Knowledge-Aided Target Detection for Multistatic Passive Radar

This paper studies the detection problem for multistatic passive radar. We consider scenarios where prior knowledge about the spectrum and peak-to-average ratio (PAR) of the non-cooperative illuminators of opportunity (IOs) is available. We develop several knowledge-aided (KA) detectors within the framework of the generalized likelihood ratio test (GLRT) to exploit such prior knowledge. Particularly, the knowledge about the bandwidth of the transmitted signal is employed to suppress the out-of-band noise, and the knowledge about the PAR constraint is exploited to eliminate the remaining high-power noise. The challenge of unknown spectrum condition is also addressed, where block sparse Bayesian learning (BSBL) is exploited to derive the maximum-likelihood estimates (MLEs) of the unknown, temporally correlated signal. The numerical results indicate that the proposed KA detectors offer significant performance improvements compared with the traditional detectors, which do not exploit such prior information.