High Altitude Platform-Based Caching and Multicasting for Rural Connectivity

TL;DR

This paper proposes a hierarchical deep reinforcement learning and convex optimization framework for energy-efficient content delivery using high-altitude platforms with caching and multicasting, improving rural connectivity.

Contribution

It introduces a novel joint optimization approach combining DRL and convex optimization for caching and resource allocation in HAP-based rural networks.

Findings

Significant power cost reduction compared to baselines

Effective use of network coding-based multicasting

Enhanced rural connectivity through optimized HAP deployment

Abstract

Providing efficient and reliable content delivery in rural areas remains a significant challenge due to the lack of communication infrastructure. To bridge the digital divide, this paper investigates the potential of leveraging multiple high-altitude platforms (HAPs) for energy-efficient content delivery in wide rural regions. Each caching-enabled HAP is equipped with both Free-Space Optical (FSO) transceivers for backhaul links and Radio Frequency (RF) antenna arrays for access links. To further enhance network efficiency, we consider a network coding-based multicasting scheme, where different types of content are treated as distinct multicast sessions. With the objective of minimizing long-term power cost, we propose a hierarchical framework that integrates deep reinforcement learn-ing (DRL) and convex optimization to jointly optimize dynamic caching strategies and resource allocation across the network. Simulation results demonstrate that our approach significantly reduces power cost compared to several baseline approaches, providing a practical solution for improving rural connectivity.