Semi-blind maximum a posteriori multipath fast fading channel estimation for TDMA systems

We propose a maximum a posteriori burst-by-burst multipath fast fading channel estimation algorithm for TDMA systems. This algorithm performs a semi-blind iterative channel estimation using the SAGE algorithm. It uses both CAZAC reference sequences and unknown symbols in the optimization of this channel estimation. It requires a convenient representation of the multipath fading channel using a discrete version of the Karhunen-Loeve expansion theorem. It also requires a trellis representation of the channel for an optimum characterization of data symbols by their a posteriori conditional probabilities. The computation of these probabilities can be realized via the Bahl or BCJR algorithm. The performance of our algorithm is evaluated by simulation for 4-PSK modulated TDMA systems.