Diffusion Least Mean Square Algorithms with Zero-Attracting Adaptive Combiners

Distributed algorithms have been proposed to address the problem of detection and estimation over wireless sensor networks. For diffusion algorithms, the combination weights play an important role. The estimates collected from some neighbors are less reliable than the others due to low signal-to-noise ratio (SNR) conditions. Thus, smaller weights can be given to the noisy estimates, with larger weights placed on those in higher SNR conditions. To improve the robustness to such cases and reduce the communication and computational complexity of sensor networks, a zero-attracting adaptive combination strategy is derived in this work.