Distorted sparse signal estimation from distributed sign measurements

Abstract

A novel algorithm called Cooperative Binary Iterative Hard Thresholding (CB-IHT) based on distributed 1-bit compressive sensing is proposed in this paper. Taking advantage of the correlated nature of distributed signal processing, the proposed algorithm is aimed to fight against the error floor in the estimation of distorted sparse signal, with an array of agents recovering the target signal cooperatively. The principles of convex optimization, consistent reconstruction and greedy pursuit algorithm are combined in the algorithm design. With two joint sparsity models representing distortion of equivalent parallel AWGN channels and parallel fading channels separately, the algorithm is performed through extensive simulations, which show that with severe distortion and large bit-budget, estimation accuracy can be improved by simply increasing the array scale.