A Spiking Neural Network Model for Indoor 2D Wireless User Positioning Based on CSI

Accurate and precise knowledge of the wireless user position is fundamental for both beamforming and beamtracking mechanisms. By accurately estimating the user's exact position, the base station (BS) or reconfigurable intelligent surface (RIS) can fine-Tune its beamforming vectors for optimal signal transmission. Furthermore, as the inference models migrate towards the network edge i.e., to the edge users, the need for energy efficient inference is imperative. Spiking neural networks (SNNs) and neuromorphic computing provide an energy-efficient framework for implementing and performing inference with significantly lower power consumption than conventional artificial neural networks (ANN) methods. This paper introduces an SNN architecture for wireless user positioning based on channel state information. Simulation results demonstrate that the proposed spiking model is capable of achieving significant accuracy, while consuming less power compared to an ANN-based model.