Performance of SWIPT Systems Employing Practical TX Pulse Shaping Filters and Nonlinear EH Circuits with Memory

Simultaneous Wireless Information and Power Transfer (SWIPT) is a promising technology that enables wireless recharging of device batteries along with information transmission. A SWIPT system usually comprises a transmitter (TX) that sends a common signal to an information receiver and an energy harvester (EH). Therefore, the choice of the pulse shaping filter employed at the TX is an essential part of the SWIPT system design.

Recent studies for optimal SWIPT system design have shown that modeling of EH circuits based on Markov decision processes combined with deep learning techniques allows for studying the impact of nonlinear effects, such as EH circuit imperfections and memory, on the system performance. However, the models proposed in the literature to date, assume rectangular pulse shaping filter employed at the TX which is not feasible in practical transmission schemes due to bandwidth constraints. The goal of this project is to overcome this limitation and to study the performance of SWIPT systems employing practical pulse shaping filters at the TX.

Guidelines for the project:

  • To conduct a literature survey on SWIPT, Markov decision processes, and machine learning
  • To generalize the existing model for EH circuits based on a dense neural network and Markov decision process to take into account an arbitrary waveform
  • To analyze the performance of SWIPT system employing practical pulse shaping filters at the TX

Type of Project: Master thesis

If successful, this work may lead to a journal and/or conference paper.