AI-based Intrusion Detection using Current Profiling in IoT and Edge Devices

Co-organizers:

Objectvies

1. Introduce hardware-intrinsic threats and their signatures in electrical current profiles.
2. Demonstrate how to simulate cyberattacks on IoT devices using Covert-channel, Depletion, DoS, and MiM tactics.
3. Guide participants through building a current-monitoring setup using ESP32 and Raspberry Pi 5.
4. Train participants to develop, train, and deploy AI-based models (in MATLAB/Python) to detect these attacks.
5. Discuss performance trade-offs, false-positive mitigation, and future directions including federated AI security.

Key Contributions

1. Novel approach: Combines current profiling with machine learning for intrinsic cyber threat detection.
2. Real-world testbed: Employs low-cost ESP32 and Raspberry Pi devices for replicable attack scenarios.
3. Attack diversity: Evaluates four distinct types of cyberattacks under realistic conditions.
4. AI integration: Leverages ensemble learning, LDA, k-NN, and SVM for real-time classification.
5. Edge-oriented: System designed for low-latency, resource-constrained IoT environments.
6. Scalable framework: Open for adaptation to federated learning and automated response strategies.

Hands-on Activities

This workshop is designed to be highly interactive, allowing participants to gain practical experience with hardware testbeds, cyberattack simulations, and AI-based threat detection workflows. Participants will work in small groups (2-3 people) at designated workstations, each equipped with a full IoT-edge setup and pre-installed software.

Activity: Setting Up the Testbed Hardware

Objective: Familiarize participants with ESP32 and Raspberry Pi 5 devices, wiring, sensor setup, and networking.

1. Unbox and connect the following components:
2. Connect ESP32 microcontrollers with DHT11, PIR, gas, soil moisture, and ultrasonic sensors.
3. Set up the Raspberry Pi 5 with a connected RGB camera.
4. Flash pre-written Arduino sketches to the ESP32s for periodic sensor data transmission via UDP.
5. Connect the Raspberry Pi 5 to local Wi-Fi or Ethernet and verify both camera streaming and data reception.
6. Verify data logging on a central receiver laptop using a MATLAB or Python-based dashboard.

Outcome: A fully functional IoT network with five sensor nodes and one edge node transmitting data.

Registration:

Visit HONET registration page here.