强化学习
继电器
计算机科学
感知
人工智能
群体行为
控制(管理)
工程类
噪音(视频)
动作(物理)
粒子群优化
控制工程
钢筋
模拟
钥匙(锁)
培训(气象学)
实时计算
特征(语言学)
出处
期刊:
[Institute of Electrical and Electronics Engineers]
日期:2026-03-20
卷期号:: 1867-1871
标识
DOI:10.1109/icaace69793.2026.11508652
摘要
The challenges inherent to the application of Integrated Sensing and Communication (ISAC) drone swarms include the fact that accuracy of sensing and quality of communication is mutually constrained, that the complexity of deploying resources makes them incompatible and unable to adapt to dynamic environments when performing collaborative tasks. This article presents a multi-UAV collaborative optimization model described as a deep reinforcement learning that balances the sensing and communication performance through the simultaneous optimization of routes of UAVs, beamforming, and power distribution. To begin with, an ISAC system model is created, and the optimization problem is categorized as a Markov Decision Process (MDP). Second, a Deep Deterministic Policy Gradient (DDPG) algorithm is to be used to address the issue. The Actor network generates the control policy of the continuous action space and the Critic network appraises the state-action value function. Lastly, the priority experience replay is presented to expedite convergence. Simulation results show that the proposed algorithm achieves a positioning error of 1.87 m in time slot 300, with a communication interruption probability as low as 3.1 %, and maintains an energy efficiency of 4.29 Mbit/J in a high-density scenario with 16 users. This scheme effectively solves the problem of multi-objective coupled optimization and provides a new approach for intelligent decision-making in ISAC UAV swarms.
科研通智能强力驱动
Strongly Powered by AbleSci AI