A Modified Levenberg-Marquardt Method for the Bidirectional Relay Channel

TL;DR

This paper introduces a modified Levenberg-Marquardt algorithm to optimize fairness in a multi-user bidirectional relay system, effectively mitigating interference and achieving near-optimal solutions with low complexity.

Contribution

It develops a novel iterative method based on a modified Levenberg-Marquardt algorithm for non-convex precoding optimization in relay channels.

Findings

The method achieves solutions close to convex relaxation approaches.

It offers a low-complexity alternative to existing optimization techniques.

The approach improves fairness in multi-user relay systems.

Abstract

This paper presents an optimization approach for a system consisting of multiple bidirectional links over a two-way amplify-and-forward relay. It is desired to improve the fairness of the system. All user pairs exchange information over one relay station with multiple antennas. Due to the joint transmission to all users, the users are subject to mutual interference. A mitigation of the interference can be achieved by max-min fair precoding optimization where the relay is subject to a sum power constraint. The resulting optimization problem is non-convex. This paper proposes a novel iterative and low complexity approach based on a modified Levenberg-Marquardt method to find near optimal solutions. The presented method finds solutions close to the standard convex-solver based relaxation approach.