A Non-Intrusive, Single-Sided Car Traffic Monitoring System Based on Low-Cost BLE Devices

Accurate and dense car traffic monitoring is a key prerequisite to cope with upcoming road infrastructure demands. We propose a vehicle detection and counting system that is based on low-cost BLE sensor devices. The variance within the Received Signal Strength inside a wireless network is fed into an adaptive peak detection algorithm in order to recognize vehicle-induced events within the stream of field strength data. Observing the traffic on a two-lane local road, a counting accuracy of 97.9 % is achieved while the false-positive rate is lower than 4.5 %. Monitoring a single-lane only, the performance can even be increased to 99.6 % and the false-positive rate is reduced to 0.8 %. Furthermore, the presented system needs sensor hardware only on one side of the observed street even when monitoring the traffic on both lanes.