MIMO radar detection with heterogeneous propagation losses

A multiple-input multiple-output (MIMO) radar can utilize distributed antennas to improve detection performance. Some existing works model bistatic path gains as identically and independently distributed random variables, and the known distinct features between propagation paths are neglected. Based on the bistatic radar equation, a path gain is a product of target reflection gain and propagation loss. Since the latter depends on geometry and system parameters, the true path gains are not necessarily identically distributed. In this paper, the propagation loss is incorporated into the MIMO radar signal model and the binary hypothesis test is explored. The test statistic is a sum of weighted Chi-square random variables under either hypothesis, and its distribution is a Gamma mixture. The probabilities of detection and false alarm can be explicitly obtained. In addition, a saddlepoint approximation based performance analysis is suggested, and it may be less sensitive to numerical errors, particularly when weighting coefficients are similar.