Abstract

Reducing the energy consumption of communication networks while maintaining reliable service is an important challenge in sustainable computing. This capstone presents GreenNet, a simulation and evaluation framework designed to investigate whether energy-aware network control can reduce estimated energy consumption without unacceptable degradation in Quality of Service (QoS). GreenNet combines a graph-based network simulator, configurable traffic scenarios, and a web-based interface for examining network behavior and comparing control strategies. Three approaches are evaluated: an All-On baseline that keeps all links active, a rule-based heuristic that selectively deactivates links, and a hybrid controller based on Proximal Policy Optimization (PPO), whose proposed actions are checked against safety constraints. The strategies are compared under normal, burst, and hotspot traffic conditions using metrics including estimated energy consumption, delivered and dropped traffic, delay, path latency, and QoS violations. A reproducible benchmarking workflow records configurations and results, allowing performance comparisons to be inspected and repeated. The final benchmark identifies the heuristic controller as the strongest overall approach, while the PPO-based hybrid controller provides modest benefits under specific conditions. These findings indicate that AI-assisted control does not consistently outperform a carefully designed heuristic and that energy savings must be assessed alongside service quality. The main contribution of GreenNet is a reproducible platform for investigating the trade-offs between network energy efficiency and performance. The project provides a foundation for further experimentation with adaptive control strategies and transparent evaluation of sustainable networking approaches.

Publication Date

9-2026

Document Type

Senior Project

Student Type

Undergraduate

Advisor

Blerta Prevalla-Etemi

Advisor/Committee Member

Valon Raca

Advisor/Committee Member

Muzafer Shala

Campus

RIT Kosovo

Download

Share

COinS