eduzhai > Applied Sciences > Engineering >

Energy Minimization in UAV-Aided Networks Actor-Critic Learning for Constrained Scheduling Optimization

  • Save

... pages left unread,continue reading

Document pages: 12 pages

Abstract: In unmanned aerial vehicle (UAV) applications, the UAV s limited energysupply and storage have triggered the development of intelligentenergy-conserving scheduling solutions. In this paper, we investigate energyminimization for UAV-aided communication networks by jointly optimizingdata-transmission scheduling and UAV hovering time. The formulated problem iscombinatorial and non-convex with bilinear constraints. To tackle the problem,firstly, we provide an optimal relax-and-approximate solution and develop anear-optimal algorithm. Both the proposed solutions are served as offlineperformance benchmarks but might not be suitable for online operation. To thisend, we develop a solution from a deep reinforcement learning (DRL) aspect. Theconventional RL DRL, e.g., deep Q-learning, however, is limited in dealing withtwo main issues in constrained combinatorial optimization, i.e., exponentiallyincreasing action space and infeasible actions. The novelty of solutiondevelopment lies in handling these two issues. To address the former, wepropose an actor-critic-based deep stochastic online scheduling (AC-DSOS)algorithm and develop a set of approaches to confine the action space. For thelatter, we design a tailored reward function to guarantee the solutionfeasibility. Numerical results show that, by consuming equal magnitude of time,AC-DSOS is able to provide feasible solutions and saves 29.94 energy comparedwith a conventional deep actor-critic method. Compared to the developednear-optimal algorithm, AC-DSOS consumes around 10 higher energy but reducesthe computational time from minute-level to millisecond-level.

Please select stars to rate!

         

0 comments Sign in to leave a comment.

    Data loading, please wait...
×