Information and Communication Technologies (ICT)

Grant Agreement Number : 216366 Project Acronym : Euro-NF Project Title : Anticipating the Network of the Future – From Theory to Design Funding Scheme : Network of Excellence Start Date of Project : 01 January, 2008 Duration : 54 months (originally 36 months)

Deliverable Number : D.SEA.10.1.3 Version Number : 1.0 Title of Deliverable : Final update of the Euro-NF vision regarding the network of the future Contractual Due Date : April, 2012 Actual Date of Completion : June, 2012 Workpackage contributing to the Deliverable : SEA 10.1 – Euro-NF Vision Lead Contractor for this Deliverable : UNI WIEN (partner 04) and BTH (partner 11) Editor{s} : Kurt Tutschku and Markus Fiedler

Nature of the Deliverable : (R/P/D/O)* R Dissemination Level : (PU, PP, RE, CO)** PU

*Nature: R-report, P-prototype, D-demonstrator, O-other **Dissemination: PU-public, PP-restricted to programme, RE- restricted to a group, CO-confidential

Project URL: http://www.euronf.org

Project FP7 ICT N° 216366 – Euro-NF 8 June 2012 Deliverable N°: D.SEA.10.1.3

With contributions from the following partners:

Partner Partner Contributor Name Contributor e-mail address Number Acronym 01 GET Daniel Kofman [email protected] 01 GET Jean-Paul Lefèvre [email protected] 04 UNI WIEN Kurt Tutschku [email protected] 10 UWUERZ Phuoc Tran Gia [email protected] 11 BTH Markus Fiedler [email protected] 13 AUEB-RC George Polyzos [email protected] 30 INESC-ID Augusto Casaca [email protected] 34 AU Wolfgang Kleinwächter [email protected]

FOREWORD:

The so-called Euro-NF vision is the result of a long-term continuous brainstorming initiated during the former project Euro-NGI (a Network of Excellence supported by the European Commission through FP6 between 2003 and 2012). An initial document [1] was produced in May 2006, and then regularly discussed and updated until the release of the First Update [2] in July 2009, followed by a Second Update [3] in December 2010, focusing on architectural issues and (r-)evolutionary approaches. The present document can be seen as a snapshot of the evolution of our views on how to extrapolate and anticipate future networking technologies and challenges, rather than providing a detailed and technically exhaustive vision. Thus, it has to be considered as complementary to its predecessors and is labelled with a version date. Moreover, being extendable and adaptive to trends, which cannot currently be foreseen, is a major feature of the Euro-NF vision and reflects also the Euro-NF ability to address emerging networking issues in a very timely manner. On this background, merely examples of future technological focus areas are provided.

[1] Euro-NGI Network of Excellence, FP6 Project N° 507613, Deliverable N° D.WP.SEA.10.1.1, “A View on Future Communications”, May 2006, available at: http://eurongi.enst.fr/p_en_Publicat_ngiview_366.html

[2] Euro-NF Network of Excellence, FP7 Grant Agreement N° 216366, Deliverable N° D.SEA.10.1.1, “First Update of the Euro-NF Vision Regarding the Network of the Future”, July 2009, available at: http://euronf.enst.fr/p_en_Publicat_ngiview_366.html

[3] Euro-NF Network of Excellence, FP7 Grant Agreement N° 216366, Deliverable N° D.SEA.10.1.2, “Second Update of the Euro-NF Vision Regarding the Network of the Future”, Dec. 2010, available at: http://euronf.enst.fr/p_en_Publicat_ngiview_366.html

Dissemination level: PU 2 Project FP7 ICT N° 216366 – Euro-NF 8 June 2012 Deliverable N°: D.SEA.10.1.3

CONTENTS

1 THE FUTURE – THE FOUNDATION FOR SMART APPLICATIONS 4

2 THE NEED FOR A NEW COMPREHENSIVE RESEARCH AGENDA 6

3 THE eFIRA CONCEPT 8

4 TECHNOLOGIES AND SCIENTIFIC BACKGROUNDS ADDRESSED BY THE eFIRA CONCEPT 9

5 THE TECHNOLOGIES ADDRESSED BY eFIRA 12

LITERATURE 17

Dissemination level: PU 3 Project FP7 ICT N° 216366 – Euro-NF 8 June 2012 Deliverable N°: D.SEA.10.1.3

eFIRA: A NEW RESEARCH AGENDA FOR THE FUTURE INTERNET NETWORK MECHANISMS THAT MASTER THE INFORMATION FLOWS BETWEEN SMART APPLICATIONS AND NETWORKS

VERSION: June 8th 2012

1 THE FUTURE INTERNET – THE FOUNDATION FOR SMART APPLICATIONS

Users will consider future networked systems as a web of smart applications, services and content. The smartness of such applications is based on clever, intelligent, and almost human-like combination of information and services. In this way, the future networked systems will bring yet unforeseen benefits and values to stakeholders such as end users, municipalities, governments, service providers, and network operators.

Contributions from the network itself to the smartness of future applications might range from low to essential. The necessity for these contributions might depend on the needs of stakeholders (e.g. users are typically not interested in network efficiency as operators do), on the requirements of the applications and of course on the applied network technology. The increased understanding and application of newly developed network paradigms and design patters, such as Network Virtualization [And05, Rixn08], Network Federation [Spyr07, Tuts10], Software-defined Networking (SDN) [McKe09, Gree09, Stae11], Service-Oriented [Khon10, Müll08], Cloud [Hwan11], Over-the-Top (OTT) Services [Gree07, Eric10, Rebb11], Delay Tolerant Networking (DTN) [Fall03], or Ad-hoc and Mesh Networks [Muth04, Zhen09], however, asks for new and comprehensive knowledge in order to understand how networks and applications can collaborate for creating smart features of future applications. To achieve this knowledge, to enhance today’s and to develop future networking mechanisms we need a new research agenda for the Future Internet, which we call the agenda the European Future Internet Research Agenda: eFIRA.

