Fundamental Concepts for Information Gathering in Sensor Networks

Johns Hopkins University, Baltimore MD

Investigators

Abstract

Wireless networks of sensors will be widely deployed in the future because they greatly extend our ability to monitor and control the physical environment from remote locations. One of the most important communication primitives that has to be provided by sensor networks is information gathering: information from the sensors has to be forwarded to a single point for evaluation. Since sensors will exist with the ratio of thousands of nodes per user, it is impossible to pay attention to any individual node. Hence, it is important to use communication methods in sensor networks that are extremely simple - to allow a verification of their correct implementation - and self-reconfiguring - to allow automatic adjustments to changes in the sensor network or sensing tasks. Furthermore, since sensors are simple devices with limited processing power, storage capacity, and energy, it is important to develop routing strategies that can support information gathering as efficiently as possible. Whereas many heuristics have already been suggested for resource-efficient information gathering in sensor networks, almost no rigorous theoretical work has been performed in this area yet. Therefore, the goal of this project is to explore fundamental concepts for self-reconfigurable and resource-efficient information gathering in wireless sensor networks. This project concentrates on networks of static sensors (i.e. sensors do not move around, which is commonly the case). However, the observer may move around. Also, connections between sensors may be available or unavailable depending on the underlying medium access control protocol or temporary problems such as obstacles. In addition, sensors may fail permanently due to energy problems. To ensure that the network protocols developed in the project will work under any circumstances, models are used in which the injections of packets, the status of the edges, and the movements of the observer are assumed to be under adversarial control. The challenge is to find distributed online protocols for this case - i.e. protocols performing online decisions that can be executed locally at every sensor - that have a near-optimal throughput using a near-minimum amount of energy and storage. Simple protocols with these properties have recently been developed and analyzed by the PI for specific cases (such as specific topologies of sensors and a static observer), and the research goal of the project is to establish analytical and experimental results for much more general cases. A broad impact of the project is ensured by disseminating the research results via lecture notes, conference and journal publications, presentations, and publications on the Web. Furthermore, sensor networks is a highly interdisciplinary field. It involves people from many different areas: integrated circuits, signal processing, operating systems, and theoretical computer science. The project will stimulate interactions of the PI with people from these fields, which may lead to exciting interdisciplinary projects in the future. Finally, since no standards have been set yet for communication in sensor networks, a potential consequence of the project with high social benefit will be to provide valuable contributions to future standards.