Linear Correction of Mismatched L-values in BICM receivers

TL;DR

This paper proposes a method for linearly correcting L-values in BICM receivers by using saddlepoint approximations of the CGF, leading to improved decoding performance in interference scenarios.

Contribution

It introduces a novel correction factor based on the saddlepoint of the CGF to optimize L-values in BICM receivers, enhancing error probability performance.

Findings

Optimal correction factors are twice the saddlepoint of the CGF.

Significant performance improvement demonstrated in interference scenarios.

Method outperforms mutual information-based correction approaches.

Abstract

In this work we analyze the problem of linear correction of the reliability metrics (L-values) in BICM receivers. We want to find the correction factors that minimize the probability of error of a maximum likelihood decoder that uses the corrected L-values. To this end, we use the efficient approximation of the pairwise error probability in the domain of the cumulant generating functions (CGF) of the L-values and conclude that the optimal correction factors are equal to the twice of the saddlepoint of the CGF. We provide a simple numerical example of transmission in the presence of interference where we demonstrate a notable improvement attainable with the proposed method. The proposed method is compared with the one based on the maximization of generalized mutual information.