Smart Shielding: Using ML and AirTag RSSI Data for Better Privacy
Type
conference paper
Date Issued
2025-07-18
Author(s)
Abstract
Bluetooth Low Energy (BLE)-based trackers have become increasingly widespread due to their affordability, energy efficiency, and integration into Crowd-Sourced Finding Networks (CFNs). CFNs, such as Apple’s Find My system, leverage vast user bases to enable device location even without direct Internet connectivity. However, the expansion of such technologies has raised concerns about security and potential misuse, particularly for unauthorized tracking. This paper investigates whether Received Signal Strength Indication (RSSI) data, when combined with Machine Learning (ML) techniques, can enhance the detection and protection mechanisms for individuals targeted by such misuse. A dataset comprising 13,353 labeled entries was collected using Apple AirTags to simulate various tracking scenarios and used to train and evaluate multiple classification models. Among these, a Decision Tree classifier demonstrated a strong balance between accuracy, F1-score, and overfitting resilience, achieving an accuracy of 85.35%. The model was subsequently integrated into HomeScout, marking a promising step toward proactive misuse mitigation in BLE-tracking ecosystems.
File(s)![Thumbnail Image]()
Name
LCN25-SmartShielding.pdf
Size
1.53 MB
Format
Adobe PDF
Checksum (MD5)
4497c58cd7e20e2b4fa9a88160ad8d83