A Wireless-Based Approach for Transit Analytics

We propose Trellis --- an in-vehicle WiFi-based tracking system that passively observes mobile devices and provides various analytics for transit operators. Our infrastructure is fairly low-cost and can be a complementary, yet efficient, mechanism by which such operators collect various information, e.g., popular original-destination stations of passengers, waiting times of passengers at stations, occupancy of vehicles, and more. A key challenge in our system is to efficiently determine which device is actually inside (or outside) of a transit vehicle, which we are able to address through contextual information. While our current system cannot provide accurate actual numbers of passengers, we expect the relative numbers and general trends to be still fairly useful from an analytics perspective. We have deployed a preliminary version of Trellis on two city buses in Madison, WI, and report on some general observations on transit efficiency over a period of four months.