA SP 1. INTRODUCTION...... 5 1% 0% 1% Other 2. SEAMLESS INTELLIGENCE 1% SCENARIO...... 8 NPS 2% 3. 23 TECHNOLOGIES IN 2022...... 14 IM 4. DRIVERS AND DISRUPTORS...... 112 IEEE 2% 5. TECHNOLOGY COVERAGE IN CE 2% IEEE XPLORE AND BY 4% IEEE SOCIETIES...... 115 6.CONTENTS IEEE COMPUTER SOCIETY SMC IN 2022...... 120 8% 7. SUMMARY AND NEXT STEPS.....124 8. AUTHORS...... 126 APPENDIX I. 23 Technologies Coverage Computer in IEEE Publications...... 134 52%

EMBS 27%

AE=Aerospace & Electronic Systems AP=Antennas & Propagation rcuits BIO=Biometrics Council CAS=Ci & Systems CE=Consumer AE=Aerospace & Electronic Electronics Systems AP=Antennas CEDA=Council & Propagation on Electronic BIO=Biometrics Design Council Automation CAS=Circuits CI=Computational & Systems CE=Consumer Intelligence COMM=Communications Electronics CPMT=Components, Packaging, & Manufacturing Technology CSS=Control Systems EC=Electromagnetic Compatibility Ed=Education CEDA=Council on Electronic ED=Electron Design Automation Devices CI=Computational EMB=Engineering Intelligence in Medicine COMM=Communications & Biology CPMT=Components, GRS=Geoscience & Remote Sensing IE=Industrial Electronics Packaging, & Manufac- turing Technology CSS=ControlIM=Instrumentation Systems EC=Electromagnetic & Measurement Compatibility IT=Information Ed=Education Theory ED=Electron ITS=Intelligent Transportation Systems DevicesMAG=Magnetics EMB=Engineering in Medicine & MTT=Microwave Theory & Techniques NANO=Nanotechnology Council NPS=Nuclear & Plasma g Sciences OCEAN=Oceanic Engineerin Biology GRS=Geoscience PE=Power & Remote & Sensing Energy IE=Industrial PHO=Photonics Electronics IM=Instrumentation RA=Robotics & & Measurement Automation IT=Information SEN=Sensors Council SMC=Systems, Man, & Cybernetics Theory ITS=Intelligent Transportation Systems MAG=MagneticsSP=Signal Processing MTT=Microwave -‐ SSC=Solid State Circuits Theory & Techniques SSIT=Social NANO=Nanotechnology Implications of Technology Council NPS=Nuclear SUPERC=Council on Superconductivity & Plasma Sciences TM=Technology Management Council UFFC=Ultrasonics, Ferroelectrics, and Frequency ular Control VT=Vehic Technology OCEAN=Oceanic Engineering PE=Power & Energy PHO=Photonics RA=Robotics & Automation SEN=Sensors Council SMC=Systems, Man, & Cyber- Figure 36 . The breakdown of natural user interfaces by sponsoring societies. netics SP=Signal Processing SSC=Solid-State Circuits SSIT=Social Implications of Technology SUPERC=Council on Superconductivity TM=Technol-

ogy Management Council UFFC=Ultrasonics, Ferroelectrics, and Frequency Control VT=Vehicular Technology

Figure 36. The breakdown of natural user148 interfaces by sponsoring societies. 154 IEEE CS 2022 , Report DRAFT APPENDIX I. 23 Technologies Coverage in IEEE Publications 1/26/2014 5:18 PM

17 3D Printing CONTENTS Sponsoring Societies 1. INTRODUCTION...... 5 2. SEAMLESS INTELLIGENCE PHO SCENARIO...... 8 OTHER 5% 3. 23 TECHNOLOGIES IN 2022...... 14 5% 4. DRIVERS AND DISRUPTORS...... 112 IEEE CPMT 5. TECHNOLOGY COVERAGE IN 5% IEEE XPLORE AND BY 26% CONTENTSIEEE SOCIETIES...... 115 UFFC 6. IEEE COMPUTER SOCIETY 6% IN 2022...... 120 7. SUMMARY AND NEXT STEPS.....124 EMBS 8. AUTHORS...... 126 6% APPENDIX I. 23 Technologies Coverage in IEEE Publications...... 134 MTT 9% AP 16% EC 10% Computer 12%

