Detection Capacities of Distributed and Centralized Systems: A Comparative Study

Distributed systems using many inexpensive sensors widely distributed over a large area present an alternative way for target detection and potential paradigm change in environmental sensing. The diversity of opportunities for detection by widely distributed sensors seems attractive, but how one compares the detection performance based on the observations made from many distributed sensors, each with small gain, with that from a centralized system with a high array gain has not been studied theoretically or experimentally. Treating the target-radiated signal as a communication signal, transmitting continuous Gaussian-distributed alphabets, the Shannon channel capacity yields the maximum information that the receivers can learn about the (target) transmitted signal. For this idea case, the channel capacity can then be used as a metric to compare the performance of various sensor systems. Matched track processing is introduced to motivate a capacity-based detector and the corresponding detection capacity. Based on that, it is found that the distributed systems can achieve, in principle, an area of coverage two to three times larger than that of a centralized system under the right conditions, and the area of coverage by the entire system can be significantly larger than the sum of detection areas of individual nodes for distributed systems.