sign in sign up
<<
Home , Ubiquitous computing, Wearable computer

Education & Training Editor: Scott F. Midkiff ■ Virginia Tech ■ [email protected]

Wearable and Ubiquitous

Tom Martin

EDITOR’S INTRODUCTION ture and thickness of the user’s device primarily depends on the relative power This issue’s column continues its coverage of innovative courses in pervasive computing. consumption and expense of computa- Tom Martin of Virginia Tech describes a course on wearable and that tion, communication, and local storage. he developed and has taught twice. He describes the course’s scope, assignments and grad- Because wearable and ubiquitous ing, and design projects and his experiences with them. Please send me your comments and computing face significant suggestions for future columns. —Scott Midkiff power consumption challenges, the course spends a good amount of time on low-power design and -level irginia Tech has offered a course on • User input/output devices power management. These topics are V wearable and ubiquitous comput- • Location and also some of my main research areas ing in the Bradley Department of Elec- • Application case studies and let me connect teaching and re- trical and Engineering twice: search. I begin this section of the class one at the senior or master’s level in The wearable and ubiquitous com- with the power consumption mecha- spring 2002 (15 students) and another puting overview begins with Mark nisms in digital CMOS (complementary at the master’s level in spring 2003 (11 Weiser’s papers from the early 1990s1–2 metal-oxide semiconductor) circuits and students). The course aims to provide and more recent articles by Mahadev then move on to higher-level power students with an appreciation of current Satyanarayanan3 and .4–5 management problems, such as dynamic wearable and ubiquitous computing These provide a road map and motiva- CPU speed-setting, low-power compi- research issues and give them hands-on tion for topics covered later in the lation and source code modification, design experience. The course features a course. The papers also give the stu- and power management state transition group project (see the related sidebar) dents an historical perspective about the strategies. We also spend time studying that reinforces the readings and lectures. evolution of the research issues since batteries—particularly characteristics of The project also teaches students about Weiser’s early work and about advances various battery chemistries and battery the design process in general, including in hardware and networking. life estimation. refining a specification, partitioning Many of the students are in their early functionality, creating interfaces between twenties. For most of their lives, cell subsystems, working in teams, and plan- phones and personal digital assistants QUICK FACTS ning their work. have been widely available, and proces- sor clock speeds have been measured in Courses: ECE 4984 and ECE 5984, COURSE BACKGROUND hundreds of megahertz and main mem- Wearable and Ubiquitous Computing The course comprises about 25 lec- ory in hundreds of megabytes. They Department: Electrical and Computer tures covering have heard about Moore’s Law, but this Engineering is their first concrete lesson in its con- Institution: Virginia Tech • A wearable and ubiquitous comput- sequences. Another major lesson from Instructor: Tom Martin ([email protected]) ing overview these papers concerns the interdepend- Level: Undergraduate and graduate • Low-power design and power ence of the user’s personal device and URL: www.ece.vt.edu/~tlmartin/ece5984_ management the infrastructure available in the user’s wearable • Hardware case studies environment. The required infrastruc-

