Complexity based design for iterative joint equalization and decoding

Abstract

We motivate the need for a complexity based design for performing joint iterative equalization and decoding. This joint iterative process, which requires the exchange of soft information, incurs a huge complexity increase over hard-decision based algorithms. We introduce complexity as a design parameter and provide two different methodologies. The first approach is a combination of SOVA (soft output Viterbi algorithm) and DFSE (decision feedback sequence estimation), and is called soft-output DFSE (SO-DFSE). The second approach, called soft-decision DFSE (SD-DFSE) generalizes the notion of reliability to soft-decisions through the use of appropriately chosen functions. By varying the design parameters in both approaches, the module can range from being as simple as a soft output DFE to being as complex as a SOVA or APP (a posteriori probability). We conclude by presenting performance curves of iterative algorithms that utilize these modules.