Distributed Coordination for Signal Detection in Sensor Networks

点击翻译按钮获取中文摘要

Distributed Coordination for Signal Detection in Sensor NetworksHypothesis testing, often called signal detection, is essential for Engineering and Science while being one of the most important applications of signal processing. Though signal detection has been studied for many years, only recently has energy efficient signal detection been consideredfor sensor networking applications. Energy efficiency appears to be the most critical barrier for sensor networking. Sensor networks hold great promise for solving many important problems including improved monitoring, control and repair of the human body, buildings, bridges, energy production facilities, forests and other critical infrastructure, while also providing important contributions to homeland security, law enforcement, disaster prediction/avoidance and defense related problems.The investigators intend to demonstrate it is possible to coordinate the operation of multiple dispersed sensors in a distributed manner, without inter-sensor communication, such that significant energy is saved without any loss in detection performance. The key is to jointly consider the signal detection and the communications. This can be achieved by having the sensors with themost informative observations transmit first, while either the sensors or the fusion center listen to keep track of the total likelihood of the possible hypotheses. This allows the sensor transmissions to be halted when the evidence for one hypothesis becomes overwhelming. The approach generalizes previous order statistic and sequential testing approaches while optimizing performance in a specific way that models a highly distributed mode of operation for the signal processing, communication, and networking. The goal of this project is to fully understand the theory of energy efficient signal detection for sensor networks employing distributed processing. New approacheswill be developed and tested with an emphasis on performance and complexity estimation.