On Precoding for Constant K-User MIMO Gaussian Interference Channel with Finite Constellation Inputs

TL;DR

This paper investigates linear precoding strategies for the K-User MIMO Gaussian Interference Channel with finite constellation inputs, deriving conditions for optimality, and proposing algorithms that outperform existing interference alignment methods in sum-rate performance.

Contribution

It introduces a high SNR approximation for finite constellation inputs, characterizes precoders that achieve maximum rates, and develops gradient-based algorithms for sum-rate optimization.

Findings

Precoders satisfying interference alignment conditions achieve CCSC.

Proposed algorithms improve sum-rate by over 0.1 bits/sec/Hz at moderate SNRs.

Finite constellation inputs significantly impact precoding design and performance.

Abstract

This paper considers linear precoding for constant channel-coefficient $K$-User MIMO Gaussian Interference Channel (MIMO GIC) where each transmitter-$i$ (Tx-$i$), requires to send $d_i$ independent complex symbols per channel use that take values from fixed finite constellations with uniform distribution, to receiver-$i$ (Rx-$i$) for $i=1,2,\cdots,K$. We define the maximum rate achieved by Tx-$i$ using any linear precoder, when the interference channel-coefficients are zero, as the signal to noise ratio (SNR) tends to infinity to be the Constellation Constrained Saturation Capacity (CCSC) for Tx-$i$. We derive a high SNR approximation for the rate achieved by Tx-$i$ when interference is treated as noise and this rate is given by the mutual information between Tx-$i$ and Rx-$i$, denoted as $I[X_i;Y_i]$. A set of necessary and sufficient conditions on the precoders under which $I[X_i;Y_i]$ tends to CCSC for Tx-$i$ is derived. Interestingly, the precoders designed for interference alignment (IA) satisfy these necessary and sufficient conditions. Further, we propose gradient-ascent based algorithms to optimize the sum-rate achieved by precoding with finite constellation inputs and treating interference as noise. Simulation study using the proposed algorithms for a 3-user MIMO GIC with two antennas at each node with $d_i=1$ for all $i$, and with BPSK and QPSK inputs, show more than 0.1 bits/sec/Hz gain in the ergodic sum-rate over that yielded by precoders obtained from some known IA algorithms, at moderate SNRs.