8 PERVASIVEcomputing Published by the IEEE CS and IEEE ComSoc ■ 1536-1268/03/$17.00 © 2003 IEEE Case studies of the Itsy6 and the IBM SAMPLE PROJECT DESCRIPTIONS wristwatch7 unite many of the low-power issues. Itsy’s main design goal was high performance with low The group design project aims to give students hands-on experience designing a wearable power consumption, so a large research and ubiquitous computing system and to teach them about aspects of the design process community would find it useful; size that will help them with any computing system’s design. Here I give some representative sam- was a secondary consideration because ple project descriptions. of the PDA form factor. However, size drove the IBM Linux wristwatch’s de- front-end for emergency response sign because of the form factor. IBM When responding to a disaster, you must keep track of emergency personnel and transmit also wanted an intuitive timely information in the field. In this project, students should create a front-end system such in an aesthetic package. Consequently, that someone in the command center will have a map showing the locations of all personnel the two research prototypes illustrate and be able to select an individual on the map to communicate with them or find out about the trade-offs that designers can make that individual’s environment. The individual will have a body or head-mounted video cam- between computational performance, era, a microphone and headset, and perhaps other sensors (for example, temperature, oxy- power consumption, and physical size. gen, and physiological). Another major problem wearable com- A project by-product should be a product-feature matrix for commercially available wear- puting faces is the impracticality of tradi- able that will suit this task. tional user I/O devices such as screens and keyboards. The lectures cover alternative Camera for historical surveys forms of interaction, including tactile dis- Before beginning a major construction project such as a road, the construction company plays, manipulative user interfaces, and must perform a historical survey of the area where the construction will take place. This survey movement-aware clothing. I also ask stu- documents any structures and features that might have historical significance. Currently, an dents to use an alternative text input inter- individual takes field notes by hand describing the location and structure details and takes face, Dasher (www.inference.phy.cam.ac. photographs of the structure and surroundings. Back in the office, the individual locates the uk/dasher), and compare it to typing and structure on a map, types the notes, and integrates the photographs into the notes. The sur- writing by hand. veyor then enters this into a GIS (geographic ) and perhaps creates a com- Several of the user I/O devices provide puter model of the landscape. a smooth transition into context-aware- The project should create a historical survey device for a digital camera that records the ness because they can be sensitive to user photograph’s location and the direction it was taken from. The device should automatically actions that are explicitly meant to be place the photograph on an electronic map of the area. input or to implicit actions that an appli- cation can use to infer the user’s current Ultrasonic building-mapping garment context. For example, an explicit action This project aims to model and create an electronic-textile garment that maps a building as would be a hand signal that tells a stereo the user walks through it using ultrasonic emitters and detectors, and perhaps other sensors to lower its volume, while an implicit facing in all directions. The device should be able to estimate the distance to objects in all action would be walking upstairs. Con- directions (within a certain range) and create a map of the areas (of a building) that a user has text-awareness involves knowing the passed through. The system should account for the user’s movement (it might be necessary user’s location, activity, companions, to stand at an entrance to create an initial registration point). and nearby resources. Without context- Students should model the garment in Ptolemy (www.ptolemy.eecs.berkeley.edu) in two awareness, a wearable computer will dimensions. The model should take as input a floor plan and a path through that floor plan, only distract the user, and a ubiquitous and output the garment’s map along the path. The model must account for the ultrasonic computing environment will offer little sensors’ physical properties. Students should use the model to answer questions such as benefit. The course lectures discuss im- plementations using various types of •What are the for mapping and feature detection (for example, walls, corners, sensors (for example, , and openings)? magnetometers, omnidirectional video • How many ultrasonic emitters and detectors do you need? How should you place them on cameras, and microphones) and post- the body? How does the generated map’s accuracy vary with the number and placement processing. We also discuss location- of emitters and detectors? awareness, particularly methods for • What information do you need about the user’s movement (for example, velocity and rates determining location indoors. of turning)? The remaining lectures comprise Continued on p. 10 application case studies. These case

