Multi-agent Resource Allocation Strategy for UAV Swarm-Based Cooperative Sensing

WANG Zhihong,LENG Supeng,XIONG Kai

DOI: https://doi.org/10.11959/j.issn.2096-3750.2023.00326

2023-01-01

Abstract:Driven by the development of intelligent internet of things（IoT） technology, unmanned aerial vehicle（UAV）swarms have been widely used for sensing and monitoring in emergency and rescue scenarios. The UAVs automatically sense and discover mission targets in the mission area, recruiting neighboring UAVs to form perception and computation task groups to collaboratively complete the perception, acquisition and processing of data. However, repetitive sensory data and imbalance in the supply and demand of computational resources between multiple tasks cause additional computational and communication overheads and increase the end-to-end processing latency. To address this challenge, a multi-task resource allocation approach combining bionics and multi-agent independent reinforcement learning was proposed, making collaborative resource allocation decisions based on local task information. The method represents the resource requirements of individual tasks as situational information concentrations and dynamically updates the heterogeneous resource requirements of each task by spreading the situational information across task groups. At the same time, it combines multi-agent independent reinforcement learning methods for intelligent decision making in order to collaboratively allocate the heterogeneous resources of each task. Simulation results show that this solution can not only effectively reduce the task execution time, but also significantly improve the computational resource utilization.