A Learning Approach To Wireless Information and Power Transfer Signal and System Design

TL;DR

This paper introduces a deep learning-based method for designing SWIPT systems, optimizing transmitter and receiver operations jointly under a nonlinear energy harvester model, and demonstrates results consistent with theoretical expectations.

Contribution

It presents a novel DNN autoencoder approach for joint optimization of SWIPT systems considering nonlinear energy harvesters, aligning with theoretical insights.

Findings

Optimized modulation constellations vary with energy demand.

Symbols concentrate around the origin as energy demand increases.

Results match previous theoretical predictions.

Abstract

The end-to-end learning of Simultaneous Wireless Information and Power Transfer (SWIPT) over a noisy channel is studied. Adopting a nonlinear model for the energy harvester (EH) at the receiver, a joint optimization of the transmitter and the receiver is implemented using Deep Neural Network (DNN)-based autoencoders. Modulation constellations for different levels of "power" and "information rate" demand at the receiver are obtained. The numerically optimized signal constellations are inline with the previous theoretical results. In particular, it is observed that as the receiver energy demand increases, all but one of the modulation symbols are concentrated around the origin and the other symbol is shot away from the origin along either the real or imaginary subchannel.