Multiple target detection for an antenna array using outlier rejection methods

When there are multiple targets in a Gaussian noise background, single target detection methods can cause missed detections. The missed detections result because targets are present in the data used to estimate the covariance. Range bins containing targets differ from target-free range bins and are considered to be outliers. Different outlier rejection methods for multivariate data are developed. These methods include single outlier detection via hypothesis testing and multiple outlier detection using model order methods. Range bins detected as outliers are removed from the covariance estimate and target detection is performed. Simulation results for the various outlier rejection methods are presented.<>