Learning-Based Symbol Level Precoding: A Memory-Efficient Unsupervised Learning Approach

TL;DR

This paper introduces an unsupervised deep neural network approach for symbol level precoding that significantly reduces memory usage through quantization, improving efficiency in multiuser downlink transmission.

Contribution

It presents a novel unsupervised learning framework for SLP using quantized DNNs, achieving substantial memory savings and model compression compared to full-precision methods.

Findings

Achieves ~21x memory reduction with binary DNNs

Achieves ~13x memory reduction with ternary DNNs

Maintains effective interference management in multiuser downlink

Abstract

Symbol level precoding (SLP) has been proven to be an effective means of managing the interference in a multiuser downlink transmission and also enhancing the received signal power. This paper proposes an unsupervised learning based SLP that applies to quantized deep neural networks (DNNs). Rather than simply training a DNN in a supervised mode, our proposal unfolds a power minimization SLP formulation in an imperfect channel scenario using the interior point method (IPM) proximal `log' barrier function. We use binary and ternary quantizations to compress the DNN's weight values. The results show significant memory savings for our proposals compared to the existing full-precision SLP-DNet with significant model compression of ~21x and ~13x for both binary DNN-based SLP (RSLP-BDNet) and ternary DNN-based SLP (RSLP-TDNets), respectively.