Joint semiblind channel estimation and data detection for MIMO-OFDM systems

In MIMO systems, where multiple antennas are used at both transmitter and receiver to achieve high spectral efficiency, channel impulse responses are often assumed to be constant over a block or packet. This assumption of block stationarity on channels is valid for most fixed wireless scenarios. However, for communications in a high mobility environment, the assumption will result in considerable performance degradation. In this paper, we focus on channel estimation for a MIMO system with OFDM transmission technique. In our system, pilots are placed on N_P subcarriers for a novel channel estimation at the receiver with Kalman filters. With the channels estimated by a Kalman filter, we apply the QRD-M algorithm for suboptimal data detection with reasonable computational cost. The outputs of data detection are fed back to another Kalman filter for an improved channel estimation. By alternatively and iteratively using these two Kalman filters, a better channel estimation can be obtained. A simulation result of the bit error rate shows the improvement of 10 dB in SNR with four iterations over noniteration.