A research team turned ordinary smartphone GPS into a data feed for Malang's least predictable transport.
Ask anyone in Malang about the angkot and the first word is usually uncertainty. Mikrolet, as the city calls them, have no fixed stops and no published operating hours. A driver may take a shortcut, turn back early, or simply stop running for the day. Passengers cannot reliably answer the simplest questions: when is one coming, and how long will the trip take?
That unpredictability is also what makes the service hard to improve. You cannot promise a schedule you have never measured. The team behind this study set out to build the measurement first, with a design goal of changing almost nothing about how the angkot actually operates.
A smartphone in the vehicle reads GPS positions and sends them over a REST web service to a server that stores the trip data. The researchers picked a phone rather than a dedicated GPS module because phones are easier to obtain and maintain, and because their accuracy is good enough for this kind of tracking. The driver keeps driving and the app stays out of the way.
To decide when the vehicle is at a stop, the team used circle based geofencing. Instead of relying on physical bus stops that do not exist, they defined pseudo-stops as circles along the route. Entering a circle registers a stop event, which is what turns a raw trail of coordinates into a usable trip record.
Two settings dominate the results: how often the app samples location, and how wide the pseudo-stop circles are. A shorter tracking interval tends to allow a smaller radius. That combination is efficient but less effective, because events slip past. A wider interval pushes the radius up, which improves effectiveness and costs efficiency.
Measured across the trials, effectiveness reached around 90 percent while efficiency sat near 30 percent, so both have room to grow. Energy use, notably, stayed low even though the app polls location services periodically. The authors report that consumption settled at a modest level after the initial start up spike.
The value of the study is that it produces the dataset that later prediction work will need. Before anyone can forecast angkot arrivals, someone has to record what the angkot actually did.
Agi Putra Kharisma, Eriq Muh. Adams Jonemaro, and Issa Arwani (2021). Paratransit Trip Data Collection System with Smartphone GPS and REST Web Service in Malang, Indonesia. Indonesian Journal of Electrical Engineering and Informatics.