AE=Aerospace & ElectronicAE=Aerospace Systems AP=Antennas & Electronic & Systems Propagation BIO=Biometrics AP=Antennas & Council Propagation CAS=Circuits & BIO=Biometrics Systems CE=Consumer Council umer CAS=Circuits & Systems CE=Cons Electronics Electronics CEDA=Council on Electronic Design Automation CI=Computational Intelligence COMM=Communications CEDA=Council on Electronic CPMT=Components, Design Automation Packaging, CI=Computational & Manufacturing Intelligence Technology COMM=Communications CSS=Control Systems CPMT=Components, EC=Electromagnetic Compatibility Ed=Education Packaging, & Manufac- ED=Electron Devices EMB=Engineering in Medicine & Biology GRS=Geoscience & Remote Sensing IE=Industrial Electronics turing Technology CSS=ControlIM=Instrumentation Systems EC=Electromagnetic & Measurement Compatibility IT=Information Ed=Education Theory ED=Electron ITS=Intelligent Devices Transportation EMB=Engineering Systems MAG=Magnetics in Medicine & Biology GRS=GeoscienceMTT=Microwave & Remote Sensing Theory & IE=Industrial Techniques Electronics IM=Instrumentation NANO=Nanotechnology Council & Measurement NPS=Nuclear IT=Information & Plasma Theory Sciences OCEAN=Oceanic Engineering ITS=Intelligent PE=Power & Energy PHO=Photonics RA=Robotics & Automation SEN=Sensors Council SMC=Systems, Man, & Cybernetics Transportation Systems SP=Signal MAG=Magnetics Processing MTT=Microwave SSC=Solid-‐State Circuits Theory & Techniques SSIT=Social NANO=Nanotechnology Implications of Technology Council NPS=Nuclear SUPERC=Council on Superconductivity & Plasma Sciences OCEAN=Oceanic EngineeringTM=Technology PE=Power Management & Energy PHO=Photonics Council UFFC=Ultrasonics, RA=Robotics & Ferroelectrics, Automation SEN=Sensors and Frequency Council Control VT=Vehicular Technology SMC=Systems, Man, & Cyber- netics SP=Signal Processing SSC=Solid-State Circuits SSIT=Social Implications of Technology SUPERC=Council on Superconductivity TM=Technol- Figure 37 . The breakdown of 3D printing by sponsoring societies. ogy Management Council UFFC=Ultrasonics, Ferroelectrics, and Frequency Control VT=Vehicular Technology

Figure 37. The breakdown of 3D printing by sponsoring societies. 149 155 IEEE CS 2022 , Report DRAFT APPENDIX I. 23 Technologies Coverage in IEEE Publications 1/26/2014 5:18 PM

18 Big Data Analytics CONTENTS Sponsoring Societies 1. INTRODUCTION...... 5 TM 2. SEAMLESS INTELLIGENCE PE 2% SCENARIO...... 8 2% 3. 23 TECHNOLOGIES IN 2022...... 14 COMM 5% 4. DRIVERS AND DISRUPTORS...... 112 5. TECHNOLOGY COVERAGE IN IEEE XPLORE AND BY IEEE SOCIETIES...... 115 6.CONTENTS IEEE COMPUTER SOCIETY IN 2022...... 120 7. SUMMARY AND NEXT STEPS.....124 8. AUTHORS...... 126 APPENDIX I. 23 Technologies Coverage in IEEE Publications...... 134