Amongst others, the important questions of the Future Internet research agenda are, whether the end-to-end (E2E) principle of today’s Internet [Salt84] will still be the governing networking architecture of the future or how should it be enhanced? eFIRA assumes that the highly successful E2E design pattern is challenged by new technologies, e.g. by DTN and SDN, as well as by economical needs, e.g. how will OTT services cover the costs of the data transport infrastructure? The complexity of this discussion is highlighted by the typical misperception of the E2E principle as the “dumb network, smart end-systems” concept that might imply a strong separation of networks and applications. The original E2E paper makes a more subtle statement, claiming nothing more but that correctly implemented solutions for data transport can only be achieved by the support from end points. Thus, the original E2E paper explicitly did not exclude network-based mechanisms for enhancing the performance of data transport control, cf. [Salt84, Davi09].

Dissemination level: PU 4 Project FP7 ICT N° 216366 – Euro-NF 8 June 2012 Deliverable N°: D.SEA.10.1.3

The discussion above reveals also that the locations of control functions are essential for the overall efficiency of the data exchange, i.e. for the collaboration of applications and networks, or between application and network layer control mechanisms respectively. Thus, semantics, content and processing of information flows determine the performance of current networks and applications and of course of future applications.

The Euro-NF project understands by “information flows” the generic term for data flows of user data or of signalling information. Furthermore, the Euro-NF NoE distinguishes between “data information flows” and “network information flows”. Data information flows are those actual data flows in the typical data plane, or its equivalents. The “network information flows” are flows of any kind of signalling information but also any flow of information about involved networks, their capabilities regarding data transport, and related boundary conditions. In particular, the Euro-NF project assumes that network support for smart applications needs to clearly understand the direct/indirect and explicit/implicit network information flows and their semantics. Hence, it is necessary to model the flows with appropriate scope and level of detail. More details on the modelling concepts and their semantics will be outlined in Section 5.

The above outlined brief analysis leads to the definition of the three key hypotheses of eFIRA for main topics of Future Internet research: a) The handling of “network information flows” becomes as important as handling the “data information flows”; and b) The new networking paradigms have more complex information flows than design patterns using the current layering architectures, in particular the ISO/OSI model and the TCP/IP reference architecture, cf. [Day08]. Thus, they require new methods, procedures, mechanisms, protocols and architecture for processing the information flows in networked systems; and c) The new software architectures of the Future Internet core allows for these new kinds of processing of information flows.

Thus, the main target of eFIRA should be:

 to establish an understanding of the information flows in current and future network paradigms and design patterns in a more comprehensive way than today, as an enabling basis for future smart applications.

The deeper knowledge in this area will lead to

 the design of new networking architectures, protocols, standards and mechanisms for mastering these information flows between smart applications and networks, and  the development of new methods to design, evaluate and validate these new networking techniques.

Dissemination level: PU 5 Project FP7 ICT N° 216366 – Euro-NF 8 June 2012 Deliverable N°: D.SEA.10.1.3

2 THE NEED FOR A NEW COMPREHENSIVE RESEARCH AGENDA

In addition to the technology challenges above outlined, one can observe that networked systems are increasingly impacted by disruptive technology changes at fast pace and with decreasing predictability. Examples for such changes are the rise and fall of P2P content distribution [Azzo03, Kara04], the current dominance of HTTP-based streaming [BBC11] or the recent success of smartphone platforms [benA10, Chon10].

The trend of disruptive changes is largely founded in the complex relationships in ICT systems. Game changing concepts, like single technology quantum leaps such as high capacity optical transmission [Mori11, Sano11] or intelligent mesh-ups of technologies [Bloo10], might immediately impact the interplay of components and thus the functionality of ICT systems. In the end, local or small changes may lead to surprising innovations by the system.

This trend is expected to even accelerate in the future, since ICT systems are becoming more complex and more integrated into daily life. Furthermore, the trend challenges the conventional concepts for researching networked systems. Conventional research strategies are typically based on long-term strategies, whereas the recent innovations have happened frequently and on short-term. As a result,

 new approaches for research strategies in this domain are needed, that are able to adapt quickly to new technologies or technology mesh-ups.

The core assumption of Euro-NF project for eFIRA is that an adaptive approach is required for the investigation of future complex networked systems. In particular,

 a research agenda is needed that can adapt and react quickly to emerging networking technologies and technology mesh-ups.

The complexity and diversity of networked systems requires interdisciplinary and competitive research for obtaining comprehensive solutions. Solutions of singular problems in networked systems, such as the integration of new physical layers, are difficult and require high amounts of work, but the results can typically be achieved since the solution space is determined. Solutions of deep and comprehensive problems, however, are more difficult since the solution space is virtually unlimited. Such comprehensive solutions often need the comparison of different kinds of solutions to address a problem completely. Comprehensive problems, however, arise promptly in complex networked systems. A very illustrative example might be the feature of “resilience” or “reliability” in networked systems. P2P files sharing systems [Tuts08] and overlays [Ande01] have demonstrated that application- specific resilience or reliability can easily be implemented in the application layer. P2P reliability mechanisms outperform even mechanisms on network layer [Tuts08]. Thus, an important question is where to locate these resilience/reliability functions at best for a specific application? Hence experts of network layer and application layer have to compete with their investigation or to discuss how the different resilience concepts that complement or cooperate with each other. Such

Dissemination level: PU 6 Project FP7 ICT N° 216366 – Euro-NF 8 June 2012 Deliverable N°: D.SEA.10.1.3

 complementing and competitive investigations can only be performed in the collaborative research agenda, where different teams compare different solutions using their different experiences and research methodologies.

Furthermore, the investigations of networking concepts for future smart applications require a multidisciplinary research approach that will involve experts from the domains of applications and networks. The research agenda for the Future Internet should address this requirement by including complementing teams from both domains. In this way,

 the eFIRA agenda might achieve new forms for integration of networks and applications which consider the system as a whole, i.e. eFIRA-based research can address the application and the network requirements jointly.

Thus, the eFIRA concept enables the finding of trade-offs between partial solutions and it avoids isolated solutions that might be replaced quickly, such as specific physical networks for specific applications that do not last long.