JULY–SEPTEMBER 2003 PERVASIVEcomputing 9 EDUCATION & TRAINING EDUCATION & TRAINING

SAMPLE PROJECT DESCRIPTIONS (cont’d) I also provide a few supplemental read- ings. This exposes many students to extensive readings from the research lit- Continued from p. 9 erature for the first time. To help them with their reading, I require them to Electronic buttons for e-textiles write a brief summary of each paper, This project aims to devise a method for reliably and cheaply connecting a printed circuit submitted via email at the beginning of board button to a textile. Soldering alone won’t suffice due to mechanical problems and the week. I also ask them to submit a list other incompatibilities. Even if you could solder, you’d still have to line up the button’s con- of questions about the readings, which I nections with fabric’s threads. Students will need to explore other options for attaching the try to work into the lecture if possible. battery, for example, by modifying the connectors that ribbon cables use or modifying regular The class meetings are meant to be con- snap buttons used for clothing. versational, and I encourage students to For this project, I’ll give students an audio circuit (microphone plus op amp filter) that they’ll ask questions and make comments. Con- have to implement on a button-sized PCB and connect to an e-textile. Other desirable buttons sequently, the discussion often follows include LED, display, and with analog/digital buttons. Students should answer tangents to the prepared lecture, but they questions including are usually fruitful, informative, and thought provoking. I’ve even incorpo- • What are your options for connecting the button to the fabric—the relative advantages and rated “tangents” from the course’s first disadvantages of each? offering into the prepared material for • What is the right protocol for the button to communicate with the fabric (for example, the second offering. For example, a ques- Firewire or USB)? tion about how CPU speed-setting poli- • How many connections to the fabric can a button reasonably support? cies would handle a certain situation led • How big can a button reasonably be before the mechanical properties (flexibility, strength me to prepare an extended example of attachment, and so on) become unacceptable? comparing the policies’ behavior. During the last few weeks of the course’s first offering, I no longer required studies cover a range of applications, mented that they would have preferred reading summaries, to give students more from wearable computers for UPS ware- more depth on various topics. However, time to focus on the design projects. houses and train maintenance to aug- the comments differed on which topics Unfortunately, this brought on a notice- mented reality for conferencing and should have received more depth, so the able decline in the amount of dialogue. modeling outdoor environments. Some coverage is probably about right for an Although the students were initially grate- applications were widely deployed, with introductory course. ful for the lighter workload, one of the tens of thousands of units shipped, while most common comments in the course others were deployed only for limited COURSE DETAILS evaluations at the end of the semester was field trials. These applications show I base the student’s grade for the that the summaries should have been how the topics previously covered in the course on a group design project (40 required throughout the term. As one stu- course come together in a particular percent), an individual research report dent said, “Continue to require weekly design. They also show how practical (25 percent), an examination (20 per- reading summaries. I found it harder to issues such as ergonomics impact an cent), and homework and reading sum- keep up [with the readings and class dis- implementation and require considera- maries (15 percent). The individual re- cussion] when they weren’t required.” tion before ubiquitous computing can search report is a survey paper that become truly ubiquitous. Other appli- targets the class as its audience. I DESIGN PROJECTS cations, particularly in augmented real- encourage students to find a topic re- For the design project, students work ity, are currently research demonstra- lated to their own research (such as in groups of two to four people. I base tions, but they illustrate potentially radio frequency circuitry, wireless net- their project grade on weekly written attractive future applications. working, and embedded systems), and progress reports, an oral presentation, a The first time we offered the course, many students use it to find back- demonstration, and a final written report. I included sections on wireless net- ground material for a thesis. A few weeks into the course, I hand working and privacy issues. But I didn’t No textbook adequately covers the out descriptions of possible projects. have enough time to fit in all of the top- course’s range of topics, so all reading The students have a week to look over ics. Even without covering networking assignments come from journals and the project descriptions before forming and privacy, the course’s pace is diffi- conference proceedings. Students must teams. No two teams can work on the cult, and many students have com- read three to four papers per week, and same project. The teams choose from

10 PERVASIVEcomputing http://computer.org/pervasive EDUCATION & TRAINING

about six to eight projects, but only those projects (typically four or five) that have enough students interested in them to form a team go forward. I make the project descriptions inten- tionally vague for two reasons: First, it’s good experience for the students to iter- ate on a specification with a customer (in this case, me). Second, it gives them considerable leeway in making design decisions. Having too specific descrip- tions would force students down a design path that they might not choose on their own. The related sidebar shows samples of project descriptions. The projects tend to relate to research Figure 1. A student wearing a prototype vest-based wearable computer for emergency performed on campus. For example, I response personnel, RANGER (Rapidly Accessible Network for Geospecific Emergency have based several projects on work by Response), while rescuing another from a “sleep emergency” during the final days of Virginia Tech’s electronic textiles group their project. (www.ccm.ece.vt.edu/etextiles), which is currently focusing on e-textiles for wear- able computing and a related design tool 600 framework. A few projects have sprung from the research topics of students in the course. Notable projects include 500

• A -based personal authen- tication device 400 • A wearable computer for emergency first responders and field command- ers (see Figure 1) 300 • A model of an e-textile garment for mapping a building as the user walks 200 through it (see Figure 2) • A model of a garment that can detect its own shape 100 • Electronic buttons that permit cheap and reliable connection of electron- ics to e-textiles (see Figure 3) 0 0 50100 150 200 250 300 350 Each week, students must submit written progress reports (so they don’t Figure 2. Simulation results for a building-mapping garment showing a room’s walls put work off until the last minute) and and the garment’s current approximation for them. part of one class period is spent meet- ing as a group with me. At the start of the projects, these meetings focus on weekly meetings begin focusing on spe- cludes test and measurement equip- improving the problem description. cific problems they need to solve for the ment, notebook computers, PDAs, and Early on, I advise students to break the next week. wireless network cards. Students have project into smaller parts that individ- With a relatively short design cycle to plan for the lead time needed to pur- uals in the group can handle and to (10–12 weeks), students typically use chase a component and for what they carefully delineate the interface be- off-the-shelf components for prototyp- can do while they’re waiting for it to tween the parts for easy integration ing. Each group has a small budget and arrive. They often find that components later. As the description develops, the access to my research lab, which in- don’t work as advertised or have inter-

