Comparison of Sparse Signal Separation Algorithms for Maritime Radar Target Detection

Due to the non-stationary nature of sea clutter, traditional maritime radar detection schemes utilise non-coherent processing. To further enhance the detection performance, one alternative is to use sparse signal separation. This is an alternative paradigm, whereby the different spatio-temporal characteristics of the radar signal are exploited to separate targets from the background interference. In previous work, the sparse signal separation problem has been posed in a compressive sensing framework so as to improve detection of small maritime targets. This paper investigates the performance of three different algorithms for solving the signal separation problem. These include the Split Augmented Lagrangian Shrinkage Algorithm (SALSA), adaptive Complex Approximate Message Passing (CAMP) and the Fast Sparse Functional Iteration Algorithm (FSFIA). The first contribution is to reformulate the CAMP algorithm to the framework of sparse signal separation. The suitability of each algorithm is then assessed using real data from the Ingara radar, and is based on the quality of the solutions obtained, the computational speed and the robustness to the user's choice of 'tuning' parameters.