.png){kind=icon}
.png){kind=icon}
Return
Joint trajectory and spectrum allocation optimization for UAV in complex electromagnetic environments: A multi-branch DRL approach
.png){kind=icon}
PRE
AI
Y
Y
W
W
Y
S
DOI：10.1016/j.phycom.2026.103076.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
In complex electromagnetic environments, Unmanned Aerial Vehicles (UAVs) face the dual challenges of avoiding Non-Flying Zones (NFZs) and counteracting dynamic hybrid jamming during mission execution, which necessitates the joint optimization of trajectory and spectrum allocation. To solve the problem, an Adaptive Sampling multi-Branch Channel-Enhanced Double Deep Q-Network (ASBCE-DDQN) algorithm is proposed. The algorithm integrates three key innovations: (1) the adaptive experience replay mechanism that prioritizes crucial communication experiences through a dynamic sampling strategy; (2) the multi-branch network for communication optimization that processes heterogeneous states and designs independent output branches for action dimensions; and (3) the enhanced channel selection module that adaptively adjusts channel Q-values via dynamic weighting. Furthermore, a composite reward function is designed to provide precise guidance for the UAV in balancing trajectory and spectrum allocation. Simulation results demonstrate that the proposed algorithm achieves a mission success rate of 91.93%, a communication success rate of 97.22% and an average flight steps reduction to 109.74 compared with baselines, confirming its effectiveness in joint trajectory and spectrum allocation optimization.
Keywords:
UAV
DRL
Joint optimization
Trajectory
Spectrum allocation
Journal
IF:
2.2
Papers:
279
Citations:
2.6K
Organization
Cited Papers