Deterministic and Monte Carlo approaches for joint iterative data detection and channel estimation

This work deals with joint data detection and channel estimation for single input single output systems in presence of inter symbol interference. Therefore, deterministic methods, the Gibbs-sampler and combinations between deterministic and Monte Carlo approaches are compared. The examined methods belong to the class of block by block iterative algorithms alternating between channel estimation and data detection. It will be shown that the deterministic method might get trapped in a local maximum of the likelihood function, whereas the Monte Carlo methods theoretically almost converge to a global maximum. Based on simulation results it will be shown that a performance gain can be achieved at the expense of slower convergence speed or an increased computational effort.