Computer 91%

AE=Aerospace & Electronic Systems AP=Antennas & Propagation BIO=Biometrics Council umer CAS=Circuits & Systems CE=Cons AE=Aerospace & Electronic Electronics Systems AP=Antennas CEDA=Council & Propagation on Electronic omation Design Aut BIO=Biometrics Council CI=Computational CAS=Circuits Intelligence & Systems CE=Consumer COMM=Communications Electronics CPMT=Components, Packaging, & Manufacturing Technology CSS=Control Systems EC=Electromagnetic Compatibility Ed=Education CEDA=Council on Electronic ED=Electron Design Automation Devices CI=Computational EMB=Engineering Intelligenceology in Medicine & Bi COMM=Communications GRS=Geoscience & Remote CPMT=Components, Sensing IE=Industrial Electronics Packaging, & Manufac- turing Technology CSS=ControlIM=Instrumentation Systems EC=Electromagnetic & Measurement Compatibility IT=Information Ed=Education Theory ED=Electron ITS=Intelligent Devices Transportation EMB=Engineering Systems MAG=Magnetics in Medicine & MTT=Microwave Theory & Techniques NANO=Nanotechnology l Counci NPS=Nuclear & Plasma Sciences OCEAN=Oceanic Engineering Biology GRS=GeosciencePE=Power & Remote & Sensing Energy IE=Industrial PHO=Photonics Electronics IM=Instrumentation RA=Robotics & & Measurement Automation IT=Information SEN=Sensors Council SMC=Systems, Man, & Cybernetics Theory ITS=Intelligent Transportation Systems MAG=MagneticsSP=Signal Processing MTT=Microwave -‐ SSC=Solid State Circuits Theory & Techniques SSIT=Social Implications NANO=Nanotechnology of Technology Council SUPERC=Council NPS=Nuclear on Superconductivity & Plasma Sciences TM=Technology Management Council UFFC=Ultrasonics, Ferroelectrics, and Frequency Control VT=Vehicular Technology OCEAN=Oceanic Engineering PE=Power & Energy PHO=Photonics RA=Robotics & Automation SEN=Sensors Council SMC=Systems, Man, & Cyber- netics SP=Signal Processing SSC=Solid-State CircuitsFigure SSIT=Social38 . The Implications breakdown of big of Technology data analytics SUPERC=Council by sponsoring on societies. Superconductivity TM=Technol- ogy Management Council UFFC=Ultrasonics, Ferroelectrics, and Frequency Control VT=Vehicular Technology

Figure 38. The breakdown of big data analytics by sponsoring societies. 150 156 IEEE CS 2022 , Report DRAFT APPENDIX I. 23 Technologies Coverage in IEEE Publications 1/26/2014 5:18 PM

19 Machine Learning, Intel. Sys. Sponsoring Societies CONTENTS ITS IE IT PE 1% 1% IEEE BIO 1% 1% 1% 1. INTRODUCTION...... 5 1% 2. SEAMLESS INTELLIGENCE SCENARIO...... 8 3. 23 TECHNOLOGIES IN 2022...... 14 OTHER 4. DRIVERS AND DISRUPTORS...... 112 6% 5. TECHNOLOGY COVERAGE IN EMB IEEE XPLORE AND BY 6% IEEE SOCIETIES...... 115 6.CONTENTS IEEE COMPUTER SOCIETY IN 2022...... 120 7. SUMMARY AND NEXT STEPS.....124 SP Computer 8. AUTHORS...... 126 11% 47% APPENDIX I. 23 Technologies Coverage in IEEE Publications...... 134

SMC 11%

CI 13%

AE=Aerospace & Electronic Systems AP=Antennas & Propagation BIO=Biometrics Council umer CAS=Circuits & Systems CE=Cons AE=Aerospace & Electronic Electronics Systems AP=Antennas CEDA=Council & Propagation on Electronic BIO=Biometrics Design Automation Council CAS=Circuits CI=Computational Intelligence COMM=Communications & Systems CE=Consumer Electronics CEDA=Council on ElectronicCPMT=Components, Design Automation Packaging, CI=Computational & Manufacturing Intelligence Technology COMM=Communications CSS=Control Systems CPMT=Components, EC=Electromagnetic Compatibility Ed=Education Packaging, & Manufac- ED=Electron Devices EMB=Engineering in Medicine & Biology GRS=Geoscience & Remote Sensing IE=Industrial Electronics turing Technology CSS=ControlIM=Instrumentation Systems & EC=Electromagnetic Measurement Compatibility IT=Information Ed=Education Theory ED=Electron ITS=Intelligent Devices Transportation EMB=Engineering Systems MAG=Magnetics in Medicine & Biology GRS=GeoscienceMTT=Microwave & Remote Sensing Theory IE=Industrial & Techniques Electronics IM=Instrumentation NANO=Nanotechnology Council & Measurement IT=Information NPS=Nuclear AN=Oceanic & Plasma Sciences OCE Engineering Theory ITS=Intelligent PE=Power & Energy PHO=Photonics RA=Robotics & Automation SEN=Sensors Council SMC=Systems, Man, & Cybernetics Transportation Systems SP=Signal MAG=Magnetics Processing MTT=Microwave -‐ SSC=Solid State Circuits Theory & Techniques SSIT=Social NANO=Nanotechnology Implications of Technology SUPERC=Council Council on Superconductivity NPS=Nuclear & Plasma Sciences OCEAN=Oceanic EngineeringTM=Technology PE=Power Management & Energy PHO=Photonics Council UFFC=Ultrasonics, RA=Robotics & Ferroelectrics, Automation SEN=Sensors and Frequency Council Control VT=Vehicular Technology SMC=Systems, Man, & Cyber- Figure 39 . The breakdown of machine learning and intelligent systems by sponsoring societies. netics SP=Signal Processing SSC=Solid-State Circuits SSIT=Social Implications of Technology SUPERC=Council on Superconductivity TM=Technol- ogy Management Council UFFC=Ultrasonics, Ferroelectrics, and Frequency Control VT=Vehicular Technology

Figure 39. The breakdown of machine learning and151 intelligent systems by sponsoring societies.

