Electrical and Computer Engineering ETDs

Publication Date

Summer 7-7-2017

Abstract

Spectrum utilization is vital for mobile operators. It ensures an efficient use of spectrum bands, especially when obtaining their license is highly expensive. Long Term Evolution (LTE), and LTE-Advanced (LTE-A) spectrum bands license were auctioned by the Federal Communication Commission (FCC) to mobile operators with hundreds of millions of dollars. In the first part of this dissertation, we study, analyze, and compare the QoS performance of QoS-aware/Channel-aware packet scheduling algorithms while using CA over LTE, and LTE-A heterogeneous cellular networks. This included a detailed study of the LTE/LTE-A cellular network and its features, and the modification of an open source LTE simulator in order to perform these QoS performance tests. In the second part of this dissertation, we aim to solve spectrum underutilization by proposing, implementing, and testing two novel multi-agent Q-learning-based packet scheduling algorithms for LTE cellular network. The Collaborative Competitive scheduling algorithm, and the Competitive Competitive scheduling algorithm. These algorithms schedule licensed users over the available radio resources and un-licensed users over spectrum holes. In conclusion, our results show that the spectrum band could be utilized by deploying efficient packet scheduling algorithms for licensed users, and can be further utilized by allowing unlicensed users to be scheduled on spectrum holes whenever they occur.

Keywords

LTE-Advanced (LTE-A), Radio Resource Management, Packet Scheduling, Cognitive Radio, Multi-agent Q-learning

Document Type

Dissertation

Language

English

Degree Name

Computer Engineering

Level of Degree

Doctoral

Department Name

Electrical and Computer Engineering

First Committee Member (Chair)

Prof., Manel Martinez-Ramon

Second Committee Member

Prof., Gregory Heileman

Third Committee Member

Prof., Nasir Ghani

Fourth Committee Member

Prof., Christopher Lamb

Download

Share

COinS