EM Based Channel Estimation and Decoding in OFDM Turbo Blast Detectors

Channel parameter estimation is an important task for coherent detection and constitutes a key challenge in MIMO systems. Turbo-Bell Labs Laboratory Architecture Systems (T-BLAST) with orthogonal frequency division multiplexing (OFDM) can be used to improve communication quality and capacity. This paper proposed a modified expectation maximization (EM) algorithm for T-BLAST systems in frequency selective channels. It is based on a combination of data and pilot based channel estimate with approximated coefficients. These coefficients are obtained by resolving an optimization problem using the Lagrangian multipliers. The algorithm we present takes advantage of the exchange message between the two soft input soft output (SISO) modules of the turbo receiver to attain the best estimation performance. The effectiveness of this technique is demonstrated through the simulation of an OFDM T-BLAST system with two-transmit and three-receive antennas.