Supervised Contrastive CSI Representation Learning for Massive MIMO Positioning

TL;DR

This paper introduces a supervised contrastive learning approach using deep CNNs to improve similarity measurement in CSI data for massive MIMO positioning, resulting in enhanced accuracy over existing methods.

Contribution

It presents a novel contrastive learning framework for CSI similarity that outperforms current state-of-the-art positioning techniques.

Findings

Significant improvement in positioning accuracy.

Effective learning of CSI similarity metrics.

Robustness demonstrated on real-world datasets.

Abstract

Similarity metric is crucial for massive MIMO positioning utilizing channel state information~(CSI). In this letter, we propose a novel massive MIMO CSI similarity learning method via deep convolutional neural network~(DCNN) and contrastive learning. A contrastive loss function is designed considering multiple positive and negative CSI samples drawn from a training dataset. The DCNN encoder is trained using the loss so that positive samples are mapped to points close to the anchor's encoding, while encodings of negative samples are kept away from the anchor's in the representation space. Evaluation results of fingerprint-based positioning on a real-world CSI dataset show that the learned similarity metric improves positioning accuracy significantly compared with other known state-of-the-art methods.