A Novel Multilayer Methodology for Congestion Estimation Using Sequence of Bluetooth MAC Scanners

Bluetooth Media Access Control Scanner (BMS) data are now increasingly used for travel time/speed estimation. In practice, travel time estimation for a link uses two BMS (at the upstream entrance and the downstream exit) only. The availability of other scanners is overlooked. Besides data quality, the accuracy and reliability of the estimation depend on the sample size and distance between the scanners. This research proposes a novel approach to enhance the estimation accuracy by utilising a sequence of scanners. The proposed multilayer methodology considers different travel time estimation models defined at different scales on the corridor. The single point of truth is established by fusing the estimates from these models using a negative exponential function based weighted average. The proposed methodology is validated using synthetic data, and its applicability is demonstrated through estimating congestion on two arterial corridors from Brisbane, Australia. The results show that the proposed methodology can increase the accuracy of the single layer estimation by around 4%. It also provides seamless congestion estimation even when data from a layer are missing.