157 IEEE CS 2022 , Report DRAFT APPENDIX I. 23 Technologies Coverage in IEEE Publications 1/26/2014 5:18 PM

20. Computer Vision and Pattern CONTENTS Recognition Sponsoring Societies 1. INTRODUCTION...... 5 RAS, 1% 2. SEAMLESS INTELLIGENCE SCENARIO...... 8 ITS, 2% 3. 23 TECHNOLOGIES IN 2022...... 14 4. DRIVERS AND DISRUPTORS...... 112 CI, 6% BIO, 2% 5. TECHNOLOGY COVERAGE IN IEEE XPLORE AND BY GRS, 5% IEEE SOCIETIES...... 115 CONTENTS EMB, 6% Other, 6. IEEE COMPUTER SOCIETY 38% IN 2022...... 120 7. SUMMARY AND NEXT STEPS.....124 8. AUTHORS...... 126 SMC, 7% APPENDIX I. 23 Technologies Coverage in IEEE Publications...... 134

SP, 14%

Compute r, 19%

AE=Aerospace & Electronic Systems AP=Antennas & Propagation BIO=Biometrics Council umer CAS=Circuits & Systems CE=Cons Electronics CEDA=Council on Electronic Design Automation CI=Computational ons Intelligence COMM=Communicati AE=Aerospace & ElectronicCPMT=Components, Systems AP=Antennas Packaging, & Propagation & Manufacturing BIO=Biometrics Technology Council CSS=Control CAS=Circuits & Systems Systems CE=Consumer EC=Electromagnetic Compatibility Ed=Education Electronics CEDA=Council on ElectronicED=Electron Design Automation Devices CI=Computational EMB=Engineering Intelligence in Medicine COMM=Communications & Biology GRS=Geoscience CPMT=Components,ectronics & Remote Sensing IE=Industrial El Packaging, & Manufac- IM=Instrumentation & Measurement IT=Information Theory ITS=Intelligent Transportation Systems MAG=Magnetics turing Technology CSS=ControlMTT=Microwave Systems Theory EC=Electromagnetic & Techniques Compatibility NANO=Nanotechnology Ed=Education Council ED=Electron Devices NPS=Nuclear EMB=Engineering & ing Plasma Sciences OCEAN=Oceanic Engineer in Medicine & Biology GRS=GeosciencePE=Power & Remote & Sensing Energy IE=Industrial PHO=Photonics Electronics IM=Instrumentation RA=Robotics & & Measurement Automation IT=Information SEN=Sensors Council SMC=Systems, Man, & Cybernetics Theory ITS=Intelligent SP=Signal Processing -‐ SSC=Solid State Circuits SSIT=Social Implications of Technology vity SUPERC=Council on Superconducti Transportation Systems TM=Technology MAG=Magnetics Management MTT=Microwave Council Theory & UFFC=Ultrasonics, Techniques NANO=Nanotechnology Ferroelectrics, and Frequency Council NPS=Nuclear Control VT=Vehicular Technology & Plasma Sciences OCEAN=Oceanic Engineering PE=PowerFigure 40 &. EnergyThe breakdown PHO=Photonicsc of omputer RA=Robotics vision and p&attern Automationa nalysis SEN=Sensorsby sponsoring Council societies. SMC=Systems, Man, & Cyber- netics SP=Signal Processing SSC=Solid-State Circuits SSIT=Social Implications of Technology SUPERC=Council on Superconductivity TM=Technol- ogy Management Council UFFC=Ultrasonics, Ferroelectrics, and Frequency Control VT=Vehicular Technology

Figure 40. The breakdown of computer vision and pattern analysis by sponsoring societies. 152 158 IEEE CS 2022 , Report DRAFT APPENDIX I. 23 Technologies Coverage in IEEE Publications 1/26/2014 5:18 PM

