About OpenKnowledge@NAU | For NAU Authors

Spectrum sharing and management in Unmanned Aerial Vehicle networks

Shamsoshoara, Alireza (2021) Spectrum sharing and management in Unmanned Aerial Vehicle networks. Doctoral thesis, Northern Arizona University.

[thumbnail of Shamsoshoara_2021_spectrum_sharing_management_unmanned_aerial_vehicle_.pdf] Text
Shamsoshoara_2021_spectrum_sharing_management_unmanned_aerial_vehicle_.pdf - Published Version
Restricted to Repository staff only

Download (24MB) | Request a copy


Small autonomous vehicles received a lot of attention in commercial, military, and personal applications in the recent era. The fast growth in the number of these vehicles raises many challenges in terms of safety, primary, security, communications, control and coordination in such large scale autonomous systems. One of these challenges is to secure the required radio spectrum for reliable communication and data transmission among the vehicles and the ground station. Spectrum scarcity is one of the key challenges toward developing emerging applications such as Unmanned Aerial Vehicle (UAV) networks, where the agents are mobile, have limited energy and computation capabilities while often requiring high data transmission rates. In many previous works, the UAVs are considered as flying base stations (BSs) to extend the coverage of cellular networks or as a relay to enable device-to-device (D2D) communications. In this research, we investigate the operations of UAV networks in disaster scenarios, where the communication infrastructure may be damaged and develop different coordination mechanisms between the UAV networks and the ground users. We study a scenario where the UAVs suffer from spectrum scarcity in their network and they require to lease additional spectrum from ground users in exchange to deliver their packets. Many parameters such as mobility, throughput, fairness, and reliability affect the task of spectrum assignment. This study aims at developing new approaches using reinforcement learning to solve this spectrum assignment problem. Moreover, packet scheduling is one of the techniques that the relay UAV considers into account for choosing the ground users to service. Many factors such as the Quality-of-Service (QoS), queue length, application service time, and energy can play an important role in packet scheduling. We studied an imitation learning approach (Behavioral cloning) to mimic an expert behavior to perform an autonomous packet scheduling task. On the other hand, compared to the applications such as flying base station or flying relay, UAVs can connect to the cellular BS as flying User Equipment (UE). In applications like this, it is crucial to have an interference management scheme to minimize the interfering effect on the ground UEs as well. In this study, we address this challenge by using apprenticeship learning via inverse reinforcement learning to account for the interference concern based on expert behavior. Finally, a practical implementation of drones is performed during a prescribed pile fire at Northern Arizona Coconino forest to collect aerial imagery using 4K cameras and a thermal camera. The collected data is used to define a dataset for different challenges such as "Fire and NoFire'' classification and fire segmentation.

Item Type: Thesis (Doctoral)
Publisher’s Statement: © Copyright is held by the author. Digital access to this material is made possible by the Cline Library, Northern Arizona University. Further transmission, reproduction or presentation of protected items is prohibited except with permission of the author.
Keywords: Reinforcement Learning; Spectrum management; Spectrum sharing; Unmanned Aerial Vehicles; Wireless Communication; Drones
Subjects: T Technology > TJ Mechanical engineering and machinery
NAU Depositing Author Academic Status: Student
Department/Unit: Graduate College > Theses and Dissertations
College of Engineering, Informatics, and Applied Sciences > School of Informatics, Computing, and Cyber Systems
Date Deposited: 14 Feb 2022 18:52
Last Modified: 14 Feb 2022 18:52
URI: https://openknowledge.nau.edu/id/eprint/5700

Actions (login required)

IR Staff Record View IR Staff Record View


Downloads per month over past year