Researchers have developed a method by which the constant noise of WiFi attempting to connect to devices could provide an accurate picture of how well public transport systems are performing, despite such systems generally operating closed and proprietary monitoring systems, with only select information dispensed beyond the confines of the company.
The paper Real-time public transport service-level monitoring using passive WiFi: a spectral clustering approach for train timetable estimation – a collaboration between Baoyang Sang of France’s École Polytechnique and Singapore-based IBM researcher Laura Wynter – establishes a system that uses WiFi’s default connection attempts to determine the arrival and departure of trains, and which could be applied to other transportation methods.
The key to establishing activity that indicates the sudden arrival (or departure) of a mass of passengers is that their numerous mobile devices are constantly sending probe requests to available WiFi networks, and these requests do not abate even in the presence of a successful log-on attempt.
Since these probe requests are so public, no closed systems are required to log and monitor them, and open sniffer tools such as Wireshark or tcpdump can provide accurate and live measurement of pre-connection activity.
The researchers propose WiFi over a number of other recently-explored possibilities to gauge ‘crowd digital miasma’, since the signal is robust and strong, noise is minimal, and the protocol is not subject to the variations in signal strength which can occur across mobile platforms such as iOs and Android in users’ individual devices, and is not affected by varying power management schemes either – a factor in other experimental schemes of a similar nature.
The technique developed by the researchers utilises spectral clustering – traces of probe requests from passengers – to estimate the true fidelity of the timetable of trains.
They note that only train operators themselves have access to location data, and that this information flows through closed and proprietary systems which are not useful for real-time analysis.
The technique can also be used as a tool for incident detection – to identify increased dwell times on platforms which may indicate network blockages or other problems.