An Agent-Based Model of a Responsive Network Monitoring System for Internet Service Providers

Source abstract

The purpose of this project was to develop a responsive Agent based network monitoring system designed to reduce the time taken by network administrators to initiate monitoring of network nodes and making any changes and updates on the monitoring parameters. This work presents an agent-based model (ABM) for a responsive network monitoring system tailored to Internet Service Providers (ISPs). The model employs autonomous agents that monitor traffic, detect anomalies, coordinate remediation, and adapt to changing network conditions. Agents interact using a publish/subscribe communication bus, and the model supports evaluation of detection accuracy, time-to-respond, false alarm rate, overhead, and resilience. The ABM is implemented with an event-driven simulation usign Mesa in Python and is validated through scenario experiments (normal load, flash crowds, link failures, DDoS). The existing system does not auto respond to changes and modifications in the network as hard coding is required to make the necessary changes. This system used an agent-based model (ABM) that provides a solution that works by reducing the time taken to initiate and make any changes in the monitoring parameters by introducing intelligence to the system. The methodology used in the development of this system is object oriented and design methodology. Smart Python Agent Development Environment was used in the system implementation. Results show how decentralized agent coordination enables faster and more scalable responses than monolithic polling solutions.