Machine Learning-based Channel Estimation for EEG Signal Transmission in MIMO-OFDM System

In wireless communication system, the most widely used modulation approach is Orthogonal Frequency Division Multiplexing (OFDM) for time fading and frequency selective channels. The channel estimation (CE) process, one of the most important techniques, makes the OFDM system robust and noise-free. Large amount of training samples is required by Deep Learning algorithms, which are normally computationally demanding. Therefore, the feasibility of using efficient machine learning is investigated in this work for estimating the frequency-selective and time- varying channels. Neural Network (NN) technique is implemented for CE, and it is integrated with MIMO-OFDM system. The proposed system’s BER performance is compared with other systems to decide optimal value for transmitter and receiver antennas in the MIMO system. By varying the transmitter and receiver number, the NN-MIMO-OFDM is validated and compared with existing MIMO-OFDM.