A Finite-Range Search Formulation of Maximum Likelihood MIMO Detection for Coherent Ising Machines

TL;DR

This paper introduces a novel Ising formulation and an enhanced Coherent Ising Machine model for MIMO detection, significantly improving BER performance and spectral efficiency in large MIMO systems with high-order modulations.

Contribution

It presents the first Ising solver that achieves substantial BER and throughput gains for large MIMO systems and high-order modulations, outperforming existing physics-inspired methods.

Findings

Significant BER improvements for 16-QAM and higher modulations.

Achieves 2x throughput at SNR ≤ 25 dB in large MIMO systems.

Maintains performance gains at 256-QAM modulation.

Abstract

The last couple of years have seen an emergence of physics-inspired computing for maximum likelihood MIMO detection. These methods involve transforming the MIMO detection problem into an Ising minimization problem, which can then be solved on an Ising Machine. Recent works have shown promising projections for MIMO wireless detection using Quantum Annealing optimizers and Coherent Ising Machines. While these methods perform very well for BPSK and 4-QAM, they struggle to provide good BER for 16-QAM and higher modulations. In this paper, we explore an enhanced CIM model, and propose a novel Ising formulation, which together are shown to be the first Ising solver that provides significant gains in the BER performance of large and massive MIMO systems, like $16\times16$ and $16\times32$, and sustain its performance gain even at 256-QAM modulation. We further perform a spectral efficiency analysis and show that, for a $16\times16$ MIMO with Adaptive Modulation and Coding, our method can provide substantial throughput gains over MMSE, achieving $2\times$ throughput for SNR $\leq25$ dB, and up to $1.5\times$ throughput for SNR $\geq 30$ dB.