Objectives

In the future, it is expected that mobile and wireless communications will play a central role in the lives of European citizens, providing the backbone for the knowledge economy. The industry sector will contribute substantially to the European business prosperity, and technology will greatly evolve from the current concept of anywhere, anytime to a new paradigm of any network on any device, right content in the appropriate context, in a secure manner. Great expectations start with the promise of a hefty increase in data transmission rates that enable real-time services addressed at areas of societal interest such as:

 Health care.
 Medical monitoring and diagnosis.
 Disability and senior citizen inclusion.
 Energy efficiency.
 Climate change.
 Security.
 Emergency services and first responders.
 Automotive industry applications.
 E-commerce.

We are facing a new era of distributed communications, where autonomous processing, communication, sensing, and actuating, are replacing traditional centralized architectures. These distributed architectures can neither be analyzed nor designed using the theory and the heuristics developed within the traditional centralized framework. A prime example of this revolution is that of dense distributed sensor networks measuring the physical world. The transition to this new era is taking place by virtue of the convergence of three key technologies: integration of digital circuitry, wireless communications, and micro/nano sensor technology. This convergence is crystallizing in the emergence of devices that communicate wirelessly, store data, process information locally, and can actuate over the environment. All these capabilities are integrated in a unit that is compact, cheap, autonomous, and destined to become ubiquitous. Wireless sensor networks, based on self-organized communications and information systems technology, are on the verge of a technological breakthrough.

There are several hurdles in the way of the above vision. High power efficiency is crucial to ensure autonomous operability in battery-powered wireless terminals. Although the circuit density on chips has consistently doubled every 18 months (as per Moore´s law), the energy density of batteries only seems to double every 10 years. This strongly motivates the need for processing and communication schemes that use energy efficiently when transferring information through a wireless medium. Scalability and self-organization are also crucial. As wireless services become pervasive, users are bound to employ a substantial number of wireless terminals and devices. People and all their things will communicate. Dwarfing the evolution that has taken the number of transmitters from one per thousands of people in the broadcast age to one per person in this mobile phone age, a new transformation will take the number of transmitters to hundreds of tiny autonomous devices per individual in the upcoming networking age. As a result, networking technology will need to become a commodity that everyone can afford, use and deploy. Traditional architectures, usually characterized by rigid and heavy infrastructures, are to be substituted by architectures that are either totally or mostly decentralized and ad-hoc, and where the various tasks of signal processing and communications are carried out in a robust, adaptive, efficient, self-organized and highly scalable manner.

As communication and sensor device technologies continue to march relentlessly with regard to shrinking sizes and increasing computing capabilities, both the fundamental theory and the algorithmic design that are critical to fully harvest these technologies are lagging behind. While, for instance, there is an exciting buzz around wireless ad-hoc networks today, much of it revolves around pushing their operational envelopes for specific applications from the ground up. A fundamental understanding of the capabilities and limitations of these complex systems of networked embedded communication and sensing devices remains far from mature. It is necessary to create new and multidisciplinary theories, algorithms, and designs, that leverage a world where the boundaries between distinct fields such as physics, signal processing, information theory, networking, communications, game theory, biology, computation, control, and machine learning, are being removed and research has to span disciplines traditionally studied in separation. A crucial objective of this proposal is to bring together different such fields in a novel way. Hand in hand with that objective, the grand daunting challenge of pushing the fundamental frontiers for wireless sensor systems is what motivates this proposal.

The effort to address most of these key issues has witnessed, thus far, very weak collaboration among different, yet necessary and complementary, tools and disciplines. This has precluded the emergence of a theoretical framework that captures the relevant problems and provides satisfying solutions. An example of this situation is found in the fields of networking and information Theory. Both areas have long promised interesting synergies. Yet, for many years, the most fruitful exchange between them has come from the hand of researchers who work in both areas or who have migrated from one to the other. The absence of interplay between these fields has reduced very importantly the amount of cross-fertilization. On the one hand, information theory has provided key basic models and has raised fundamental questions. On the other hand, the reality of networking has addressed increasingly complex networks. In their early days, both fields tackled and solved several closely related network problems contemporarily and from different angles. But then, as progress on multi-terminal information theory slowed down and networking tackled ever more complex problems, the gap between these disciplines has widened rather than narrowed. Nevertheless, networking and information theory have always shown a strong conceptual connection, and exciting recent theoretical and algorithmic developments point towards a new world of rewarding cooperation. On the networking side, the complexity of physical layer issues, particularly in wireless networks, has motivated a cross-layer approach that fits information theory well. On the information-theoretic side, classical approaches to multiuser information theory have been enhanced by an active interest in casting practical networking problems in an information-theoretic setting. In particular, developments in information theory have drastically changed the angle of attack on several networking problems.

