We introduce the notion of diversity order in distributed radar networks. Our goal is to analyze the tradeoff between distributed detection, using $K$ sensors, and centralized detection, using collocated antennas. The diversity order is representative of the degrees of freedom available in the system. In contrast with the asymptotically high signal-to-noise ratio (SNR) definition in wireless communications, we define the diversity order of a distributed radar network as the slope of the probability of detection ($P_{rm{D}}$) versus SNR curve at $P_{rm{D}}=0.5$. We analyze an optimal joint detection system and prove that its corresponding Neyman-Pearson (NP) test statistic follows a Gamma distribution and that, for large $K$, its diversity order grows as $sqrt{K}$. For a fully distributed system using the NP fusion rule, we prove that the test statistic follows a binomial distribution and that the diversity order is also on the order of $sqrt{K}$. In more practical systems where the fusion center uses a fixed fusion rule, the largest growth in diversity order is achieved by the OR rule, and it only grows as $log(K)$. We provide the results of simulations to confirm the theory developed.
展开▼
