Project Overview
What is IntelliHome SDN?
IntelliHome SDN is a smart-home network framework built on Software Defined Networking (SDN), combining manufacturer-specific access control policies, real-time machine learning-based threat detection, and flow trust ranking. The system dynamically adapts to changing traffic behavior, ensuring both performance and security in residential IoT environments.
Motivation
As smart homes become more device-rich, traditional networking approaches fall short in enforcing access boundaries or detecting malicious behavior. Common routers lack visibility into device intent, often exposing home networks to IoT-based threats like botnets, data leaks, or unauthorized lateral access. Traditional firewalls lack device-awareness and adaptability. By combining SDN’s centralised control with ML and MUD policies, this project aims to offer a proactive, adaptive, and transparent security mechanism tailored for IoT environments.
Key Features
- SDN-based Central Control: Centralised policy enforcement and flow monitoring through Ryu.
- MUD Policy Enforcement: Automatic behaviour profiling and policy generation for IoT devices.
- ML-based Anomaly Detection: Real-time classification of flows as benign or malicious.
- Trust Scoring with PageRank: Continuous recalculation of flow trustworthiness.
- Live Monitoring & Reporting: Visual and statistical reports for network events.
Scope of Work
- Design network architecture using Mininet to simulate IoT environments.
- Develop Ryu controller modules for flow collection, policy enforcement, and anomaly response.
- Integrate ML model for flow classification.
- Implement PageRank algorithm for dynamic trust scoring.
- Build web-based front-end for visualising network health, blocked flows, and trust scores.
- Conduct performance evaluation with varied IoT device scenarios.
Relation to TELE4642 Concepts
The project draws heavily on Network Performance topics such as:
- Queuing Models & Traffic Analysis: Using arrival/service rate concepts from M/M/1 models.
- QoS Principles: Traffic classification, policing, and scheduling.
- Markov Chains: Underpinning PageRank’s random walk model.
- Measurement & Evaluation: Empirical performance testing for latency, throughput, and detection accuracy.
Expected Outcomes
- A functional prototype of an adaptive IoT network firewall and traffic monitor.
- Demonstrated improvement in malicious flow detection accuracy over static policies.
- Clear visualisation of trust scores and blocked flows.
- Comprehensive performance evaluation against multiple traffic scenarios.
Tools & Technologies
- Ryu SDN Controller
- Mininet Network Emulator
- Python (for controller logic and ML model integration)
- Scikit-learn / RandomForestClassfier (for ML inference)
- Flask / HTML / CSS / JavaScript (for web-based interface)