CE COMM IT 21 Life Sciences 0% 0% 0% NANO Sponsoring Societies 0% CONTENTS SuperC SSC AP SMC CSS 1% 1% 1% 0% 1. INTRODUCTION...... 5 MTT 1% RA OCEAN 1% 1% 2. SEAMLESS INTELLIGENCE IM 0% SCENARIO...... 8 2% CI 1% 3. 23 TECHNOLOGIES IN 2022...... 14 BIO PHO 1% 4. DRIVERS AND DISRUPTORS...... 112 2% MAG 5. TECHNOLOGY COVERAGE IN 2% IEEE XPLORE AND BY CAS IEEE SOCIETIES...... 115 SEN CONTENTS 2% 2% 6. IEEE COMPUTER SOCIETY EMB IN 2022...... 120 IEEE 7. SUMMARY AND NEXT STEPS.....124 3% 40% 8. AUTHORS...... 126 APPENDIX I. 23 Technologies Coverage SP in IEEE Publications...... 134 7%

Computer 12%

UFFC NPS 8% 10%

AE=Aerospace & Electronic Systems AP=Antennas & Propagation BIO=Biometrics Council umer CAS=Circuits & Systems CE=Cons Electronics CEDA=Council on Electronic Design Automation CI=Computational Intelligence COMM=Communications AE=Aerospace & ElectronicCPMT=Components, Systems AP=Antennas Packaging, & Propagation & Manufacturing BIO=Biometrics Technology Council CAS=Circuits CSS=Control & Systems Systems CE=Consumer EC=Electromagnetic Compatibility Ed=Education Electronics CEDA=Council on ElectronicED=Electron Design Automation Devices CI=Computational EMB=Engineering in Medicine Intelligence & Biology COMM=Communications GRS=Geoscience & CPMT=Components, Remote Sensing IE=Industrial Electronics Packaging, & Manufac- IM=Instrumentation & Measurement IT=Information Theory ITS=Intelligent Transportation Systems MAG=Magnetics turing Technology CSS=ControlMTT=Microwave Systems Theory EC=Electromagnetic & Techniques Compatibility ology NANO=Nanotechn Council Ed=Education ED=Electron NPS=Nuclear Devices & Plasma EMB=Engineering Sciences OCEAN=Oceanic Engineering in Medicine & Biology GRS=GeosciencePE=Power & Remote & Sensing Energy IE=Industrial PHO=Photonics Electronics IM=Instrumentation RA=Robotics & & Measurement Automation IT=Information SEN=Sensors Council SMC=Systems, Man, & Cybernetics Theory ITS=Intelligent SP=Signal Processing -‐ SSC=Solid State Circuits SIT=Social S Implications of Technology SUPERC=Council on Superconductivity Transportation Systems MAG=MagneticsTM=Technology Management MTT=Microwave Council Theory & Techniques UFFC=Ultrasonics, NANO=Nanotechnology Ferroelectrics, and Council Frequency NPS=Nuclear Control VT=Vehicular Technology & Plasma Sciences OCEAN=Oceanic Engineering PE=Power & Energy PHO=PhotonicsFigure 41 . The RA=Robotics breakdown of life & Automation sciences by sponsoring SEN=Sensors societies. Council SMC=Systems, Man, & Cyber- netics SP=Signal Processing SSC=Solid-State Circuits SSIT=Social Implications of Technology SUPERC=Council on Superconductivity TM=Technol- ogy Management Council UFFC=Ultrasonics, Ferroelectrics, and Frequency Control VT=Vehicular Technology

Figure 41. The breakdown of life sciences by sponsoring societies. 153 159 IEEE CS 2022 , Report DRAFT APPENDIX I. 23 Technologies Coverage in IEEE Publications 1/26/2014 5:18 PM

22 Computation Biology Sponsoring Societies CSS NANO CAS CONTENTS IT 1% 1% -1% 1. INTRODUCTION...... 5 1% RA 2. SEAMLESS INTELLIGENCE SMC 1% SCENARIO...... 8 3% OTHER 3. 23 TECHNOLOGIES IN 2022...... 14 3% 4. DRIVERS AND DISRUPTORS...... 112 IEEE 5. TECHNOLOGY COVERAGE IN 4% IEEE XPLORE AND BY SP Computer IEEE SOCIETIES...... 115 4% 28% 6.CONTENTS IEEE COMPUTER SOCIETY IN 2022...... 120 7. SUMMARY AND NEXT STEPS.....124 8. AUTHORS...... 126 CSS APPENDIX I. 23 Technologies Coverage 13% in IEEE Publications...... 134