In addition, by the comparison of solutions and methods from different areas,

 the eFIRA concept will lead to the development of integrated methodologies for the collaboration of networks and applications.

By the approach of integrated Future Internet research,

 the eFIRA concept might produce quickly results on the identification of network and service requirements by future smart applications, and thus,

 the eFIRA knowledge will pave the way for the economical exploitation of future smart applications by the European Internet economy and their stakeholders, i.e. users, software developers, services providers, or network operators.

Furthermore, the structured knowledge obtained in the eFIRA agenda can be considered as results of pre-standardization work, which means that

 eFIRA concept is able to feed its knowledge to standardization processes for network technology for future smart applications, for example by initiating workshops and seminars in conjunction with standardization bodies such as ETSI.

All in all,

 the eFIRA concept for Future Internet research implements a highly dynamic, flexible, adaptive, interdisciplinary, complementing and competitive research approach to generate new knowledge for the collaboration of networks and smart application and for understanding the information flows among them.

 a future eFIRA concept should gathers a focused group of major European research major institutions in the domains of networking and networked applications to develop future networking mechanisms, evaluation models and methods for future smart application.

Dissemination level: PU 7 Project FP7 ICT N° 216366 – Euro-NF 8 June 2012 Deliverable N°: D.SEA.10.1.3

3 THE eFIRA CONCEPT

The eFIRA agenda aims at a systematic design and engineering approach that considers networks and applications as a whole. Thus, the research work integrated by eFIRA will address networking hardware and software, but also the needs of applications, users and stake-holders such as personal usage, societal and business impacts, or ecological constraints. Moreover, the eFIRA approach aims at not only providing foundations for research on future networks for smart applications, but also at engineering of systems to become operable in real-world environments. These aims lead to the question, how can such a systematic research and engineering approach for mastering information flows between smart applications and networks be facilitated?

In order to enable a new systematic design paradigm for mastering the information flow between smart applications and networks, the eFIRA agenda assumes that the Dijkstra paradigm of the separation of concerns, cf. [Dijk82], has to be re-applied to the design pattern [Day08] and building blocks of future systems as well as to the design methodologies for networks and smart applications. We suggest that the usage and technology areas can be separated in three areas of concern for usage and technology, cf. Figure 1:

 Future smart applications (such as future environmental sensing networks or future intelligent transportation systems, cf. Section 4 below);

 Future smart mediation techniques (such as former routing tasks, now enabling mediation for Publish/Subscribe techniques, delay-tolerant networking, application- specific topologies and resource management);

 Future smart connectivity techniques (such as convergence of high speed optical and wireless transmission and energy efficient transmission techniques).

The separation of design methodologies for future networks has to be changed in the light of the new requirements from smart applications in order to facilitate a coherent systematics and the holistic nature of the design. Such a separation might divide the design methods into:

 New design methods for networking architectures, e.g. which separation is appropriate (e.g. layering vs. heaps);

 New methods for comparing and evaluating architectures, e.g. for comparing different Future Internet architectures in term of new qualitative (e.g. flexibility, adaptivity, or expandability) and quantitative metrics (e.g. the quality of cooperation, quality of pricing, quality of protection);

 New design methods for smart (e.g. models for self-organization, smartness and autonomy);

 New design methods for including socio-economic needs in future networks, e.g. for assessment of reliability, resource efficiency, security and Quality of Experience, or methods for incorporating network and application governance concepts as well as business models into network mechanisms.

Dissemination level: PU 8 Project FP7 ICT N° 216366 – Euro-NF 8 June 2012 Deliverable N°: D.SEA.10.1.3

s Future Smart a

e

r Application d A n

a y

g Future Smart e o g l a o Mediation s n h U c Future Smart e T Connectivity s g d n - i e

e d

e g u s g m l n n s

s i

d t n r c t .

i n n n e e o e r o n a r r n g i f s g g g h n

i i i a

u u u o n t t

l r t t i s s s s s p r m c e e a o c c k e e e d d a r f h e r v m

t e e - d m d d i o o u t t o

e s i i o m r t o

h h i c s w d t t h h o u w w w w

t d f c c c r r o e e g e e e e e r n r l o o o h n N m n a N f a a N m f a N t s i

Design methods Figure 1 - eFIRA’s conceptual matrix for interconnecting technologies for smart applications with networks with the appropriate design methods

Thus, the concept of the research and integration work of the eFIRA agenda is to weave a new fabric for combining future technologies and new design methods for mastering the information flows between smart applications and networks. A first definition of the threads of this fabric is given by rough and generic specification of the new areas of the separations given above. These threads do not exclude an evolution of today’s successful network paradigms such as the E2E concept. These threads will contribute to a new understanding of the separation of concerns.

Technically, the weaving of the fabric is materialized by enablers, which are physical entities and mechanisms or intellectual methods and algorithms for the coordinated operation of future networks and future smart applications. In particular, the methodologies to achieve smart combinations of future networking and application technologies for supporting smart applications are at the core of the eFIRA concept.

4 TECHNOLOGIES AND SCIENTIFIC BACKGROUNDS ADDRESSED BY THE eFIRA CONCEPT

The eFIRA agenda expects that the main usage domains of smart applications are within everyday life. Examples domains are health services; energy usage and distribution; environmental sensing, monitoring and analysis; logistics and transportation; and personal entertainment, [EC09]. The embedding of future smart applications in daily life will provide a basis for reinforced social, economical and ecological stability, as required by internationals societies such as the European Commission [EC10a, EC10b]. The embedding makes the smart applications a true commodity for everyone.

However, being a commodity increases the implicit and explicit requirements on applications and networks, such a , safety, privacy, security, ease-of-use, ease-of-operation, efficiency, or performance. A comprehensive description of the requirements of future smart applications however, is that at this point of time practically not feasible and perhaps even not advised. The major difficulty is that future applications and thus their requirements are very difficult to predict. The eFIRA approach to resolve this deadlock is to establish strong collaborations with users and industry to identify future smart applications, identify their requirements, spread this knowledge, and develop network mechanisms based on this

Dissemination level: PU 9 Project FP7 ICT N° 216366 – Euro-NF 8 June 2012 Deliverable N°: D.SEA.10.1.3 knowledge. However, in order to understand current main requirements from future or smart applications, two examples of them are outlined next.

