Authors: Ruoyu Li, Qing Li, Tao Lin, Qingsong Zou, Dan Zhao, Yucheng Huang, Gareth Tyson, Guorui Xie, Yong Jiang

Published on: April 19, 2024

Impact Score: 7.4

Arxiv code: Arxiv:2404.12738

Summary

• What is new: DeviceRadar introduces a novel approach to IoT device fingerprinting in ISP networks, utilizing ‘key packets’ and spatial relationships for real-time, high-speed identification without the need for detailed data obscured by middleboxes.
• Why this is important: Existing IoT device fingerprinting methods struggle in ISP environments due to data obscuration by middleboxes and the challenge of processing high-speed traffic.
• What the research proposes: DeviceRadar employs a new fingerprinting method based on ‘key packets’ and a packet size embedding model for real-time, accurate device identification in ISP networks.
• Results: Achieves state-of-the-art accuracy across 77 IoT devices at 40 Gbps throughput with just 1.3% of the processing time compared to GPU-accelerated methods.

Technical Details

Technological frameworks used: DeviceRadar

Models used: Packet size embedding model, neighboring key packet distribution feature vector for machine learning

Data used: Packet sizes and directions, real-time traffic data from ISP networks

Potential Impact

ISP services, IoT security solutions, network equipment providers, and real-time data analytics markets

Want to implement this idea in a business?

We have generated a startup concept here: SecureIoT.