EMB CI 23% 18%

AE=Aerospace & Electronic Systems opagation AP=Antennas & Pr BIO=Biometrics Council CAS=Circuits & Systems CE=Consumer AE=Aerospace & Electronic Electronics Systems AP=Antennas CEDA=Council & Propagation on Electronic BIO=Biometrics Design Automation Council CAS=Circuits CI=Computational & Systems CE=Consumer Intelligence COMM=Communications Electronics CPMT=Components, Packaging, & Manufacturing Technology S=Control CS Systems EC=Electromagnetic Compatibility Ed=Education CEDA=Council on ElectronicED=Electron Design Automation Devices CI=Computational EMB=Engineering Intelligence in Medicine COMM=Communications & Biology GRS=Geoscience CPMT=Components, & Remote Sensing IE=Industrial Electronics Packaging, & Manufac- turing Technology CSS=ControlIM=Instrumentation Systems EC=Electromagnetic & Measurement Compatibility IT=Information Theory Ed=Education ITS=Intelligent ED=Electron Transportation Devices Systems EMB=Engineering MAG=Magnetics in Medicine & MTT=Microwave Theory & Techniques NANO=Nanotechnology Council NPS=Nuclear & Plasma g Sciences OCEAN=Oceanic Engineerin Biology GRS=GeosciencePE=Power & Remote & Sensing Energy IE=Industrial PHO=Photonics Electronics IM=Instrumentation on RA=Robotics & Automati SEN=Sensors & Measurement Council IT=Information SMC=Systems, Man, & Cybernetics Theory ITS=Intelligent Transportation Systems MAG=MagneticsSP=Signal Processing MTT=Microwave -‐ SSC=Solid State Circuits Theory & Techniques SSIT=Social NANO=Nanotechnology Implications of Technology Council NPS=Nuclear SUPERC=Council on Superconductivity & Plasma Sciences TM=Technology Management Council UFFC=Ultrasonics, Ferroelectrics, and Frequency Control VT=Vehicular Technology OCEAN=Oceanic Engineering PE=Power & Energy PHO=Photonics RA=Robotics & Automation SEN=Sensors Council SMC=Systems, Man, & Cyber- netics SP=Signal Processing SSC=Solid-StateFigure Circuits42 . SSIT=Social The breakdown Implications of computational of Technology biology by SUPERC=Council sponsoring on societies. Superconductivity TM=Technol- ogy Management Council UFFC=Ultrasonics, Ferroelectrics, and Frequency Control VT=Vehicular Technology

Figure 42. The breakdown of computational biology by sponsoring societies. 154 160 IEEE CS 2022 , Report DRAFT APPENDIX I. 23 Technologies Coverage in IEEE Publications 1/26/2014 5:18 PM

23 Robotics in Medicine Sponsoring Societies Educ ITS CONTENTS Ocean CE SEN 1% 1% 1% 1. INTRODUCTION...... 5 2% 1% 2. SEAMLESS INTELLIGENCE IM SCENARIO...... 8 2% 3. 23 TECHNOLOGIES IN 2022...... 14 CI 3% 4. DRIVERS AND DISRUPTORS...... 112 IEEE 5. TECHNOLOGY COVERAGE IN 4% IEEE XPLORE AND BY IEEE SOCIETIES...... 115 Computer 4% 6.CONTENTS IEEE COMPUTER SOCIETY IN 2022...... 120 EMB 7. SUMMARY AND NEXT STEPS.....124 5% 8. AUTHORS...... 126 RA APPENDIX I. 23 Technologies Coverage 48% in IEEE Publications...... 134 SMC 8%

