Sparse Bayesian learning based channel estimation in FBMC/OQAM industrial IoT networks

Abstract

The next generation of communication technology is accelerating the transformation of industrial internet of things (IIoT). Filter bank multicarrier with offset quadrature amplitude modulation (FBMC/OQAM), as a candidate wireless transmission technology for beyond fifth generation (5G), has been widely concerned by researchers. However, effective channel estimation (CE) in IIoT communication should be solved. In practice, wireless channels have block–sparse structures. For the conventional sparse channel model, the general sparse channel estimation methods do not take the potential block–sparse structure information into account. In this paper, we have investigated the sparse Bayesian learning (SBL) framework for sparse multipath CE in FBMC/OQAM communications. Block SBL (BSBL) algorithm is proposed to estimate the channel performance by exploiting the block–sparse structure of sparse multipath channel model. The BSBL method can improve the estimation performance by using the block correlation of the training matrix. Computer simulation results demonstrate the robustness of the BSBL CE approach in FBMC/OQAM systems, which can achieve lower mean square error (MSE) and bit error rate (BER) than traditional least squares (LS) method and classical compressive sensing methods. The state of art compressive sampling matching pursuit (CoSaMP) greedy algorithm with a prior knowledge of sparse degree can provide slightly better CE performance than BSBL algorithm, but the proposed method maintains robustness in practical channel scenario without the prior knowledge of sparse degree.