A Decoupled Access Scheme With Reinforcement Learning Power Control for Cellular-Enabled UAVs

This article proposes a downlink/uplink decoupled (DUDe) access scheme for cellular-enabled unmanned aerial vehicle (UAV) communication systems. To minimize interference, the proposed separates control and data links of UAVs, as well uplinks (ULs) downlinks (DLs) ground users (GUEs), onto different serving base stations operating frequencies. Since power availability is major constraint in UAV communications, two allocation schemes based on $Q$ -learning (QL) deep (DQL) are to optimize energy efficiency (EE) this DUDe network. quantify improvements achieved, compared with fractional (FPC) used 4G 5G networks, convex optimization-based optimal scheme. The results demonstrate that can achieve up several times higher sum rates EE UL direction than its coupled counterparts. Moreover, it shown performance QL DQL approach surpass conventional FPC by 80%–100% UHF band, 160%–170% mmWave band.