Future environmental sensing networks are a first example. They permit applications from accurate, local weather forecasts for farming to live, worldwide maps of diseases and human wellness. A successful illustration for them is the ViSE (Virtualized Sensing Environment) project at the University of Massachusetts - Amherst, cf. [Kuro08]. It has demonstrated how to provide access to controllable virtualized sensors, e.g. steerable and Pan-Tilt-Zoom video cameras, which export rich actuation capabilities and high-bandwidth data. ViSE demonstrated also the federation of local sensors with the national, large-scale GENI test facility for permitting end-users and researchers end-to-end control for sensor experiments and applications. Such sensing applications might not be limited to environmental applications. They also can be in industrial environments for remote control of manufacturing processes, cf. [SSSC11].

Such sensing applications combine in an intelligent way distributed high-volume data sources (e.g. HD images or weather webcams), small volume but large-scale sensor information (e.g. temperature), and real-time requirements for actuator control with smart information processing. The example of sensing networks also outlines that information processing will take place at very diverse locations. The range of architecture options for processing varies from highly centralized to extremely distributed, as in peer-to-peer computing. Integrating very different types of data sources with highly diverse quality-of-service requirements for data transport is an extremely challenging task in today’s networks. Solving this task for future network and smart application is challenging task.

Intelligent Transportation Systems (ITS) using Vehicular Ad hoc NETworks (VANETs) are becoming a reality. ITS is a key topic for both the research community and the industry, including vehicle manufacturers and the industry. In particular, VANETS that are enhanced by DTN concepts and termed as Vehicular Delay-Tolerant Networks (VDTNs) envision the capabilities of the support of networks for smart applications. They may implement in short time services that have been proposed already for a long time, for example: safety applications, such as collision avoidance, hard-break notifications or even road condition information; information applications that can provide data dissemination to determine, for example, the best routes to decrease traffic congestion; or traffic light control. On another side, an efficient planning of these systems, with fixed or mobile stations giving support to vehicular networks, in a hierarchical approach, may envision the support of entertainment applications, namely content distribution related to advertisement, video streaming and even multi-player games between passengers. A key project in this area is the Drive-In VANET project, cf. [Barr09]. The goal of DRIVE-IN project is to investigate how vehicle-to-vehicle communication can improve the user experience and the overall efficiency of vehicle and road utilization. In the future, however, such networks will be operated not as isolated, single purpose network islands. They rather will be opened to multiple applications and will be integrated and interworking with other communication networks.

Summarizing the above given examples, an initial characterization of major data, processing and communication features for future smart applications can be given. Future smart applications are characterized by:

Dissemination level: PU 10 Project FP7 ICT N° 216366 – Euro-NF 8 June 2012 Deliverable N°: D.SEA.10.1.3

 using very different types of data, e.g. from very low volume sensor data to interactive video streams and actuator information with strict real-time requirements or very high data volumes;

 flexibility, elasticity and adaptation in the use of processing resources in order to implement arbitrary workflows on data, content and services, e.g. permit seamless transition between centralized processing (cloud, client-server) and peer-to-peer computing;

 the use of very different transmission modes for data, even the use of them in parallel (yet any mix of intermittent and delay tolerant, multi-hop mesh, packet- and circuit- switched transmission modes is permitted);

 being agnostic whether wireless, mobile or wired connections are used;

 being able to interact with infrastructure for improved data collection/content distribution;

 smart provider and connectivity selection (e.g. based on economic and ecological needs);

 machine-to-machine communication that requires highly scalable and stable mechanisms (up to billions of communicating devices);

 the use of application-specific virtual network, with own topologies, naming schemes, and routing and resource management techniques, that can be embedded in more general network infrastructures;

 highly autonomous operation for reduced human interaction in processing and the operational loop;

 considering the system as whole, i.e. being able to integrate multi-disciplinary approaches for information processing and ;

 being easily able to addresses the different socio-economic interests of the different stake-holders in the ecosystem of networked applications.

Finally, future smart applications have to fulfil of course long-time desired performance features, such as being highly reliable and assuring transmission quality. The absence of the latter features in today’s network increases even further the expectations and the pressure on the future systems for being novel and smart.

Dissemination level: PU 11 Project FP7 ICT N° 216366 – Euro-NF 8 June 2012 Deliverable N°: D.SEA.10.1.3

5 THE TECHNOLOGIES ADDRESSED BY eFIRA

The eFIRA agenda assumes that the support of future smart applications rely closely on the availability and combination of new and even old network paradigms, cf. Section 1. The integration of the paradigms into future applications is achieved by designing of the future Internet core as a new software architecture with variable and adaptive capabilities for information flow processing.

The evolution of information flows in network architectures is outlined in Figure 2. Figure 2(a) depicts the ideal layer concept of ISO/OSI reference model [Zimm01]. This well-known model assumes a linear, vertical information flow between adjacent layers with predefined information processing at the layers according to the functions provided by them. Figure 2(b) indicates the growth of functionality in certain layers in the TCP/IP reference model [Brad89]. This growth leads to the overload of IP layer, denoted as the “fat waist” this architecture [Deer01, Auir08]. However, the vertical data information flow is still maintained in the TCP/IP model. Recently, an analytical model was provided [Akhs11] that explains why a layered protocol stack evolves to a hourglass-shaped architectures as depicted in Figure 2(c). The model also outlines that alternative protocols dependencies and the related adaption of the processing of flows might lead to network architectures with higher efficiency. The new separation of concerns from Section 1 is shown in Figure 2(d). This separation leads to distinguish responsibilities for creating, maintaining, and providing services in an architecture for networked systems. The architecture considers the domains of “Application” (representing applications or re-useable services), “Mediation” (mapping of application demands to transport capabilities), and “Connectivity” (represent services to related transport technologies), cf. [Tuts07, Tuts08, Müll11].

