The NCCR MICS research agenda

Instrumenting a landscape, using wireless sensors to monitor a watershed from glaciers to river mouths, allows us to better understand the mechanisms governing the circulation of water and thus to improve prediction and management of this valuable resource. Incorporating wireless sensors and actuators in the walls of a building has the potential to reduce its construction and maintenance costs (e.g. its energy consumption) while increasing the comfort of the users by creating individual microclimates.
These are just two examples of how wireless sensor networks will change the way we use information technology: information becomes embedded into our physical environment by means of miniature devices and computers, providing dense sensing close to physical phenomena. This information is processed and transmitted, and actions can be coordinated. The physical environment thus becomes intertwined with the Internet information space, evolving into what we envision as the Smart Earth.

Embedded instruments and computers have existed for a long time, but their reduction in size and cost, as well as the spread of wireless communication facilities, has made it possible for these technologies to enjoy a much wider range of use. Furthermore, sensor technology is expected to have a significant impact on personal communication and society: on the Web, people will be able to share not only documents, but also their multisensory experience of events.

Many technological challenges must be addressed to realize this potential. The devices are resource-constrained by their size and by available energy sources. They have limited processing speeds, storage capacities and communication bandwidths. They must be inexpensive to produce, deploy and maintain. Finally, their use raises tricky security and privacy issues. As applications quickly involve very large numbers of sensors, scalability is a major issue as well, leading to the use of self-organization principles.

Communication is one of the most energy-consuming tasks. For information to be forwarded over long distances, sensor networks route it from node to node, from the point of production to the point of use, and process data locally as much as possible. Various, sometimes conflicting optimization techniques are being investigated in order to minimize the use of energy, depending on the applications and platforms.

The NCCR MICS is tackling this wide range of problems, from the study of fundamental principles (network structures, distributed algorithms, information and communication theory) to the development of platforms (wireless sensor technology, ad-hoc networks, in-network information processing, software verification), and their deployment in applications.

Theory of self-organized, distributed communication and information

The projects dealing with this topic address theoretical questions in four areas: information theory, network theory, distributed signal processing and distributed algorithms.

Information theory considers the ultimate limits of communication systems and points out the basic tradeoffs such as those between data rate, reliability, bandwidth and energy consumption. This aspect is one of the mathematical foundations of communication systems design.

Network theory looks at the performance and the behavior of large-scale wireless networks, considering the feasibility of such networks under various constraints, as well as algorithmic questions.

The third area of research relates to fundamental signal processing principles. Scientists want to derive precisely the structure of distributed signals for some important classes of signals (acoustics, for example) and, in turn, derive bounds on distributed compression and communication. They also want to investigate the interaction of sources and channels.

It is desirable that algorithms for wireless and mobile networks require as little communications as possible and run as fast as possible. Both goals can only be achieved by developing local algorithms, requiring a small number of communication rounds only.

Related projects:	Information and coding theory for wireless multi-hop networks
	Network theory for wireless multi-hop networks
	Distributed signal processing and communication in sensor networks
	Algorithmic foundations of ad hoc and sensor networks
	SensorScope and its application to environmental monitoring

Mobile communication and processing platforms

This topic integrates all the expertise that is required to cope with the challenges linked to the implementation and deployment of wireless ad-hoc networks.

One of the projects aims at gaining insight into the dynamics of rapid gravity-driven flows, such as avalanches and earth mass movements, by using a sensor network-based monitoring system. The data will make it possible to build fluid-mechanics models describing the flow behavior in a realistic way.

A second project is concerned with the design, modeling, and control of a distributed, self-organized, networked olfactory system for chemical plume mapping and odor source localization. The main difficulty will be to design a system which will integrate the appropriate combination of sensitivity, response speed and size to be embedded in a miniature robotic platform.

Scientists involved in this cluster are also exploring how energy reduction can be achieved by using a more systematic cross-layer optimization, including additional upper and lower layers. In that respect, radio ultra-wide band communication offers interesting opportunities.

Related projects:	Very low radiated power UWB communication
	Deployment of sensor networks
	Modular and composable platform for sensor and actuator networks
	Distributed odor source localization using a miniature multi-robot system
	Real-time avalanche and landslide analysis through sensor networks

Networked software systems

The construction of software systems (protocols, middleware, application development tools) for ad hoc and sensor networks remains difficult. Researchers in this cluster investigate key technologies to allow the efficient and safe construction of such software systems.   The key idea is to develop techniques to check the properties of software modules, using a combination of compile-time (off-line) and run-time (dynamic) analysis.

One thrust is the analysis of multi-threaded (Java )programs to detect errors or to issue warnings.   Current approaches provide either no detection or do not support modularity (where every module can be replaced and new classes can be added).   Another thrust is to prove communication protocols correct and secure, e.g., to confirm that a protocol keeps private the location of the sender or any node that relays a message.

In addition to these two approaches that address fundamental problems that affect software development, the cluster includes two application projects that deploy sensors in difficult-to reach regions: one project focuses on the permafrost region in the Swiss alps (some sensors are reachable only by climbing in adverse terrain), the other project's focus is on water and humidity monitoring and management in an arid part of the Indian subcontinent, many kilometers away from Switzerland.

Related projects:	Checking properties of flexible programs in the presence of modularity
	VerSePro: verification of security and privacy protocols for wireless networks
	Secure stream ciphers
	Spam detection based on self-organization
	PermaSense
	WaterSense

In-network information management

Future information systems will operate on data acquired in real time by a collection of sensor network and RFID antennas. Projects within this cluster aim at supporting end-to-end information management for sensor and mobile networks. They concern all system layers and processing levels.

However, adequate models, tools, and mechanisms for developing applications in such environments are still lacking. In terms of models, the scientists want to come up with a "declarative middleware" language which is supported by a flexible communication infrastructure. Thus sensors become services used by potentially many applications.

For optimizing data flow and data aggregation in sensor networks, the scientists develop as well protocols and a data management middleware for sensor networks. This middleware will allow to disseminating knowledge on information needs by applications and in reaction to dynamically reorganize data flows.

Finally, several projects within this cluster address issues in information processing across different networks, in particular bridging sensor networks with internet.

Related projects:	Xtream
	Mobile publish-subscribe
	Distributed event detection and localization architecture for wireless sensor networks
	Data dissemination in mobile ad hoc sensor environments
	Sensor awareness
	Serious building games
	Idea futures market for MICS technology foresight