A BBL-Net based OFDM Signal Detection in the Presence of RF Impairments

Channel estimation is an essential technology for any multi-carrier modulation (MCM) framework. Traditionally, least square (LS) and minimum mean square error (MMSE) estimators are used based on the availability of channel state information (CSI). However, in the presence of radio frequency (RF) impairments, i.e., carrier frequency offset (CFO) and phase offset(PO), the statistical models lose their optimality. This paper proposes a batch normalized Bi-directional long short memory network (BBL-Net) based joint channel estimation and signal detection approach for orthogonal frequency division multiplexing (OFDM) system in the presence of CFO and PO at the receiver. A combination of experiments has been conducted to find the optimum activation function for the BBL-Net architecture. Simulation results show that the trained model can recall and predict the information efficiently even in high interference scenarios and also provide robust performance in lower pilot numbers.