A sequential decoder for linear block codes with a variable bias-term metric

A sequential decoder for linear block codes that performs maximum-likelihood soft-decision decoding is described. The decoder uses a metric computed from a lower bound on the cost of the unexplored portion of the code tree. It is shown that for certain block codes the average computational complexity of this metric is superior to that of the Fano metric. A new function, the cumulative column distance function, is introduced for linear block codes. This function is an important factor that determines the average computational effort of a sequential decoder for a linear block code with an arbitrary maximum-likelihood metric. Simulation results show that a sequential decoder for linear block codes with a fast growing cumulative column distance function achieves a low computational complexity, a result analogous to that for convolutional codes.