McMatcher: A symbolic representation for matching random BLE MAC addresses

Abstract

Bluetooth Low Energy (BLE) is a widely used wireless technology which offers a wide range of applications. However, the introduction of MAC address randomization to preserve the users’ privacy makes it more challenging to leverage all its potential. In this paper, we present McMatcher, a privacy-preserving, novel methodology for matching random MAC addresses from BLE devices. McMatcher uses a symbolic representation of the RSSI time series to build characterizing vectors, embedding both the temporal as well as the signal strength (RSSI) properties of the BLE signal. Our methodology achieves 100% accuracy in matching 92 MAC addresses from 16 smartphones, in a dataset containing 332 MAC addresses in total. As opposed to previous works, our methodology does not require any model training, and relies only on the RSSI measurements. The computational simplicity of McMatcher allows matching MAC addresses in realtime, taking only 230ms for a set of 18 MAC addresses.