Distributed Coordination for Signal Detection in Sensor Networks

传感器网络中信号检测的分布式协调假设检验，通常称为信号检测，是工程和科学中必不可少的，同时也是信号处理最重要的应用之一。虽然信号检测已经研究了很多年，但直到最近才考虑将能量有效的信号检测用于传感器网络应用。能源效率似乎是传感器网络最关键的障碍。传感器网络为解决许多重要问题提供了巨大的希望，包括改善对人体，建筑物，桥梁，能源生产设施，森林和其他关键基础设施的监测，控制和修复，同时还为国土安全，执法，灾害预测/避免和防御相关的问题。调查人员打算证明，在分布式环境中协调多个分散的传感器的操作是可能的。方式，而没有传感器间通信，从而节省了大量能量，而没有检测性能的任何损失。关键是要把信号检测和通信结合起来考虑。这可以通过让具有最多信息观测的传感器首先发送来实现，而传感器或融合中心则监听以跟踪可能假设的总可能性。这使得传感器传输可以在一个假设的证据变得压倒性时停止。该方法概括了以前的顺序统计量和顺序测试方法，同时以特定的方式优化性能，该方式为信号处理，通信和网络的高度分布式操作模式建模。这个项目的目标是充分理解传感器网络采用分布式处理的节能信号检测理论。新的方法将被开发和测试，重点是性能和复杂性估计。

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.