Reduced-complexity MAP equalizer for dispersive channels

Abstract

We present a computationally efficient soft-output equalizer for communication over dispersive channels. The proposed equalizer has a forward and backward recursion structure similar to the classic BCJR MAP algorithm, but its complexity is reduced by constructing a reduced-state trellis, as in the decision feedback sequence estimator (DFSE). We are able to achieve consistent state definitions by running the backward pass on the same trellis as the forward pass. The performance of this reduced-complexity equalizer is evaluated within an iterative receiver in which the intersymbol interference (ISI) channel and channel code are treated as serially concatenated codes. The EDGE air interface is used as an illustrative example, for which we are able to reduce the number of states from 4096 to 8.