Design and Evaluation of Interference Handling Mechanisms in Flow-Level Simulators

Abstract

Wireless distributed systems face significant challenges when multiple devices share limited spectrum resources, leading to signal interference that corrupts data and impacts system performance. Flow-level network simulators, while computationally efficient, typically struggle to accurately model wireless interference scenarios, requiring users to implement complex collision handling mechanisms themselves. This complexity creates a barrier for researchers seeking to use flow-level simulation for studying distributed systems in wireless environments.

This thesis addresses the challenge of simplifying interference management in the Extensible Simulator for Distributed Systems (ESDS), a flow-level simulator for wireless networks. A novel interference handler plugin is designed and implemented that abstracts away collision detection and management, incorporating core principles from the IEEE 802.11 MAC protocol including medium sensing, acknowledgment mechanisms, and exponential backoff schemes. The implementation adapts packet-level MAC concepts to the flow-level simulation environment while addressing the inherent abstraction gap between these two approaches.

The research reveals fundamental limitations when adapting packet-level MAC protocols to flow-level simulation environments. Key challenges include ESDS's inability to continuously sense the medium, which prevents direct implementation of priority access mechanisms and precise timing features found in the 802.11 protocol.