D2D-Based Cellular-Connected UAV Swarm Control Optimization via Graph-Aware DRL

Research output: Contribution to conference typesPaperpeer-review

1 Citation (Scopus)

Abstract

Cellular-connected unmanned aerial vehicle (UAV) swarm is a promising solution for diverse applications, including cargo delivery and traffic control. However, it is still challenging to communicate with and control the UAV swarm with high reliability and low latency. In this paper, we propose a two-phase command and control (C&C) transmission scheme in cellular-connected UAV swarm network, where the ground base station (GBS) broadcasts the common C&C message in Phase I, the UAVs that have successfully decoded the C&C message will then relay the message to the rest of UAVs via device-to-device (D2D) communications under individual latency constraint. To maximize the number of UAVs that receive the message successfully within the latency constraint, we formulate the problem as a decentralized and partially observable Markov process for finding the optimal policies. To address this problem, we further develop a fully decentralized graph attention network (GAT)-based reinforcement learning algorithm to optimize the D2D pair selection, where the GAT is utilized to exploit the dynamic topology information of cellular-connected UAV swarm network. Simulation results show that our algorithm outperforms the other two baselines and could achieve a cooperative target under a mobile UAV swarm scenario.

Original languageEnglish
Pages1326-1331
Number of pages6
DOIs
Publication statusPublished - 2022
Event2022 IEEE Global Communications Conference, GLOBECOM 2022 - Virtual, Online, Brazil
Duration: 4 Dec 20228 Dec 2022

Conference

Conference2022 IEEE Global Communications Conference, GLOBECOM 2022
Country/TerritoryBrazil
CityVirtual, Online
Period4/12/20228/12/2022

Keywords

  • Cellular-connected UAV swarm network
  • D2D
  • graph attention
  • multi agent reinforcement learning

Fingerprint

Dive into the research topics of 'D2D-Based Cellular-Connected UAV Swarm Control Optimization via Graph-Aware DRL'. Together they form a unique fingerprint.

Cite this