Joint Uplink-and-Downlink Optimization of 3D UAV Swarm Deployment for Wireless-Powered NB-IoT Networks

TL;DR

This paper proposes a joint 3D deployment optimization of UAV swarms for wireless-powered NB-IoT networks, enhancing uplink throughput and fairness by considering energy harvesting, channel models, and scheduling policies.

Contribution

It introduces a comprehensive joint optimization framework for UAV deployment, device association, and scheduling in 3D space for IoT networks with energy and channel constraints.

Findings

NF policy yields higher sum throughput than FF policy.

FF policy provides better system fairness.

Joint 3D optimization improves UAV deployment efficiency.

Abstract

This paper investigates a full-duplex orthogonal-frequency-division multiple access (OFDMA) based multiple unmanned aerial vehicles (UAVs)-enabled wireless-powered Internet-of-Things (IoT) networks. In this paper, a swarm of UAVs is first deployed in three dimensions (3D) to simultaneously charge all devices, i.e., a downlink (DL) charging period, and then flies to new locations within this area to collect information from scheduled devices in several epochs via OFDMA due to potential limited number of channels available in Narrow Band IoT, i.e., an uplink (UL) communication period. To maximize the UL throughput of IoT devices, we jointly optimizes the UL-and-DL 3D deployment of the UAV swarm, including the device-UAV association, the scheduling order, and the UL-DL time allocation. In particular, the DL energy harvesting (EH) threshold of devices and the UL signal decoding threshold of UAVs are taken into consideration when studying the problem. Besides, both line-of-sight (LoS) and non-line-of-sight (NLoS) channel models are studied depending on the position of sensors and UAVs. The influence of the potential limited channels issue in NB-IoT is also considered by studying the IoT scheduling policy. Two scheduling policies, a near-first (NF) policy and a far-first (FF) policy, are studied. It is shown that the NF scheme outperforms FF scheme in terms of sum throughput maximization; whereas FF scheme outperforms NF scheme in terms of system fairness.