IE 10% CSS 10%

AE=Aerospace & Electronic Systems AP=Antennas & Propagation BIO=Biometrics Council CAS=Circuits & Systems CE=Consumer AE=Aerospace & Electronic Electronics Systems AP=Antennas CEDA=Council & Propagation on Electronic BIO=Biometrics Design Automation Council CAS=Circuits CI=Computational & Systems CE=Consumer Intelligence COMM=Communications Electronics CEDA=Council on ElectronicCPMT=Components, Design Automation Packaging, CI=Computational & Manufacturing Intelligence Technology COMM=Communications CSS=Control Systems EC=Electromagnetic CPMT=Components, Compatibility Packaging, Ed=Education & Manufac- ED=Electron Devices EMB=Engineering in Medicine & Biology GRS=Geoscience & Remote Sensing IE=Industrial Electronics turing Technology CSS=ControlIM=Instrumentation Systems EC=Electromagnetic & Measurement Compatibility IT=Information Ed=Education Theory ED=Electron ITS=Intelligent Transportation Systems Devices EMB=Engineering MAG=Magnetics in Medicine & MTT=Microwave Theory & Techniques NANO=Nanotechnology ncil Cou NPS=Nuclear & Plasma Sciences OCEAN=Oceanic Engineering Biology GRS=GeosciencePE=Power & Remote & Sensing Energy IE=Industrial PHO=Photonics Electronics IM=Instrumentation RA=Robotics & & Measurement Automation IT=Information SEN=Sensors Council SMC=Systems, Man, & Cybernetics Theory ITS=Intelligent Transportation Systems MAG=MagneticsSP=Signal Processing MTT=Microwave -‐ SSC=Solid State Circuits Theory & Techniques l SSIT=Socia Implications NANO=Nanotechnology of Technology Council SUPERC=Council NPS=Nuclear on Superconductivity & Plasma Sciences OCEAN=Oceanic EngineeringTM=Technology PE=Power Management & Energy PHO=Photonics Council RA=Robotics UFFC=Ultrasonics, & Automation Ferroelectrics, SEN=Sensors and Frequency Council Control VT=Vehicular Technology SMC=Systems, Man, & Cyber- netics SP=Signal Processing SSC=Solid-StateFigure Circuits43 . SSIT=Social The breakdown Implications of robotic ofs Technology in medicine by SUPERC=Council sponsoring societies. on Superconductivity TM=Technol- ogy Management Council UFFC=Ultrasonics, Ferroelectrics, and Frequency Control VT=Vehicular Technology

155 Figure 43. The breakdown of robotics in medicine by sponsoring societies.

161 CONTENTS We also did the search on Google Scholar and plan to do similar effort on MS Academic Research. We present the former in the table below. 1. INTRODUCTION...... 5 Table 4. Google and IEEE Xplore search results combined. 2. SEAMLESS INTELLIGENCE SCENARIO...... 8 # Xplore Google Google/ Technology Boolean search query 3. 23 TECHNOLOGIES IN 2022...... 14 articles Scholar (K) IEEE ratio 4. DRIVERS AND DISRUPTORS...... 112 1. Security Cross- ((Privacy OR Security OR Intrusion) OR Intrusion OR “Security 12,389 523.0 9.2 5. TECHNOLOGY COVERAGE IN Cutting Issues legislation”) OR (((cyber) OR cybersecurity) OR cyber-security) OR IEEE XPLORE AND BY “cyber security” IEEE SOCIETIES...... 115 CONTENTS 2. Open Intellectual (((“Crowd sourcing”) OR “Open IP”) OR Open AND “Intellectual 1,416 16.2 13.3 6. IEEE COMPUTER SOCIETY Property Movement Property”) OR “Open standards” IN 2022...... 120 3. Sustainability ((((“Energy usage”) AND computing)) OR (“Sustainability”) OR 882 11.7 10.7 7. SUMMARY AND NEXT STEPS.....124 (“Green computing”) OR (“Carbon footprint”) OR (“Earth friendly”)) 8. AUTHORS...... 126 OR (Green ICT)) OR (Sustainable Computing) APPENDIX I. 23 Technologies Coverage 4. Massively Online ((((“Open Courses”) OR (“Massively Online”) NOT “Games”) OR 458 0.4 0.9 in IEEE Publications...... 134 Open Courses “Massively” AND “Courses”) OR “Online learning”) OR “Automated grading” 5. Quantum Computing “Quantum Computing”) OR (“Quantum” AND “mechanical phenom- 2,823 26.1 1.3 ena”) OR (“Quantum properties”) OR (“Quantum annealing”) OR (“factorization algorithm” OR “Shor”) OR (“Qubit” 6. Device and (((“Microelectromechanical systems”) OR “Nano-technology” OR 7,546 335.0 5.6 Nano-technology “Nanotechnology” OR “Nano technology”) OR “Microelectrome- chanical systems”) OR “Micro machine” OR “Micro machines” OR “Micromachines” OR “Micromachine” OR “Micro-machine” OR “Micro-machines” 7. 3D Integrated Circuits ((((((“2.5D chip” OR “2.5-D chip” OR “2.5D chips” OR “2.5-D 1,579 22.8 14.4 chips”))) OR (“3D chip” OR “3-D chip” OR “3D chips” OR “3-D chips”)) OR “System on a Chip”) OR “System in a Package”) 8. Universal Memory ((((“Non-volatile memory”) OR Memristor) OR “Spin Trans- 460 10.4 44.4 fer Torque” RAM) OR “Phase Change Memory”) OR “Universal Memory” 9. Multicore (((((“Multicore”) OR “Multiprocessor”) OR GPU) OR (“Accelerators”) 2,276 72.6 15.8 AND “processor”)) OR GPGPU) OR “Manycore” 10. Photonics ((“Photonics interconnect”) AND “Silicon photonics”) OR VCSEL OR 1,313 20.0 22.6 “Vertical Cavity Surface Emitting Laser”)