JULY–SEPTEMBER 2003 PERVASIVEcomputing 11 EDUCATION & TRAINING EDUCATION & TRAINING

tions. Similarly, for context-awareness, a mini project might involve having the students use sets of sensors to determine a user’s current activity or location. Another area for improvement would be to make the course more multidisci- plinary. Very real issues of industrial design, ergonomics, human-computer interaction, security, and economics come into play in this field. I discuss (a) (b) such issues briefly during the lectures when appropriate, but the students Figure 3. Electronic buttons for e-textiles: (a) unpopulated and (b) finished on an would benefit from more detailed, con- e-textile sweater. centrated treatments. Today’s students will be tomorrow’s engineers and operability problems with the rest of ument. The design document discusses researchers; they must be ready to their system or that they overlooked an the project’s major design decisions and tackle all the issues if ubiquitous com- important detail. Many projects require trade-offs, product-feature matrices for puting is to fulfill its promise. using state-of-the-art hardware and major components (both hardware and that is not well tested, sup- software), and test methods and results ported, or stable. These practical issues for both individual subsystems and the REFERENCES provide valuable lessons for students overall system. I also ask them to write 1. M. Weiser, “The Computer for the 21st that they don’t learn from reading a section entitled “If I could do it all Century,” Scientific American, vol. 265, no. research papers. It also gives them a over again.” 3, Sept. 1991, pp. 94–104. greater appreciation for the work 2. M. Weiser, “Some Issues behind research papers that describe the in Ubiquitous Computing,” Comm. ACM, design, implementation, and deploy- first offered the course at the 4000 vol. 36, no. 7, July 1993, pp. 75–84. ment of tens or hundreds of prototypes I level, which is for both undergraduate 6 3. M. Satyanarayanan, “Pervasive Comput- (such as Itsy and Xerox PARC’s tabs, (seniors) and graduate students. I ing: Vision and Challenges,” IEEE Per- pads, and boards2). intended the course to be a mixture of sonal Communications, vol. 8, no. 4, Aug. Despite widespread gnashing of teeth seniors and graduate students, either 2001, pp. 10–17. and sleeplessness in the last few days split evenly or having slightly more 4. T. Starner, “The Challenges of Wearable before project demonstrations, students graduate students. I ended up, however, Computing: Part 1,” IEEE Micro, vol. 21, often feel that the project is one of the with mainly seniors. While seniors for no. 4, July/Aug. 2001, pp. 44–52. best parts of the course. As one student the most part performed well in the 5. T. Starner, “The Challenges of Wearable put it: “I liked the idea of working on course, a number of them had difficulty Computing: Part 2,” IEEE Micro, vol. 21, group projects … as it allowed for some handling the pace and the required read- no. 4, July/Aug. 2001, pp. 54–67. sort of interest that isn’t totally present ings. I taught the second course offering 6. W. Hamburgen et al., “Itsy: Stretching the in pre-made engineering projects.” at the 5000-level, which is for graduate Bounds of ,” Computer, Once students complete their project, students only, and we were able to sus- vol. 34, no. 4, Apr. 2001, pp. 28–36. they must demonstrate it, make an oral tain a higher pace of study. Seniors can 7. C. Narayanaswami et al., “IBM’s Linux presentation, and submit a final writ- still take the course if they have a high , the Challenge of Miniaturization,” ten report. The report has two major enough GPA. Computer, vol. 35, no. 1, Jan. 2002, pp. 33–41.EXT ISSUE pieces, a user’s guide and a design doc- To improve the course, I could reduce the group design project’s scope and add one or two individual mini proj- next issue ects. These mini projects would permit more depth in certain topic areas. For Tom Martin is an assistant -Scale Sensor example, numerous architectural sim- professor in the Bradley De- Systems: Design and Policy at ulation tools exist for estimating soft- partment of Electrical and Carnegie Mellon University ware energy consumption that I could Computer Engineering at use to reinforce concepts on low-power Virginia Tech. Contact him compilation and source code modifica- at [email protected].

12 PERVASIVEcomputing http://computer.org/pervasive