// Trust-Based Path Optimization · Random Forest ML · QoS-Aware · Anomaly Detection
Network Setup
5
No connections yet
Simulation Parameters
🚨 DDoS / Chaos Test
Live Trust Scores
Node
Score
Status
—
Nodes5
Edges0
Avg Trust0.500
ML ModelRForest ✓
Network Topology Visualization
⚡
Configure your network, add connections, then click Run ML Prediction
📋Events will appear here after prediction
🔐Trust comparison appears after simulation
🛡️Anomaly detection & QoS metrics appear after simulation
Developed By
👨💻 Development Team
Yug Patel
Registration: 24BYB1062
Full-Stack Developer
Adarsh Saripaka
Registration: 24BYB1093
Full-Stack Developer
👨🏫 Guided By
Dr. Swaminathan Annadurai
Project Guide
Professor, Department of Computer Science
This project demonstrates the implementation of a Machine Learning based approach for optimizing network
routing, utilizing trust models for anomaly detection and QoS prioritization in data transmission.
Help - How to Use Smart Routing System
🚀
Setup Network
Set Number of Nodes using the
slider (3-15).
Create Connections by selecting
exactly 2 nodes in the grid.
Configure Simulation Parameters
like Routing Strategy, QoS Priority, Packet Count, Source and Destination nodes.
▶ Run
Simulation
Click Run ML Prediction to
simulate packet transmission over the network.
View Results in the right panel to
see Success Rate, Path Selected, and routing metrics.
🛡️
Chaos Engineering (DDoS test)
Select a target node from the dropdown.
Click Inject Node Failure to
artificially drop its trust value to near zero.
Run another prediction to see how the ML model intelligently routes around the
compromised node.
Learn About Smart Routing
📚
Theoretical Foundation
What is the Smart Routing ML System?
A networking simulator that uses a Machine Learning (Random Forest) classifier and a custom trust-based
mathematical formula to optimize routing paths dynamically, rather than relying solely on traditional
shortest-path algorithms.
⚙️ How
it Works
1. Path Discovery: Simple paths between source and
destination are enumerated.
2. Trace Metrics: Computes simulated delay, packet loss,
bandwidth, and link load based on path length and congestion.
3. Predict Probability: Metrics undergo a Random Forest
classifier evaluation to deduce the overall success probability.
4. Cost Factorization: Paths are evaluated considering QoS
class weights combined with ML likelihoods.
Where:
• α (Alpha) = Learning rate
• M = ML Confidence Multiplier
• R = Simulated Reward Signal
• β (Beta) = High-trust dampening factor
📶
QoS Prioritization
Real-Time (VoIP/Video): Emphasizes ultra-low latency. Premium: Balances speed and high reliability. Best Effort: Classical bulk data transfer focused on throughput.