162 # Xplore Google Google/ Technology Boolean search query articles Scholar (K) IEEE ratio 11. Networking and (((“Interconnects”) OR (((“Inter-connectivity”) OR “Interconnectivi- 2,939 16.6 221.4 CONTENTS Inter-connectivity ty”) OR “Inter connectivity”) AND Networking) OR (“Ethernet”) AND “internet”) OR “Ethernet” AND Networking 1. INTRODUCTION...... 5 2. SEAMLESS INTELLIGENCE 12. Software Defined (((((((“Software Defined Networks”) OR “Software defined network- 496 5.3 9.1 SCENARIO...... 8 Networks ing”) OR “Index Terms”:SDN) OR OpenFlow) OR “Software radio”) OR “Active networking”) OR “Virtual Local Area Networks”) OR 3. 23 TECHNOLOGIES IN 2022...... 14 VLAN 4. DRIVERS AND DISRUPTORS...... 112 13. High Performance ((((((“High Performance Computing” OR HPC)) OR Supercom- 1,068 51.1 47.8 5. TECHNOLOGY COVERAGE IN Computing puters) OR “Message Passing Interface”) OR GPGPU) OR “Com- IEEE XPLORE AND BY pute-intensive”) OR Petascale) OR Exascale IEEE SOCIETIES...... 115 14. Cloud Computing ((((((((Cloud Computing) OR “Grid computing”) OR “Cluster com- 4,252 521.0 31.9 6.CONTENTS IEEE COMPUTER SOCIETY puting”) OR Virtualization) OR “-as-a-Service”) OR IaaS) OR PaaS) IN 2022...... 120 OR SaaS) OR “Pay as you go” 7. SUMMARY AND NEXT STEPS.....124 15. Internet of Things (((((“Internet of Things”) OR “Smart homes”) OR Ubiquity) OR Per- 442 262.0 122.5 8. AUTHORS...... 126 vasiveness) OR Interconnectivity) OR “Smart dust” APPENDIX I. 23 Technologies Coverage 16. Natural User (“Natural User Interfaces” OR “NUI”) OR ((“gesture recognition”) OR 3,581 13.4 178.6 in IEEE Publications...... 134 Interfaces (“Speech and gesture recognition”)) OR (“Graphical user interface” OR “NUI”) OR (“Human Computer Interface” OR “HCI”) OR (“Multi sensor input” OR “Multiple sensor input”) OR (“Augmented reality”) 17. 3D Printing ((((3-D) OR 3D) AND Printing) OR “Additive manufacturing”) OR 215 38.4 42.2 “Selective laser sintering” 18. Big Data Analytics ((“Big data”) OR “Massive Data”) AND Analytics 42 9.3 11.4 19. Machine Learning (((((“Artificial intelligence”) OR “Machine Intelligence”) OR “Intelli- 13,199 17.6 3.7 and Intelligent gent systems”) OR “Machine Learning”) OR “Supervised learning”) Systems OR “Reinforcement learning” 20. Computer Vision and (“Computer Vision” OR “Pattern Recognition” OR “Object Rec- 11,728 18.0 1.53 Pattern Matching ognition” OR “Object Detection” OR “Visual Reconstruction” OR “Image Based Rendering” OR “Structure From Motion” OR “Scene Understanding” OR “Image Understanding” OR “Machine Vision” OR “Robot Vision”) 21. Life Sciences (((((((Bioinformatic) OR Biology) OR Biomedical) OR Biometrics) OR 28,510 450.0 592.8 (Health) OR “Health care”) OR Healthcare) OR “Life Sciences”) OR Medical) OR Medicine 22. Computational ((((“Computational Biology”) OR Bioinformatics) OR “Structural 2,145 19.5 15.2 Biology and bioinformatics”) OR Phylogenetics and evolutionary modeling) OR Bioinformatics Phylogenetics 23. Robotics (Robotics) OR Robot 8,817 658.0 74.6

163