BCJR algorithm

The BCJR algorithm, a name derived from the initials of its designers L. Bahl, J. Cocke, F. Jelinek and J. Raviv from, was developed in 1974 for decoding of block and convolutional codes. It is the optimal in the sense of minimum symbol error probability decoding algorithm (maximum a posteriori probability, MAP) and is used in particular in the iterative decoding of parallel or serial concatenated convolutional and block codes, such as turbo codes.

The advantage of the BCJR algorithm for decoding convolutional codes by means of so-called soft-decision is the efficient use of the information of the composite probability of successive code symbols, referred to as Markov chains. It may be as well as the Viterbi algorithm in the form of a trellis diagram graphically.