Optimal detector of OFDM signals for imperfect channel estimation

An optimal pilot assisted detector of OFDM signals in frequency selective fading channels is proposed and investigated. This detector does not estimate the channel explicitly but jointly processes the received data and pilot symbols to recover the data with minimum error. It outperforms the traditional mismatched detectors that treat channel estimates as perfect. We use complex exponential functions to approximate the channel frequency response and compare the performance of a channel estimator applying complex exponential functions and that of an estimator applying the traditional linear interpolation. We also compare the performance of mismatched detectors using maximum likelihood (ML) channel estimates with that of the optimal detector. The performance of the proposed optimal detector is better than that of the mismatched detectors and it is close to that of the detector with perfect channel information. We also investigate the performance of iterative receivers in a system transmitting turbo encoded data. Simulation results show that the iterative receiver applying the optimal detector at the initial iteration and a mismatched detector in the following iterations outperforms the iterative receivers applying mismatched detectors in all iterations.