Non-cooperative Aerial Base Station Placement via Stochastic\n Optimization

2019/05/10 by Daniel Romero, Romero, Daniel, Geert Leus +1
Engineering · Computer Science · #UAV Applications and Optimization #Distributed Control Multi-Agent Systems #Satellite Communication Systems

paper · pdf · doi:10.48550/arxiv.1905.03988

Abstract

Autonomous unmanned aerial vehicles (UAVs) with on-board base station\nequipment can potentially provide connectivity in areas where the terrestrial\ninfrastructure is overloaded, damaged, or absent. Use cases comprise emergency\nresponse, wildfire suppression, surveillance, and cellular communications in\ncrowded events to name a few. A central problem to enable this technology is to\nplace such aerial base stations (AirBSs) in locations that approximately\noptimize the relevant communication metrics. To alleviate the limitations of\nexisting algorithms, which require intensive and reliable communications among\nAirBSs or between the AirBSs and a central controller, this paper leverages\nstochastic optimization and machine learning techniques to put forth an\nadaptive and decentralized algorithm for AirBS placement without inter-AirBS\ncooperation or communication. The approach relies on a smart design of the\nnetwork utility function and on a stochastic gradient ascent iteration that can\nbe evaluated with information available in practical scenarios. To complement\nthe theoretical convergence properties, a simulation study corroborates the\neffectiveness of the proposed scheme.\n

Related

• Beyond Empirical Models: Pattern Formation Driven Placement of UAV Base Stations
• Learning in the Sky: An Efficient 3D Placement of UAVs
• Deployment Strategies of Multiple Aerial BSs for User Coverage and Power Efficiency Maximization
• Analysis of Reinforcement Learning Schemes for Trajectory Optimization of an Aerial Radio Unit
• Placement Optimization of Aerial Base Stations with Deep Reinforcement Learning