Electrical and Computer Engineering ETDs
Publication Date
Fall 11-15-2024
Abstract
Multi-access Edge Computing (MEC) is crucial for Internet of Things (IoT) applications by optimizing data processing and reducing latency. This thesis presents contributions to resource allocation and decision-making in edge computing environments. The TRUST-ME model is introduced, involving multiple edge servers and IoT devices (users) offloading computing tasks to MEC servers. A utility function is designed to assess latency and cost benefits for IoT devices using server resources. The core innovation is a novel trust model that evaluates IoT devices’ confidence in MEC servers by integrating both direct and indirect trust, based on interactions and feedback from other devices. In addition, a reinforcement learning framework with optimistic Q-learning and upper confidence bounds is proposed for autonomous IoT server selection. Additionally, a multilateral bargaining model ensures fair resource allocation based on computing demands. Simulations demonstrate the model’s effectiveness, scalability, and convergence, validated through real-world scenarios and comparisons with existing approaches.
Keywords
Multi-access Edge Computing, Internet of Things, Reinforcement Learning, Game Theory
Document Type
Thesis
Language
English
Degree Name
Computer Engineering
Level of Degree
Masters
Department Name
Electrical and Computer Engineering
First Committee Member (Chair)
Eirini Eleni Tsiropoulou
Second Committee Member
James Plusquellic
Third Committee Member
Symeon Papavassiliou
Recommended Citation
Tsikteris, Sean. "TRUST-ME: Resource Allocation and Server Selection Based on Trust in Multi-Access Edge Computing." (2024). https://digitalrepository.unm.edu/ece_etds/696
