Skip to content

This repository contains the ML Classifiers that I had trained for my project on IoT Device Security. I used multiple open-source datasets and performed Network Traffic Analysis to create efficient classifiers.

Notifications You must be signed in to change notification settings

khushbhuta/IoT-Device-Fingerprinting-and-Classification

Repository files navigation

IoT-Device-Fingerprinting-and-Classification

This project focuses on building a robust classification system for IoT Devices. I have created device fingerprints for IoT Devices using network traffic analysis to classify them and detect anomalies. The goal is to create an efficient way to classify anomalies through new techniques that involve hardware-level features too. I have mainly worked on frequency-domain analysis(applying FFT to network-level features) and multimodal feature integration to detect anomalies and classify non-attack devices. The CNN method did not yield efficient results and was hence scrapped.

Testing:

For validating this method, I have tested my models on multiple open-source IoT-Network-Traffic datasets. The results can be seen as a section in my comprehensive report.

Datasets used:

  1. UNSW NB-15 (Anomaly Detection Dataset)
  2. UNSW Traffic Traces Dataset

About

This repository contains the ML Classifiers that I had trained for my project on IoT Device Security. I used multiple open-source datasets and performed Network Traffic Analysis to create efficient classifiers.

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published