Constructing Origin-Destination Matrix using Wi-Fi and AFC Data

Abstract

Transportation systems planning and management rely heavily on Origin-Destination (OD) demand matrices. However, the traditional approach of creating these matrices using household travel survey data is not only time-consuming but also expensive, making it challenging to apply them to detailed and time-sensitive analyses. Automated fare collection (AFC) systems can provide a solution to these challenges. However, many transit fare systems are entry-only with no requirement to tap at the exit station, which makes it challenging to determine the location of alighting stations and analyze spatial demand patterns. This study proposes a framework that uses Wi-Fi traces and passenger counts at AFC entry/exit gates to construct OD matrices for entry-only Urban Rail Transit (URT) systems. The City of Toronto's subway system was used as a case study, and the framework was compared to 2016 Transportation Tomorrow Survey (TTS), which is the primary source of OD-matrix estimation in the city. The generated OD-matrices were found to be very close to the OD-matrix from the household travel survey, with a cosine similarity close to one for most subway regions. Our estimated OD-matrices offer several advantages over traditional methods. They have low matrix sparsity and fast computational time and convergence, and they exhibit strong capability of recognizing demand patterns at the station-level.