We have accordingly assembled a first-rate research team (the COMONSENS consortium) with interdisciplinary expertise and where the interaction of different research groups will be strongly motivated. This multidisciplinary research will allow finding novel formulations to new and existing problems as well as verifying certain theories and even posing original problems in other disciplines, thereby generating a great amount of new knowledge. Our research will enable fundamental insights of the performance bounds and limitations of large-scale networks. Today´s embryonic ad-hoc and sensor networks, built from the ground up, are hamstrung by scaling and robustness problems: our study will put forth concrete design guidelines and algorithmic prescriptions. In short, we will provide the expertise on theory and algorithms that is needed to render large-scale robust ad-hoc and sensor networks a reality. We will leverage the various experimental testbeds already developed within the COMONSENS consortium to both validate our assumptions and operating constraints, as well as to effect timely technology transfer of the theory and algorithms coming out of our study. Moreover, we will conduct real deployments that demonstrate the practical impact in several important areas of societal interest.

It should be mentioned that, although there is an important number of groups that are currently conducting research in various areas related to communications, networks, signal processing, and data fusion, which are connected with some aspects of this proposal, only recently, and specially in the U.S., has there been an important effort to migrate from more traditional techniques to a more decentralized communication paradigm. This has been reflected in the recent funding approval by several public and private sponsors, including DARPA, U.S. National Science Foundation, Australian Research Council, INTEL, IBM, and Microsoft, of ambitious projects, as well as by the appearance of several start-up companies. In Europe, although there is clear awareness of the need to advance in this emerging field, there is a significant delay. Only a few countries have recently become active in this area, namely Switzerland (e.g., Swiss Federal institute of Technology, EPFL, ETH), Germany (e.g., Technical University of Munich, Fraunhofer), Sweden (e.g., Swedish Defense Research Agency, FOI), Finland (e.g., Technical Research Center of Finland, VTT), and the Netherlands (e.g., Delf University of Technology). Moreover, in most cases, the amount of multidisciplinary research has been either nonexistent or very limited.

In Spain, and to the best of our knowledge, this strategic area of research is at the moment tackled by few groups that are working separately and consequently lack a sufficient critical mass. Based on this rationale, the COMONSENS consortium has gathered groups of researchers with a strong background in complementary disciplines of interest to the proposal. The added value of the COMONSENS consortium is the network of researchers spanning a wide variety of topics that range from mathematics to applications. This allows a level of versatility that is unusual in academic research, and long-term vision and risk taking that are uncommon in industrial research. We therefore believe that this project provides a unique opportunity to both put this strategic research field center stage within the Spanish scientific community and to achieve clear international impact.

Our proposal focuses on three main interconnected goals:

 Theory: Derivation of novel theoretical tools capable of combining disciplines such as networking, information theory, signal processing, and other diverse fields, with the aim of shedding light on the understanding of the capabilities and limitations of complex communication and processing systems, and specifically on multipoint-to-multipoint communication scenarios. These tools are then to provide theoretical guidelines for the subsequent design of optimum algorithms, in the same way in which many derivations of the fundamental limits in information theory furnish intuition towards designing the corresponding practical coding algorithms. A crucial challenge will be to merge tools from other disciplines, such as random graph and random matrix theory, statistical physics, percolation theory, developmental biology, evolutionary algorithms, game theory, general theory of distributed systems, machine learning, Monte Carlo statistical methods, and advanced optimization theory.

 Algorithms: Design of efficient signal processing algorithms and coding techniques for multipoint-to-multipoint communication and sensor networks based on the previously obtained theoretical guidelines. On the one hand, special emphasis will be placed on distributed algorithms for multipoint-to-multipoint communication, optimizing (separately or jointly) the various processes involved, such as channel estimation, synchronization, coding for multi-input multi-output (MIMO) antenna systems, cognitive sensing and radio transmission, routing, distributed source coding, etc. On the other hand, cooperative algorithms will be designed, subject to communication constraints, for different tasks of interest (event detection, statistical inference, location, estimation, distributed storage, control, etc).

 Testbeds & Applications: Testing and validation of algorithms in testbeds and applications. This project will make extensive use of two testbeds already available within the COMONSENS consortium: a testbed for MIMO communications with several transmit and receive antennas, and a testbed for wireless ad-hoc sensor networking. This will allow, beyond the simulation of algorithms, the comparison of techniques in a realistic setup, the evaluation of their robustness and capabilities, and the provision of valuable feedback into the formulation of further theory. Our testbeds will both keep our research grounded and facilitate the timely impact of the underlying theory and algorithms. Furthermore, during this project, the operational functionalities of these testbeds will be drastically increased. Also, we intend to build demonstrators for important applications such as ambient intelligence and environmental monitoring. These will provide scenarios for the deployment of results obtained throughout the project.