Soft Input Soft Output MMSE-SQRD Based Turbo Equalization for MIMO-OFDM Systems under Imperfect Channel Estimation

Abstract

In this paper, a turbo equalization scheme for MIMO-OFDM systems under imperfect channel estimation based on soft-input soft-output (SISO) minimum mean-square error (MMSE) sorted QR decomposition (SQRD) is proposed. A turbo structure consists of a SISO detector and a SISO decoder where extrinsic information is exchanged between the two SISO modules. Turbo equalization schemes are preferable in practical communication systems due to their good performance and acceptable computational complexity. MMSE-SQRD based SISO detection derives from SISO MMSE detection, and successive interference cancellation (SIC) is performed using a posteriori information obtained from previous detected symbols. Compared to SISO MMSE detection, MMSE-SQRD based SISO detection is of low complexity but has significant bit error rate (BER) performance enhancement. However, the derivation of the MMSE-SQRD based SISO detection scheme is under perfect knowledge of channel information at receivers. When channel estimation errors are presented, it has been pointed out that the system performance will degrade. In this paper, we studied this practical issue, and proposed the SISO MMSE-SQRD based turbo equalization under imperfect channel estimation. We first model the channel estimation error as added random Gaussian noise over the channel estimation matrix; based on that, we rederive the SISO MMSE detection for the data, and then redefine the extended channel matrix and receive vector by taking into account of channel estimation errors; after that, the SQRD algorithm is adjusted in accordance; MMSE-SQRD based data detection algorithm is finally performed. Numerical simulation results show that the proposed SISO MMSE-SQRD based turbo equalization for MIMO-OFDM systems under imperfect channel estimation outperforms the traditional MMSE based SISO detection with imperfect channel estimation in terms of BER performance and computational complexity.