Joint channel and impulsive noise estimation for underwater acoustic OFDM systems based on sparse Вayesian learning

Impulsive noise (IN) often occurs in underwater acoustic channels and severely degrades the performance of underwater acoustic orthogonal frequency division multiplexing (OFDM) systems. To address this issue, a joint channel and IN estimation (JCIE) algorithm based on generalized approximate message passing-based sparse Bayesian learning (GAMP-SBL) using all subcarriers and its improved version are proposed. Firstly, utilizing the joint sparsity of the channel and IN, and regarding the data symbols as unknown parameters, a compressed sensing (CS) model is conceived using all subcarriers. Subsequently, the GAMP-SBL algorithm is applied to recover the joint sparse vector, and the symbol estimation is incorporated into the GAMP-SBL framework to jointly estimate the channel impulse response, IN and data symbols. Moreover, exploiting the prior information of the channel and IN attained during the data symbol initialization, redundant atoms are pruned from the dictionary matrix of the all subcarrier-based CS model to further decrease the computational complexity. Simulation results demonstrate that the proposed algorithms can effectively enhance the performance in terms of the channel estimation, IN estimation, and system bit error rate in comparison with the existing GAMP-SBL-based channel and IN estimation algorithms. And compared with the corresponding SBL-based counterparts, the proposed methods can maintain comparable system performance with lower complexity.