Abstract

With the unprecedented increase in mobile data demand and limited usable spectrum to provide for it, a paradigm shift towards spectrum sharing is a promising solution. However, there are many challenges that limit current spectrum sharing practices. One challenge is that proper spectrum sharing requires engaging devices to have an understanding of the impact they have on the ecosystem while transmitting in terms of spacial interference footprint and the implications to devices in their interference range. Another is that operators, especially licensed ones, have strict quality of service requirements for their subscribers, discouraging them from allowing unlicensed access of their purchased spectrum unless the sharing scheme can guarantee minimal impact to their systems. This dissertation seeks to address these challenges across three distinct works.

First, we used geographical features of a region to reduce in-field propagation experimentation by predicting the number of measurements required to accurately characterize its path loss, which can then be used to model coverage of arbitrarily positioned base stations. By exploiting the relationship between terrain feature complexity and measurement requirements, we found that the number of measurements collected to achieve a certain path loss accuracy over the entire region can be reduced by up to 58% in a high density drive testing scenario.

Next, we looked at applying Listen-Before-Talk (LBT) schemes in Citizens Broadband Radio Service (CBRS) networks for increasing the spatial reuse at secondary users while minimizing the interference footprint on incumbent and primary users. We used a novel Q-learning scheme to adapt the contention EDT to the changing network topology and traffic conditions, providing up to 350\% gains in average secondary node user perceived throughput (UPT) in certain difficult topologies with merely a 4% reduction in primary node UPT.

Finally, we studied channel selection in unlicensed Long-Term Evolution (LTE) cellular systems. We propose a mechanism for unlicensed LTE channel selection that not only takes into account interference to and from Wi-Fi access points but also considers other LTE operators in the unlicensed band. By collecting channel utilization statistics and sharing this information periodically with other unlicensed LTE base stations (eNBs), each eNB can improve their channel selection given their limited knowledge of the full topology via a proposed statistical and machine learning approach. We simulate operation in the unlicensed band using our channel selection algorithm and show how Wi-Fi load and inter-cell interference estimation can jointly be used to select transmission channels for all eNBs in the network.

Degree Date

Fall 2019

Document Type

Dissertation

Degree Name

Ph.D.

Department

Electrical and Computer Engineering

Advisor

Joseph Camp

Second Advisor

Dinesh Rajan

Third Advisor

Carlos Davila

Fourth Advisor

Ping Gui

Fifth Advisor

James Dunham

Number of Pages

121

Format

.pdf

Creative Commons License

Creative Commons Attribution-Noncommercial 4.0 License
This work is licensed under a Creative Commons Attribution-Noncommercial 4.0 License

Share

COinS