Pareto Boundary of the Rate Region for Single-Stream MIMO Interference Channels: Linear Transceiver Design

TL;DR

This paper addresses the challenging problem of characterizing the Pareto boundary of the rate region in single-stream MIMO interference channels, proposing an algorithm for exact and approximate solutions with convergence guarantees.

Contribution

It introduces an alternating optimization algorithm to compute the Pareto boundary for two-user and multi-user MIMO interference channels, including exact and approximate solutions.

Findings

The algorithm computes well-distributed operating points on the Pareto boundary.

Numerical results show the algorithm's effectiveness and convergence.

The method provides a complete inner bound of the strict Pareto boundary.

Abstract

We consider a multiple-input multiple-output (MIMO) interference channel (IC), where a single data stream per user is transmitted and each receiver treats interference as noise. The paper focuses on the open problem of computing the outermost boundary (so-called Pareto boundary-PB) of the achievable rate region under linear transceiver design. The Pareto boundary consists of the strict PB and non-strict PB. For the two user case, we compute the non-strict PB and the two ending points of the strict PB exactly. For the strict PB, we formulate the problem to maximize one rate while the other rate is fixed such that a strict PB point is reached. To solve this non-convex optimization problem which results from the hard-coupled two transmit beamformers, we propose an alternating optimization algorithm. Furthermore, we extend the algorithm to the multi-user scenario and show convergence. Numerical simulations illustrate that the proposed algorithm computes a sequence of well-distributed operating points that serve as a reasonable and complete inner bound of the strict PB compared with existing methods.