Communication-constrained distributed radar detection in spiky clutter

We analyze a decentralized radar detection system, operating under communication constraints. The interfering background is assumed to be Weibull distributed. Each sensor transmits its local detection to a fusion center, which takes the final decision. In most studies of distributed detection systems, optimal local receivers operating points and decision fusion rule are determined according to the Neyman-Pearson criterion, for maximum global detection performance. However, unexpected large bandwidths of the data links between peripheral sensors and the fusion center can result from this approach, since the optimal solution can imply operating with high peripheral false alarm probabilities. We present an algorithm for the design of the decentralized detection system under communication constraints. We also perform an analysis of the detection performance as a function of channel bandwidth, for various detection environments. Results show that bandwidth limitations affect significantly the performance of decentralized radar detection systems when the clutter is spiky. In such situations one should employ broadband data links; our results can be useful for a trade-off analysis between communication cost and detection performance