Optimal DoF Region for the Asymmetric Two-Pair MIMO Two-Way Relay Channel

TL;DR

This paper characterizes the optimal degrees of freedom region for an asymmetric two-pair MIMO two-way relay channel, proposing a new transmission scheme that accounts for asymmetric data streams and antenna configurations.

Contribution

It derives the DoF region outer bound, proposes a scheme to achieve it, and analyzes the impact of antenna asymmetry on DoF and data exchange symmetry.

Findings

Optimal DoF region derived for asymmetric MIMO TWRC.

Achievability scheme with joint relay and source processing.

Redundant antennas identified in asymmetric configurations.

Abstract

In this paper, we study the optimal degrees of freedom (DoF) region for the two-pair MIMO two-way relay channel (TWRC) with asymmetric antenna setting, where two pairs of users exchange information with the help of a common relay. Each user $i$ is equipped with $M_i$ antennas, for $i=1,2,3,4$, and the relay is equipped with $N$ antennas. First, we derive an outer bound of the DoF region by using the cut-set theorem and the genie-message approach. Then, we propose a new transmission scheme to achieve the outer bound of the DoF region. Due to the asymmetric data exchange, where the two users in each pair can communicate a different number of data streams, we not only need to form the network-coded symbols but also need to process the additional asymmetric data streams at the relay. This is realized through the joint design of relay compression matrix and source precoding matrices. After obtaining the optimal DoF region, we study the optimal sum DoF by solving a linear programming problem. From the optimal DoF region of this channel, we show that in the asymmetric antenna setting, some antennas at certain source nodes are redundant and cannot contribute to enlarge the DoF region. We also show that there is no loss of optimality in terms of the sum DoF by enforcing symmetric data exchange, where the two users in each pair are restricted to communicate the same number of data streams.