ISO/OSI Model , www, VoIP, .. Application Model description EuroView Application SMTP, HTTP, RTP, ... Architecture Presentation L (2007) TCP, UDP, ... Protocol …

Session DiffServ dependencies

as edges Application . IP/MobileIP 4 Products: P(u) t Transport m

IPsec 3 u Layer of u: l(u) g M Network Protocols 2 Substrates:S(u) Mediation , PPP, ... as nodes Data Link Copper, fiber, radio, ... 1 Physical Infrastructure Physical CSMA, sonet, async,.. Layered acyclic network Connectivity Every layer provides a service 11" (a) (b) (c) (d) (e)

Figure 2 - Evolutions of Networking Architecture While the architecture of Figure 2(d) is rather generic and might allow for non-linear information flow between layers, the information flow graph architecture in Figure 2(e) relaxes this requirement. It allows for adaptive, dynamic and non-linear workflows in the processing of information flows. This dynamic and adaptation feature can be derived from the concept of Service Oriented Architectures (SOA)1, cf. [Müll11]. eFIRA will validate the assumption whether such dynamically interacting services might be able to replace the concept of layers.

1 However, the SOA concept should not be set equal to Web Services. We consider SOA as an approach to design computing systems, while Web Services is a technology to implement such systems.

Dissemination level: PU 12 Project FP7 ICT N° 216366 – Euro-NF 8 June 2012 Deliverable N°: D.SEA.10.1.3

Moreover, the control and management of data information flows is not of sole importance. The handling of information about the network, its state and the quality of the data transport is significant for smart applications. Accessing and processing of such network information enables smart applications to operate truly smart. User data information and network information can be combined and the application can consider the status of the system as a whole. The simplest example of such a holistic concept is the knowledge of connectivity options for the current context of an application. The application may select the most reliable, cheapest or most ecological connectivity. However, the smart handling of network information requires a more complex information flow processing.

More evolved and complex application and network scenarios have emerged recently that require the combined processing of application data and network information flows. Such scenarios, for example, may comprise the use of cognitive radio technology for sensor networks or the application of polymorphic, virtualized or federated networks for providing new cloud services. In these scenarios, the semantic of network information has to be understood and modelled in sufficient detail. These models might comprise complex models for network topologies, network behaviour, resource availability, or for relating Quality-of- Service with Quality-of-Experience. Moreover, the understanding of application and network information may also require the use W3C’s semantic web techniques like Resource Description Framework (RDF), RDF schema (RDFS), or Web Ontology Language (OWL) for improved semantic descriptions and analysis. In addition, more complex matching mechanisms are needed for comparing the current and required state of a network system and to make appropriate making decisions for routing, flow or overload control.

Network information can easily be communicated directly between applications and networks when the information is of Boolean type, e.g. access is granted by the network to a resource. Many signalling protocols and architectures achieve such a direct communication in an elegant and highly efficiently way. Major examples for such signalling systems are SIP or IMS. Indirect exchange of network information between networks and applications is also often used. The TCP protocol can be consider as a prominent example. Here, the network indicates (i.e. “”) overload indirectly to the application-oriented TCP transport protocol by dropping packet. Upon recognition of the packet loss, TCP protocol may then perform an adaptation of the window size. Figure 3 outlines the data information flows between data producer and data consumer and summarizes the ways of direct and indirect communication of network information. The eFIRA agenda assumes that the mapping of the processing location of the data and network information flows is flexible.

In addition, the recent success of P2P-based file sharing and HTTP-based streaming have demonstrated that application-layer transport control mechanisms (source/P2P selection, flow and congestion control on application layer) which use indirectly network information, e.g. peer quality, may perform equally or even superior to conventional transport layer control mechanisms. However, these application-layer mechanisms may increase the decoupling of applications from networks. Although the decoupling might be a design paradigm and an aim for network architectures, little research work has been carried out a) to understand how the above outlined complex applications perform indirect or direct interaction with of the data transport mechanisms on the various levels or b) to understand where to place these information flow control mechanisms most efficiently. Conventional approaches of cross-layer optimization might often also proof inappropriate to the complexity of the scenarios.

Dissemination level: PU 13 Project FP7 ICT N° 216366 – Euro-NF 8 June 2012 Deliverable N°: D.SEA.10.1.3

More technical solutions for the “smart” mastering of information flows have emerged recent lately. However, an exhaustive and structured description of them can’t yet be achieved since this knowledge is yet missing. It is major aim of the eFIRA agenda is to achieve a classification of the concepts and to provide methodologies of how to combine them efficiently. Another prominent example are middleboxes in mobile networks, cf. [Wang11].

Figure 3 - Data Information and Network Information Flows These boxes perform in addition to security and NAT (Network Address Translation) functions, tasks like TCP state tracking, response filtering or packet mangling for influencing jointly the behaviour of networks and applications. Another recent result of relating the applications and networks was the identification of functional relationships between application-orient Quality-of-Experience (QoE) and network-oriented Quality-of-Service (QoS) performance metrics. The IQX hypothesis for an exponential dependency of QoE and QoS was validated in [Fied10]. Network Virtualization (NV) and Network Federation (NF) can also achieve co-operation of future networks and applications [Tuts09, Tuts10]. NV aims at the definition and concurrent operation of application-specific networks with own topology, naming schemes, routing and resource management. NV permit the parallel operation of multiple achieves in a single physical infrastructure and, thus, a vertical integration of networks. NF aims at the collaboration of sub-networks, which are willing to achieve the applications requirements. Currently, the collaboration of the network in the Internet is achieved by the BGP protocol. However, this protocol has to be reconsidered and eventually enhanced for future networks [Agap11].

However, a comprehensive understanding of these mechanisms in the light of today’s and future networks and applications hasn’t yet been achieved. The current solutions of handling data or network information flows are rather isolated. Their efficiency is often not investigated in the context of future applications. Thus, eFIRA is aiming at a comprehensive understanding, combination and processing of application data flows and network information flows. eFIRA targets at new mechanisms and techniques for mastering the information flow between smart applications and networks.

