Research Project

Context-Aware Behavioral Fingerprinting for IoT Devices

Description

  • Collected & analyzed 600K+ packets (Wireshark); engineered 18 packet-level features (e.g., TCP window size, payload entropy).
  • Achieved ~94% accuracy via Random Forest for device identification in variable contexts.
  • Deployed an IoT testbed (Raspberry Pi gateway + 8 NodeMCU nodes), lowering per-device cost by ~90% vs prior setups.

Stack: Python, C, Raspberry Pi, NodeMCU, scikit-learn, Wireshark