A Cooperative Game Theoretic Framework for Collaborative Information Processing in Distributed Sensor Nets

University Of New Mexico, Albuquerque NM

Investigators

Abstract

CCF- 0830545: A Cooperative Game Theoretic Framework for Collaborative Information Processing in Distributed Sensor Nets Distributed sensor networks can be used to monitor or track moving targets, environmental processes, and leaked chemical or biological agents. When distributed sensor nets are used to monitor such phenomenon of interest, each node in the network can come up with an estimate based on its own observation of the process. However, nodes may be able improve their estimates if they were to share information with each other, by forming coalitions. The price for such inter-node collaboration can be the cost of communication resources. Unless only the nodes that are the best candidates join a coalition, there is the drawback of waste of individual node power as well as causing unwanted excessive interference to everyone else. These concerns become critical when we consider, as in this research, sensor nets formed by battery-powered nodes operating in a limited bandwidth. But, a.) How do we define best nodes to join a coalition at any given time in a dynamic sensor net? b.) What should be a fair resource commitment from each node to a given coalition? In this research we formulate the above problem of collaborative information processing in distributed sensor networks as a cooperative game theoretic problem. The research covers different types of sensor network architectures as well as various fairness criteria among nodes in a network. By casting in a cooperative game theoretic framework allows us to apply solution concepts from game theory to achieve different fairness criteria among nodes, in developing collaborative processing schemes. The research is carried out in conjunction with an outreach program that focuses on providing international educational experience to students.