Smart applications will have higher chances of success if the new network architecture allows locating, customizing, reserving, provisioning and monitoring network resources in a smart, coherent and comprehensive way. To this end, the network architecture has to be built having such functionalities and concerns in mind in order to provide enhanced

Dissemination level: PU 14 Project FP7 ICT N° 216366 – Euro-NF 8 June 2012 Deliverable N°: D.SEA.10.1.3 control, monitoring and measurement planes that can be easily supported, easily extended (e.g. in response to technology shifts) and seamlessly integrated to effectively close the measurement-optimization-control cycle.

In this way, the eFIRA agenda might weave a new fabric of interconnections of application and networks and might answer the question, whether today’s and future network architecture are smart enough for future smart applications?

The research work in eFIRA should aim at the investigation of the following list of focus areas:

1) Virtualization and Federation for Polymorphic Networks Smart applications and services demand for well-adapted network infrastructures. Consequently, we observe a growing degree of polymorphism in NF, offering many different possibilities to allocate and integrate network resources. Network Virtualization [Ande05, Rixn08] enables the parallel operation (respectively consolidation) of multiple and possibly application specific networks in a single physical system and enables vertical integration. Network Federation aims at the combination of willingly cooperating network resources, also denoted as stitching [Falk10, Moti08], and is thus achieving horizontal convergence of diverse technical or administrative domains. The successful implementation of virtualization and federation builds upon performance classification of resources and quantification of their interaction as a basis for design and management of virtualized and stitched solutions.

2) Cooperative Algorithms for Seamless Wired and Wireless Connectivity A typical end user disposes of a set of network accesses. Seamless switches between mobile, wireless and wired technologies and operators, augmented by new communication paradigms such as mesh networks and cognitive radio has opened a tremendous choice of possibilities to make smart applications and services and thus users “Always Best Connected” (ABC) [Gust03]. However, this plethora of choices is a mixed blessing, as the meaning of “best” depends to a significant extent on content and context. Local, remote and cross-layer information need to be combined in order to yield optimal decisions, which means that the corresponding algorithms need to act in a distributed, cooperative way upon knowledge and comparison of static and dynamic characteristics of alternatives, thus enabling user-centric ABC [PERI10].

3) Engineering for Cyber-Physical Systems Cyber physical systems combine computing and networking power with physical components [Lee08]. Thus, they can be seen as a materialization of the paradigm in a wide range of domains, including robotics; smart homes, vehicles, and buildings; medical implants; and future-generation sensor networks. The efficient design and operation of such systems requires a clear view of the characteristics of the components and devices and their interplay, given the specific conditions under which the systems operate. Thus, modelling of components and their communication, and analysis and synthesis of information flows are of key importance in order to enable reliable and secure functioning.

4) Design and Performance of Network-located Enablers for Smart Applications We observe a growing diversity and multitude of network-located services (e.g. web, authentication, location, or billing but also routing-as-a-service or data fusion services) and specific entities (e.g. gateways, firewalls or cloud facilities). Intended as enablers for smart applications and services, they can easily turn into disablers when facing performance

Dissemination level: PU 15 Project FP7 ICT N° 216366 – Euro-NF 8 June 2012 Deliverable N°: D.SEA.10.1.3 problems, e.g. due to untimely reactions. Already the design part of these enablers, which includes definitions of architecture, roles, communication, etc., needs to carefully take care of the specific requirements of smart applications and services in order to proactively fight availability, reliability, performance and security issues. This requires a clear view on the impact of different design options, to be obtained from performance experience, experiments, studies and analyses. On the other hand, suitable control interfaces between enabler and applications need to be designed and deployed. Successful design needs to be accomplished by effective operations of the enablers, which includes appropriate reactions upon requests and carefully synthesized control loops.

5) Cyber Assurance Smart application and services must perform in a reliable and secure way in order to become accepted for mission-critical tasks, for which current “one-size-fits-all” Internet never had been designed. On the other hand, “smart” devices and applications have frequently shown their capability to overload network resources and to open the door to adversaries, with unsatisfactory user experience and performance or even economic damage as consequence. Obviously, a reliable functional mapping between user experience and network and service design and operation needs to be established. Such knowledge – preferably in form of generic and quantitative relationships between network, system and user performance and security – has to be advanced and disseminated in order to successfully guide the whole life cycle of applications, services and resources.

6) Multi-stakeholder approaches to governance in Networks of the Future So far, Internet governance (e.g. address assignment) has been dominated by a comparably small set of players, with concentration to the “birthplace” of Internet. This dominance is questioned by emerging stakeholders. For smart applications as well as for the Internet of Things, there is a fierce debate of which players shall be involved in setting the rules, which in particular refers to the Object Name System (ONS). Thinking of the plethora of issues related to addressing, accessibility, security and performance within smart applications and the polymorphic NF, there is a need to include multiple stakeholders such as manufacturers, policy makers and user organizations in the corresponding decision-making according to their place in the polymorphic network landscape.

7) Networking for sustainable energy use and sustainable energy use in networks It is claimed that ICT stands for 2 % of the carbon footprint [Gart2007]. The new trend in networking is “energy saving is in, over-provisioning is out”. It is expected that smart applications and services will enable energy savings in an energy-efficient way. The study [Eura08] mentions a potential of 15 % carbon savings until 2020, and it is challenging for future smart applications, services and networks to even improve this percentage. In order to yield such savings, ICT architectures in general and smart applications in specific, have to come closer to energy producers and consumers, both physically and logically. Energy spending for energy saving is a compromise in itself, which highlights the demand for energy consumption models, related to cost and performance models, and the corresponding assessment and optimization methods.

