Lightweight IoT Device Fingerprinting Approach using Locality-Sensitive Hashing

The resource constraints of Internet of Things (IoT) devices limit the use of heavy security mechanisms, which leaves them more exposed to cyberattacks. Detecting malicious devices from network traffic is therefore critical. This paper proposes a lightweight fingerprinting method based on locality-sensitive hashing (LSH) that builds device-specific signatures from combined packet headers rather than payloads, avoiding payload variability and the overhead of full packet processing. Experiments on the LSIF and IoTSentinel datasets show that combined headers reach up to 99.3% and 95.8% accuracy, respectively, outperforming payload-based methods, while reducing processing time, making it suitable for real-time processing.