Resource Allocation in V2I Link Between Connected Autonomous Vehicles and 5G mm-Wave Band Small-Cells Using Machine Learning

Raeisi Ziarani, Mostafa

Resource Allocation in V2I Link Between Connected Autonomous Vehicles and 5G mm-Wave Band Small-Cells Using Machine Learning

dc.contributor.advisor	Sesay, Abu B.
dc.contributor.author	Raeisi Ziarani, Mostafa
dc.contributor.committeemember	Fapojuwo, Abraham
dc.contributor.committeemember	Far, Behrouz
dc.date	2024-05
dc.date.accessioned	2024-04-30T15:47:34Z
dc.date.available	2024-04-30T15:47:34Z
dc.date.issued	2024-04-23
dc.description.abstract	This dissertation presents innovative contributions aimed at enhancing resource allocation for high-speed vehicular users within the Fifth-Generation (5G) networks. The study addresses challenges in segregating users based on velocities and introduces a customized distance metric. When applied in K-means clustering, this metric yields optimal results, particularly in scenarios involving the separation of high-speed and low-speed users. The validation of this metric is rigorously examined through mathematical analysis and exhaustive search on distance metric criteria. Furthermore, numerical simulations illustrate the efficacy of the K-Means algorithm when utilizing the proposed distance metric to segregate users across various scenarios and dimensions. A new user-centric channel allocation scheme, known as Vehicular Frequency Reuse (VFR), is introduced for 5G networks, with a specific focus on millimeter-wave (mm-wave) band small cells. Accompanied by an innovative cell reselection procedure that is designed to adapt the network configurations to user mobility. A novel mobility management function seamlessly integrates the proposed VFR scheme and cell reselection procedure into the 5G mobility management framework. This integrated function significantly reduces handover rates and enhances link reliability in 5G network for high-speed road users, such as Connected Autonomous Vehicles (CAVs). The proposed Distance-Threshold metric is employed to assess the frequency reuse ratio within this network. Additionally, a Velocity-Threshold metric, calculated using a k-means algorithm and the proposed distance metric, simplifies the segregation between low-speed and high-speed users based solely on velocity comparison. This approach reduces complexity in real-time user separation processes while maintaining the clustering algorithm of the real-time pipeline of the system. This research also presents a novel approach to power control in vehicular 5G-connected networks using Deep Reinforcement Learning (DRL). Focusing on optimizing power allocation for CAVs in mm-wave bands between CAVs and Roadside Units (RSUs), the goal is to achieve the demanded uplink transmission capacity while minimizing power consumption and co-channel interference. Implemented through the Proximal Policy Optimization (PPO) algorithm within a modified actor-critic architecture, a Deep Neural Network (DNN) model guides decision-making. The proposed method is integrable with existing 3rd Generation Partnership Project (3GPP)-based 5G architecture with minimal changes, leveraging quantized information from cellular users’ measurement reports for compatibility. Simulation results across varied road conditions demonstrate the superior performance of the proposed algorithm compared to conventional 3GPP-based power control algorithm. Future research directions are identified to enhance these contributions. Highlighted is the integration of adaptive beamforming with the proposed resource allocation to enhance energy and spectrum efficiency further. Additionally, exploring other dimensions of resource allocation and technologies including Sixth-Generation (6G) is suggested. In conclusion, the thesis addresses critical challenges in vehicular 5G networks, offering innovative solutions for clustering, channel allocation, and power control. These contributions lay the foundation for enhanced network efficiency, reliability, and performance in high-speed vehicular environments, with future research directions poised to further push the boundaries of wireless technology.
dc.identifier.citation	Raeisi Ziarani, M. (2024). Resource allocation in V2I link between connected autonomous vehicles and 5G mm-wave band small-cells using machine learning (Doctoral thesis, University of Calgary, Calgary, Canada). Retrieved from https://prism.ucalgary.ca.
dc.identifier.uri	https://hdl.handle.net/1880/118491
dc.identifier.uri	https://doi.org/10.11575/PRISM/43333
dc.language.iso	en
dc.publisher.faculty	Graduate Studies
dc.publisher.institution	University of Calgary
dc.rights	University of Calgary graduate students retain copyright ownership and moral rights for their thesis. You may use this material in any way that is permitted by the Copyright Act or through licensing that has been assigned to the document. For uses that are not allowable under copyright legislation or licensing, you are required to seek permission.
dc.subject	Vehicular Networks
dc.subject	5G
dc.subject	mm-Wave
dc.subject	Connected Autonomous Vehicles
dc.subject	Machine Learning
dc.subject	Reinforcement Learning
dc.subject	Vehicle-to-Infrastructure
dc.subject	Channel Allocation
dc.subject	Power Allocation
dc.subject.classification	Engineering--Electronics and Electrical
dc.title	Resource Allocation in V2I Link Between Connected Autonomous Vehicles and 5G mm-Wave Band Small-Cells Using Machine Learning
dc.type	doctoral thesis
thesis.degree.discipline	Engineering – Electrical & Computer
thesis.degree.grantor	University of Calgary
thesis.degree.name	Doctor of Philosophy (PhD)
ucalgary.thesis.accesssetbystudent	I do not require a thesis withhold – my thesis will have open access and can be viewed and downloaded publicly as soon as possible.