Neural network estimation for MIMO-OFDM receivers

Orthogonal frequency division multiplexing (OFDM) has high data rate capacity and lower Inter Symbol Interference (ISI) and is considered as the best solution for next generation mobile communication. Multiple Inputs and Multiple Output (MIMO) antenna system improve reception through spatial diversity and high end coding. Combining these two, offers high interference mitigation in wireless receivers. In this paper, the effects of neural network aided estimation in such receivers are considered. Neural network acts as a pre-processing block to the estimator. The effects of training parameters of the neural network in bit error rate are presented. The training volumes and number of hidden neurons are varied and tested for wide simulation environments.