Performance/Complexity Comparison between MAP-PSP and Mixture Kalman Filtering for Joint Estimation and Detection of STTCs

This paper compares the performance and computational complexity of two joint channel estimation and data detection algorithms for space-time trellis codes (STTCs) over time-varying flat fading channels. The first algorithm, the maximum a posteriori probability-per survivor processing (MAP-PSP), employs an improved survivor branch decision technique based on the symbol by symbol MAP criterion, with a fixed delay. This delay allows future received symbols to be utilized in the decision making, resulting in a more reliable survivor branch selection than in the conventional PSP. The second one is based on a delayed mixture Kalman filtering (MKF) technique, where importance samples and weights take into account also future received symbols. Simulation results show that the MAP-PSP algorithm substantially outperforms the delayed MKF algorithm with a lower computational complexity.