L0.5-regularization based distributed channel estimation for industrial wireless sensor network

In wireless communication, channel state information (CSI) is essential for receivers to correctly demodulate the signal from transmitters. The performance of wireless sensor network is significantly affected by accuracy of channel estimation, which is obtained by exciting the channel with a probing sequence and decoding the impulse response of receivers. Considering the sparsity of response, compressive sensing theory can be applied to decrease the sampling rates at the receivers, who recover the commonly detected CSI by solving the Lp-regularization. Hence, a distributed multipath channel estimation (DMCE) approach is proposed here, utilizing iterative half thresholding algorithm to solve the L_0.5-regularization. Distributed cooperation reduces iteration times by allocating the computation burden to multiple sensors, whose spatial diversity reduces error of estimation from the noisy measurements. Simulation results shows the efficiency, effectiveness and promotion in sparsity than L₁-regularization and non-cooperative scheme of our approach.