Dissemination level: PU 16 Project FP7 ICT N° 216366 – Euro-NF 8 June 2012 Deliverable N°: D.SEA.10.1.3

LITERATURE

[Agap11] A. Agapi et al., “Routers for the Cloud: Can the Internet Achieve 5-Nines Availability?,” IEEE Internet Computing, vol. 15, no. 5, pp. 72-77, Oct. 2011. [Agui08] R. L. Aguiar, “Some comments on hourglasses,” SIGCOMM Comput. Commun. Rev., vol. 38, no. 5, pp. 69–72, Sep. 2008. [Akhs11] S. Akhshabi and C. Dovrolis, “The evolution of layered protocol stacks leads to an hourglass-shaped architecture,” SIGCOMM Comput. Commun. Rev., vol. 41, no. 4, pp. 206–217, Aug. 2011. [Ande01] D. Andersen, H. Balakrishnan, F. Kaashoek, and R. Morris, “Resilient overlay networks,” SIGOPS Oper. Syst. Rev., vol. 35, no. 5, pp. 131–145, Oct. 2001. [Ande05] T. Anderson, L. Peterson, S. Shenker, and J. Turner, “Overcoming the Internet impasse through virtualization,” , vol. 38, no. 4, pp. 34- 41, Apr. 2005. [Azzo03] N. B. Azzouna and F. Guillemin, “Analysis of ADSL traffic on an IP backbone link,” in IEEE Global Telecommunications Conference, 2003. GLOBECOM ’03, 2003, vol. 7, pp. 3742- 3746 vol.7. [Barr09] J. Barros, M. Ferreira, S. Sargento, P. Steenkiste, and O. Tonguz, “DRIVE-IN – Distributed Routing and Infotainment through Vehicular Inter-Networking.” [Online]. Available: http://www.cmuportugal.org/tiercontent.aspx?id=1552&terms=drive-in. [Accessed: 12-Dec-2011]. [BBC11] BBC News, “Netflix dominates US net traffic,” 19-May-2011. [Online]. Available: http://www.bbc.co.uk/news/technology-13439641. [Accessed: 14-Dec-2011]. [benA10] D. ben-Aaron, “Android Phones Set to Pass ’s Symbian in Europe, IDC Says,” 29-Nov-2010. [Online]. Available: http://www.bloomberg.com/news/2010-11- 29/android-phones-set-to-pass-nokia-s-symbian-in-europe-idc-says.html. [Accessed: 14-Dec-2011]. [Bloo10] Bloomberg, “Bloomberg Game Changers: Steve Jobs - Video,” 10-Feb-2010. [Online]. Available: http://www.bloomberg.com/video/66625228/. [Accessed: 14-Dec-2011]. [Brad89] R. Braden, “RFC1122: Requirements for Internet Hosts -- Communication Layers.” IETF, Oct-1989. [Chon10] S. Choney, “Is 2010 the year of wireless congestion? - Technology & science - Tech and gadgets - Consumer tech,” 01-Apr-2010. [Online]. Available: http://www.msnbc.msn.com/id/34634571/ns/technology_and_ science- tech_and_gadgets/#.TujG3Jgf_Lo. [Accessed: 14-Dec-2011]. [Davi09] B. S. Davie and J. Medved, “A programmable overlay for service provider innovation,” in Proceedings of the 2nd ACM SIGCOMM workshop on Programmable routers for extensible services of tomorrow, New York, NY, USA, 2009, pp. 1–6. [Day08] J. Day, Patterns in Network Architecture: A Return to Fundamentals. Prentice Hall, 2008. [Deer01] S. Deering, “Watching the Waist of the Protocol Hourglass - IETF 51 Plenary,” London, Aug-2001. [Dijk82] E. Dijkstra, “On the role of scientific thought,” in Dijkstra, Edsger W.. Selected writings on Computing: A Personal Perspective, New York, NY, USA: Springer, 1982, pp. 60 - 66. [EC09] Commission of the European Communities, “Communication from the Commission to the European Parliament, the Council, the European Economic and Social Committee and the Committee of the Regions - A public-private partnership on the Future Internet - COM(2009) 479 final, Brussels, 28-Oct-2009.

Dissemination level: PU 17 Project FP7 ICT N° 216366 – Euro-NF 8 June 2012 Deliverable N°: D.SEA.10.1.3

[EC10a] Commission of the European Communities, “Work Programe 2011 – Cooperation – Theme 3 – ICT – Information and Communication Technologies - COM(2010) 4900, Brussels, 19-Jul-2010. [EC10b] Commission of the European Communities, “Communication from the Commission to the European Parliament, the Council, the European Economic and Social Committee and the Committee of the Regions – ‘Europe 2020 Flagship Initiative Innovation Union’ - COM(2010) 546final, Brussels, 06-Oct-2010. [EC11] Commission of the European Communities, “Updated Work Programe 2011 and Work Programme 2012 – Cooperation – Theme 3 – ICT – Information and Communication Technologies - COM(2011) 5068 - Brussels.” 19-Jul-2011. [Eric10] , “Voice over LTE - A Step Towards Future Telephony.” Ericsson, Dec-2010. [Eura08] euractiv.com, “Study sheds light on ICT sector’s carbon footprint | EurActiv,” 27-2008. [Online]. Available: http://www.euractiv.com/climate-change/study-sheds-light-ict- sector-carbon-footprint/article-173710. [Accessed: 24-Dec-2011]. [Fall03] K. Fall, “A delay-tolerant network architecture for challenged ,” in Proceedings of the 2003 conference on Applications, technologies, architectures, and protocols for computer communications, New York, NY, USA, 2003, pp. 27–34. [Falk10] A. Falk, “Stitching Slices - Presentation of the GENI Control Framework Working Group at GEC7,” Mar-2010. [Fied10] M. Fiedler, T. Hossfeld, and P. Tran-Gia, “A generic quantitative relationship between quality of experience and quality of service,” IEEE Network, vol. 24, no. 2, pp. 36-41, Mar. 2010. [Gart07] Gartner, “Gartner Estimates ICT Industry Accounts for 2 Percent of Global CO2 Emissions,” 26-Apr-2007. [Online]. Available: http://www.gartner.com/it/page.jsp?id=503867. [Accessed: 24-Dec-2011]. [Gree09] K. Greene, Tr10: Software-defined networking. MIT Technology Review, 2009. [Gree07] W. Greene and B. Lancaster, “Over the Top Services.” LTC International, 18-Nov- 2007. [Gust03] E. Gustafsson and A. Jonsson, “Always best connected,” IEEE Wireless Communications, vol. 10, no. 1, pp. 49- 55, Feb. 2003. [Hwan11] K. Hwang, J. Dongarra, and G. C. Fox, Distributed and cloud computing: clusters, grids, clouds, and the future internet. Morgan Kaufmann Publishers, 2011. [Kara04] T. Karagiannis, A. Broido, N. Brownlee, K. C. Claffy, and M. Faloutsos, “Is P2P dying or just hiding? [P2P traffic measurement],” in IEEE Global Telecommunications Conference, 2004. GLOBECOM ’04, 2004, vol. 3, pp. 1532- 1538 Vol.3. [Khon10] M. Rahamtullah Khondoker, Bernd Reuther, Dennis Schwerdel, Abbas Siddiqui, Paul Müller: Describing and Selecting Communication Services in a Service Oriented Network Architecture. In the Proceedings of the ITU-T Kleidoscope event, Pune, India, 13-15 Dec 2010 [Kuro08] J. Kurose, P. Shenoy, D. Ganesan, M. Zink, and D. Irwin, “ViSE - a Virtualized Sensing Environment,” 2008. [Online]. Available: http://vise.cs.umass.edu/trac/. [Accessed: 12- Dec-2011]. [Lee08] E. A. Lee, “Cyber Physical Systems: Design Challenges,” in 2008 11th IEEE International Symposium on Object Oriented Real-Time (ISORC), 2008, pp. 363-369. [McKe09] N. McKeown, Software-defined Networking. Keynote Talk at IEEE INFOCOM 2009. [Mori11] T. Morioka, M. Jinno, H. Takara, and H. Kubota, “Innovative Future Optical Transport Network Technologies,” NTT Technical Review, vol. 9, no. 8, Aug. 2011. [Moti08] M. Motiwala, M. Elmore, N. Feamster, and S. Vempala, “Path splicing,” SIGCOMM Comput. Commun. Rev., vol. 38, no. 4, pp. 27–38, Aug. 2008.

