D3QN-Based IAB Resource Allocation and Tethered UAV Positioning for IoT Networks

Abstract

The use of tethered uncrewed aerial vehicles (TUAVs) is promising for addressing the energy-constraint problems associated with battery-powered aerial vehicles. In addition, integrated access and backhaul (IAB) technology allows the simultaneous exploitation of the same frequency band for both access and backhaul links, thus increasing resource utilization efficiency in air-ground integrated Internet of Things (IoT) networks. However, the joint optimization of TUAV deployment and IAB bandwidth allocation is an extremely complicated problem, particularly when considering the dynamic characteristics of TUAV-aided IAB network environments. Therefore, we herein propose a distributed double deep Q-network (D3QN)-based optimal resource allocation and a TUAV deployment algorithm to maximize the network-wide sum rate. By performing extensive simulations, it is shown that the proposed algorithm significantly improves the network-wide sum rate compared with several benchmark algorithms, such as the reward-optimal, random action, fixed channel allocation, fixed transmit power allocation, fixed TUAV positioning, distributed Q-learning, distributed DQN, and centralized DDQN algorithms.