Link to the University of Pittsburgh Homepage
Link to the University Library System Homepage Link to the Contact Us Form

Analyzing Usage Conflict Situations in Localized Spectrum Sharing Scenarios: An Agent-Based Modeling and Machine Learning Approach

Bustamante, Pedro (2022) Analyzing Usage Conflict Situations in Localized Spectrum Sharing Scenarios: An Agent-Based Modeling and Machine Learning Approach. Doctoral Dissertation, University of Pittsburgh. (Unpublished)

Download (15MB) | Preview


As spectrum sharing matures, different approaches have been proposed for a more efficient allocation, assignment, and usage of spectrum resources. These approaches include cognitive radios, multi-level user definitions, radio environment maps, among others. However, spectrum usage conflicts (e.g., "harmful" interference) remain a common challenge in spectrum sharing schemes. In particular, in conflict situations where it is necessary to take actions to ensure the sound operations of sharing agreements. A typical example of a usage conflict is where incumbents' tolerable levels of interference (i.e., interference thresholds) are surpassed. In this work, we present a new method to examine and study spectrum usage conflicts. A fundamental goal of this project is to capture local resource usage patterns to provide more realistic estimates of interference. For this purpose, we have defined two spectrum and network-specific characteristics that directly impact the local interference assessment: resource access strategy and governance framework. Thus, we are able to test the viability in spectrum sharing situations of distributed or decentralized governance systems, including polycentric and self-governance. In addition, we are able to design, model, and test a multi-tier spectrum sharing scheme that provides stakeholders with more flexible resource access opportunities.

To perform this dynamic and localized study of spectrum usage and conflicts, we rely on Agent-Based Modeling (ABM) as our main analysis instrument. A crucial component for capturing local resource usage patterns is to provide agents with local information about their spectrum situation. Thus, the environment of the models presented in this dissertation are given by the REM's Interference Cartography (IC) map. Additionally, the agents' definitions and actions are the results of the interaction of the technical aspects of resource access and management, stakeholder interactions, and the underlying usage patterns as defined in the Common Pool Resource (CPR) literature. Finally, to capture local resource usage patterns and, consequently, provide more realistic estimates of conflict situations, we enhance the classical rule-based ABM approach by using Machine Learning (ML) techniques. Via ML algorithms, we refine the internal models of agents in an ABM. Thus, the agents' internal models allow them to choose more suitable responses to changes in the environment.


Social Networking:
Share |


Item Type: University of Pittsburgh ETD
Status: Unpublished
CreatorsEmailPitt UsernameORCID
Bustamante, Pedropjb63@pitt.edupjb630000-0001-9952-5453
ETD Committee:
TitleMemberEmail AddressPitt UsernameORCID
Committee ChairWeiss, Martinmbw@pitt.eduMBW0000-0001-6785-0913
Committee MemberKrishnamurthy,
Committee MemberYurko,
Committee MemberEdward,
Date: 17 January 2022
Date Type: Publication
Defense Date: 22 November 2021
Approval Date: 17 January 2022
Submission Date: 17 December 2021
Access Restriction: No restriction; Release the ETD for access worldwide immediately.
Number of Pages: 348
Institution: University of Pittsburgh
Schools and Programs: School of Computing and Information > Telecommunications
Degree: PhD - Doctor of Philosophy
Thesis Type: Doctoral Dissertation
Refereed: Yes
Uncontrolled Keywords: spectrum sharing, governance mechanisms, machine learning, agent-based modeling, citizen broadband radio service, telecommunications regulation, common pool resources, polycentric governance, self-governance
Date Deposited: 17 Jan 2022 15:05
Last Modified: 17 Jan 2022 15:05


Monthly Views for the past 3 years

Plum Analytics

Actions (login required)

View Item View Item