Deep Convolutional Neural Networks for Massive MIMO Fingerprint-Based Positioning

TL;DR

This paper explores using convolutional neural networks to improve fingerprint-based positioning with massive MIMO channels, demonstrating that CNNs can achieve high accuracy with sufficient training data.

Contribution

It introduces the application of CNNs to massive MIMO channel fingerprints for positioning, showing their effectiveness in achieving high accuracy.

Findings

Moderately deep CNNs can achieve fractional-wavelength positioning accuracy.

CNNs effectively learn sparse structures in massive MIMO channels.

Training data quality and quantity are crucial for high accuracy.

Abstract

This paper provides an initial investigation on the application of convolutional neural networks (CNNs) for fingerprint-based positioning using measured massive MIMO channels. When represented in appropriate domains, massive MIMO channels have a sparse structure which can be efficiently learned by CNNs for positioning purposes. We evaluate the positioning accuracy of state-of-the-art CNNs with channel fingerprints generated from a channel model with a rich clustered structure: the COST 2100 channel model. We find that moderately deep CNNs can achieve fractional-wavelength positioning accuracies, provided that an enough representative data set is available for training.