Iterative Turbo Decoding Using Gibbs Sampling

This paper discusses an iterative multiuser receiver for codedivision multiple access (CDMA) with forward error control coding. The receiver is derived from the maximum aposteriori (MAP) criterion for the joint received signal. A major drawback of the MAP receiver is its heavy computational cost that grows exponentially with the number of users. An alternative solution is proposed here based on a MonteCarlo approach where the MAP probabilities are computed via the Gibbs sampler. The computational cost of the MonteCarlo based method grows linearly with the number of users times the number of Monte-Carlo runs. In a second part of the paper, we present a ‘turbo’ equalization method that combines the previous decoding technique with an EM (Expectation Maximization) estimation technique for the blind or semi-blind identification of the channel parameters. Simulation results are finally presented to illustrate the performances of the proposed method.