Adaptive Neural Signal Detection for Massive MIMO

TL;DR

This paper introduces MMNet, a deep learning-based MIMO detection method that outperforms existing approaches on realistic channels, requiring less computation and adapting online to different channel realizations.

Contribution

MMNet leverages iterative soft-thresholding theory and a novel training algorithm to improve MIMO detection performance on real-world channels with lower complexity.

Findings

Achieves near-optimal performance on i.i.d. Gaussian channels with 100x fewer operations.

Matches OAMPNet performance at 2.5dB lower SNR on correlated channels.

Outperforms classical MMSE detector by 4-8dB in error rate.

Abstract

Symbol detection for Massive Multiple-Input Multiple-Output (MIMO) is a challenging problem for which traditional algorithms are either impractical or suffer from performance limitations. Several recently proposed learning-based approaches achieve promising results on simple channel models (e.g., i.i.d. Gaussian). However, their performance degrades significantly on real-world channels with spatial correlation. We propose MMNet, a deep learning MIMO detection scheme that significantly outperforms existing approaches on realistic channels with the same or lower computational complexity. MMNet's design builds on the theory of iterative soft-thresholding algorithms and uses a novel training algorithm that leverages temporal and spectral correlation to accelerate training. Together, these innovations allow MMNet to train online for every realization of the channel. On i.i.d. Gaussian channels, MMNet requires two orders of magnitude fewer operations than existing deep learning schemes but achieves near-optimal performance. On spatially-correlated channels, it achieves the same error rate as the next-best learning scheme (OAMPNet) at 2.5dB lower SNR and with at least 10x less computational complexity. MMNet is also 4--8dB better overall than a classic linear scheme like the minimum mean square error (MMSE) detector.