Adaptive T-algorithm in MLSD/MLSDE receivers for fading channels

This paper develops a threshold method called adaptive T-algorithm to reduce the number of states in the maximum likelihood sequence detection and estimation (MLSD/MLSDE) receivers. This algorithm removes the states whose costs are larger than the minimum cost (the cost of the best survivor path) plus a threshold at each time. The threshold value is formulated based on the probability of removing the correct state in the trellis diagram and it is a function of short-term power of the channel impulse response and the variance of the additive white Gaussian noise. Simulation results for frequency flat and selective fading channels show that the adaptive T-algorithm greatly reduces the computational complexity of the MLSD/MLSDE receivers especially at high SNR with negligible loss in their performances.