Dissemination level: PU 18 Project FP7 ICT N° 216366 – Euro-NF 8 June 2012 Deliverable N°: D.SEA.10.1.3

[Murt04] C. S. R. Murthy and B. S. Manoj, Ad hoc wireless networks: Architectures and protocols. Upper Saddle River, NJ: Prentice Hall, 2004. [Müll08] Paul Müller, Bernd Reuther: Future Internet Architecture – A Service Oriented Approach. In: it – Information Technology, Jahrgang 50 (2008) Heft 6, S. 383-389 6/2008 [Müll11] P. Müller, “Futuer Internet Architecture -- A Cloud Perspective: A Services-Oriented Approach,” Poznan, Poland, 26-Oct-2011. [PERI10] FP7 STREP Perimeter, “Perimeter - User-centric paradigm for seamless mobility in future internet,” 2010. [Online]. Available: http://www.ict-perimeter.eu/. [Accessed: 24- Dec-2011]. [Rebb11] T. Rebbeck and S. Sate, “The future of mobile voice: scenarios for market evolution.” Analysys Mason, 2011. [Rixn08] S. Rixner, “Network Virtualization: Breaking the Performance Barrier,” Queue, vol. 6, no. 1, pp. 37:36–37:ff, Jan. 2008. [Sano11] A. Sano, T. Kobayashi, E. Yoshida, and Y. Miyamoto, “Ultra-high Capacity Optical Transmission Technologies for 100 Tbits/s Optical Transport Networks,” IEICE Transaction on Communications, vol. E94-B, no. 2, Feb. 2011. [Salt84] J. H. Saltzer, D. P. Reed, and D. D. Clark, “End-to-end arguments in system design,” ACM Trans. Comput. Syst., vol. 2, no. 4, pp. 277–288, Nov. 1984. [Spyr07] T. Spyropoulos, S. Fdida, and S. Kirkpatrick, “Future internet: fundamentals and measurement,” SIGCOMM Comput. Commun. Rev., vol. 37, no. 2, pp. 101–106, Mar. 2007. [SSSC11] Scottisch Sensor Systems Centre, “Manufacturing and process control : Scottish Sensor Systems Centre.” [Online]. Available: http://sensorsystems.org.uk/markets/manufacturing- and-process-control/. [Accessed: 12-Dec-2011]. [Stae11] D. Staessens, S. Sharma, D. Colle, M. Pickavet, and P. Demeester, “Software defined networking: Meeting carrier grade requirements,” in 2011 18th IEEE Workshop on Local & Metropolitan Area Networks (LANMAN), 2011, pp. 1-6. [Tuts07] K. Tutschku, “P2P-based Network and Service Operation for the Emerging Future Internet,” Praxis der Informationsverarbeitung und Kommunikation, vol. 30, no. 2, pp. 68-75, 2007. [Tuts08] K. Tutschku, “Peer-to-Peer Service Overlays – Capitalizing on P2P Technology for the Design of the Future Internet,” Habilitation Thesis, Würzburg, Germany, Jun. 2008. [Tuts09] K. Tutschku, “Towards the Future Internet: virtual networks for convergent services,” e & i Elektrotechnik und Informationstechnik, vol. 126, no. 7-8, pp. 250-259, Jul. 2009. [Tuts10] K. Tutschku, “Towards a Consolidated and Sustainable Future Internet – Using Network Virtualization and Network Federation for Vertical and Horizontal Integration,” Sophia-Antipolis, France, Mar-2010. [Wang11] Z. Wang, Z. Qian, Q. Xu, Z. Mao, and M. Zhang, “An untold story of middleboxes in cellular networks,” SIGCOMM Comput. Commun. Rev., vol. 41, no. 4, pp. 374–385, Aug. 2011. [Zimm80] H. Zimmermann, “OSI Reference Model--The ISO Model of Architecture for Open Systems Interconnection,” IEEE Transactions on Communications, vol. 28, no. 4, pp. 425- 432, Apr. 1980. [Zhen09] P. Zheng, Wireless networking complete. Morgan Kaufmann, 2009.

Dissemination level: PU 19