Energy Outage and Deep Learning Prediction for Aerial Vehicle-Based Energy Harvesting Networks with Co-Channel Interference

Abstract

This research investigates the Energy Outage Probability (EoP) performance of autonomous vehicle (AV) swarm-aided energy harvesting (EH) networks with limitations in relaying communication protocols and co-channel interference (CCI) impacts. In particular, we propose a framework that combines opportunistic AV selection with amplify-and forward (AF) relaying protocols, accounting for CCI induced by multiple interferers. Under Nakagami-m fading channels, the study examines CCI impacts on the EoP of AV swarm networks, where AV selection ensures connectivity between source and destination. Analytical approximation expressions for EoP are derived, considering full interference impacts and three special cases: CCI absent at AVs, CCI absent at the destination, and CCI absent at both. To enhance predictive capabilities, a deep learning (DL) solution is introduced for real-time coverage energy probability estimation, optimizing AV selection, and minimizing network EoP. Numerical results, validated through Monte-Carlo simulations, confirm the accuracy of the mathematical framework, the efficacy of the DL solution, and reveal insights into key design parameters influencing network performance.

Keywords: Simultaneous wireless information and power transfer (SWIPT), Co-channel interference (CCI), Unmanned aerial vehicles (UAVs), swarm intelligence, deep learning-based energy prediction.