Abstract
In this paper, statistical models are proposed to estimate trip level origin-destination (OD) matrices for public transport based on Wi-Fi data traffic. Wi-Fi monitoring equipment installed in 32 buses in Stavanger, Norway, collected Wi-Fi data during several months. The median received signal level of frames transmitted by a device and the time interval between the first and last frame are modelled as statistical distributions, conditional on whether the Wi-Fi device is on the bus or not. Based on these models and using passenger load data from Automatic Passenger Counting (APC) systems installed in the buses, the probability for each detected device being on or off the bus is estimated. When tested on large data sets, the proposed statistical method is more accurate than when hard thresholds for median received signal level and time interval of observation are applied.