Skip to Main Content
M-BCJR algorithm is a reduced state version of the BCJR algorithm and selects a set of active states in the forward recursion based on an estimation of the filtered probability distribution of states at each time. We propose to use instead an estimation of the fixed-lag smoothed probability distribution of states with a non zero lag. Our implementation uses a Gaussian approximation to estimate these distributions with a low complexity, using the principle of probabilistic data association (PDA). The performance of the M-BCJR can be seen to be greatly improved as a result while remaining robust against changes in the channel multipath profile.