Fine-grained travel time sensing in heterogeneous mobile networks

Abstract

Sensing and estimating real-time passenger travel time in urban mobile networks provides essential information for passengers to select different transportation modalities, improving their travel experience. Most existing work on travel time sensing has been focused on individual transportation modalities and its riding time based on the static time tables. However, passengers often consider different modes of transportation, e.g. taxis, subways, buses or personal vehicles, and a significant portion of the travel time is spent in the uncertain waiting or walking, which cannot be accurately obtained by static time tables. In this paper, we design a real-time data-driven framework FineTravel for fine-grained travel time sensing based on multi-source real-time data from multi-modal mobile networks including taxi, bus, subway, and private vehicle. The key challenge we address in FineTravel is to estimate implicit components (including walking, waiting and riding time) of the total travel time without direct measurement. In contrast to the existing work based on single-modal transportation networks mostly with offline data, the novelty of the FineTravel is based on its real-time multi-source data-driven (including both vehicle GPS and smartcard data) modeling for a comprehensive travel time sensing.