Joint Trajectory and Resource Optimization of MEC-Assisted UAVs in Sub-THz Networks: A Resources-based Multi-Agent Proximal Policy Optimization DRL with Attention Mechanism

Abstract: THz band communication technology will be used in the 6G networks to enable high-speed and high-capacity data service demands. However, THz-communication losses arise owing to limitations, i.e., molecular absorption, rain attenuation, and coverage range. Furthermore, to maintain steady THz-communications and overcome coverage distances in rural and suburban regions, the required number of BSs is very high. Consequently, a new communication platform that enables aerial communication services is required. Furthermore, the airborne platform supports LoS communications rather than NLoS communications, which helps overcome these losses. Therefore, in this work, we investigate the deployment and resource optimization for MEC-enabled UAVs, which can provide THz-based communications in remote regions. To this end, we formulate an optimization problem to minimize the sum of the energy consumption of both MEC-UAV and MUs and the delay incurred by MUs under the given task information. The formulated problem is a MINLP problem, which is NP-hard. We decompose the main problem into two subproblems to address the formulated problem. We solve the first subproblem with a standard optimization solver, i.e., CVXPY, due to its convex nature. To solve the second subproblem, we design a RMAPPO DRL algorithm with an attention mechanism. The considered attention mechanism is utilized for encoding a diverse number of observations. This is designed by the network coordinator to provide a differentiated fit reward to each agent in the network. The simulation results show that the proposed algorithm outperforms the benchmark and yields a network utility which is $2.22\%$, $15.55\%$, and $17.77\%$ more than the benchmarks.