CAREER: Local Information Based Distributed Optimization of Resources in Large-Scale Adhoc and Sensor Networks

Rensselaer Polytechnic Institute, Troy NY

Investigators

Abstract

Efficient usage of bandwidth and/or energy is necessary for the success of a majority of ad-hoc and sensor network applications. This project focuses on the development of distributed bandwidth/energy optimization algorithms that require only local network topology or state information, and yet guarantee a performance close to the global or local optimum. More specifically, this project involves the development of distributed node wakeup, medium access control, routing and flow control algorithms that optimize (exactly or approximately) the throughput or lifetime of the network, while posing minimal overhead of communication on the individual nodes. The project objectives also include understanding the fundamental trade-offs between optimization accuracy and communication complexity, so that the appropriate level of accuracy can be chosen based on the network and application characteristics. The questions considered in this project are crucially important for the efficient management of ad-hoc and sensor networks in a wide range of application scenarios, including environmental data gathering, military and relief operations, and health monitoring. Moreover, although the project is focused only on large-scale ad-hoc and sensor networks, the results of this research are expected to influence a larger body of science and engineering, including optimization theory, stochastic control and distributed algorithms. The expected outcomes of this project also include a publicly available simulation tool, a junior museum exhibit, and a graduate level course on distributed optimization methods in networking. A conscious effort will be made towards engaging women and other under-represented groups in the research and educational programs related to this project.