Resource Management for Transmit Power Minimization in UAV-Assisted RIS HetNets Supported by Dual Connectivity

• Published in: IEEE transactions on wireless communications 2022-03, Vol.21 (3), p.1806-1822
• Main Authors: Khalili, Ata, Monfared, Ehsan Mohammadi, Zargari, Shayan, Javan, Mohammad Reza, Yamchi, Nader Mokari, Jorswieck, Eduard Axel
• Format: Article
• Language: English
• Subjects: Algorithms; Approximation; Beamforming; deep Q-network (DQN) learning; Line of sight; Machine learning; Millimeter waves; Minimization; NOMA; non-orthogonal multiple access (NOMA); Nonorthogonal multiple access; Optimization; reconfigurable intelligent surface (RIS); Relays; Resource allocation; Resource management; Subcarriers; Trajectory; Unmanned aerial vehicle (UAV); Unmanned aerial vehicles; Wireless communication
• ISSN: 1536-1276; 1558-2248
• DOI: 10.1109/TWC.2021.3107306

This paper proposes a novel approach to improve the performance of a heterogeneous network (HetNet) supported by dual connectivity (DC) by adopting multiple unmanned aerial vehicles (UAVs) as passive relays that carry reconfigurable intelligent surfaces (RISs). More specifically, RISs are deployed under the UAVs termed as UAVs-RISs that operate over the micro-wave ( \mu \text{W} ) channel in the sky to sustain a strong line-of-sight (LoS) connection with the ground users. The macro-cell operates over the \mu \text{W} channel based on orthogonal multiple access (OMA), while small base stations (SBSs) operate over the millimeter-wave (mmW) channel based on non-orthogonal multiple access (NOMA). We study the problem of total transmit power minimization by jointly optimizing the trajectory/velocity of each UAV, RISs' phase shifts, subcarrier allocations, and active beamformers at each BS. The underlying problem is highly non-convex and the global optimal solution is intractable. To handle it, we decompose the original problem into two subproblems, i.e., a subproblem which deals with the UAVs' trajectories/velocities, RISs' phase shifts, and subcarrier allocations for \mu \text{W} ; and a subproblem for active beamforming design and subcarrier allocation for mmW. In particular, we solve the first subproblem via the dueling deep Q-Network (DQN) learning approach by developing a distributed algorithm which leads to a better policy evaluation. Then, we solve the active beamforming design and subcarrier allocation for the mmW via the successive convex approximation (SCA) method. Simulation results exhibit the effectiveness of the proposed resource allocation scheme compared to other baseline schemes. In particular, it is revealed that by deploying UAVs-RISs, the transmit power can be reduced by 6 dBm while maintaining similar guaranteed QoS.