Distributed MIMO Radar Based on Sparse Sensing: Analysis and Efficient Implementation

In sparse sensing based distributed multiple-input multiple-output radars, the problem of target estimation is formulated as a sparse vector recovery problem, where the vector to be recovered is block sparse, or equivalently, the sensing matrix is block diagonal and the sparse vector consists of equal-length blocks that have the same sparsity profile. This paper derives the theoretical requirements and performance guarantees for the application of sparse recovery techniques to this problem. The obtained theoretical results confirm previous, simulations-based observations that exploiting the block sparsity of the target vector can further reduce the amount of measurements needed for successful target estimation. For signal recovery, two low-complexity approaches are proposed. The first one is an alternating direction method of multipliers-based sparse signal recovery algorithm, which in addition to significantly reducing computations is also amenable to a parallel and semidistributed implementation. The second approach decouples the location and speed estimation into two separate stages, with each stage addressing a sparse recovery problem of lower dimension while maintaining high estimation accuracy.