Accepted paper: Detecting air travel to survey passengers on a worldwide scale

The paper Detecting air travel to survey passengers on a worldwide scale (pre-editing version), co-authored with Pierre Dillenbourg and Nicolas Nova, as been accepted for publication in the Journal of Location Based Services. It reports on a methodology for gathering mobility data at a world-wide scale, contrasting with traditional travel survey methods. The originality of this research work is to take into consideration the limitations of the technological settings as well as the complexity of human and technological environments as the source of the design solution.

Abstract. Market research in the transportation sector is often based on traditional surveys, such as travel diaries, which have well documented shortcomings and biases. The advent of mobile and wireless technologies enables new methods of investigation of passengers behaviour that can eventually provide original insights into mobility studies. Because these technologies can capture travellers’ experience in context and real time, they pave the road for new surveys methods. In this paper, we demonstrate that mobile phones can recognize air travel with a light algorithm that scans their connectivity to cellular networks. The originality of our method is that it does not rely on any GPS-like location information and runs on a large variety of mobile phones. It detects flights on a worldwide scale and asks travellers to report on their travel experiences as they occur, eliminating the recall bias of traditional solutions. Once the system detects a journey, it triggers a flight satisfaction questionnaire that sends answers to a centralized server. This approach respects the traveller’s privacy and proved a 97% success rate in detecting flights in a 12-months study involving 6 travellers who boarded on 76 planes.

Keywords: Sensing and activity recognition, mobility detection, transportation study

Unlike the traditional ways to capture travel information, our approach relies on the mobile phone to generate “automatic passive” GSM fingerprints and trigger an in-situ questionnaire. It is an hybrid solution of implicit motion detection with the air traveller’s consent and explicit disclosure of the travel experience. The motion detection is based on an algorithm that analyses the sequences of GSM network Location Area Identity. The figure below shows 3 examples: In flight, the mobile phone roams from one country code to another. If our software does not retrieve any LAI within 30 minutes, it detects an air travel. In a train, the mobile phone moves within different network providers and area codes. No survey appears if the disconnection periods do not exceed 30 minutes. Similar scenario takes place for a car that moves within different location areas.

Why do I blog this: Work conducted a couple of years ago as a fruit of the research developed for CatchBob! at the demand of a client of Simpliquity. Unlike many designs that consider practical constraints as detrimental to the elegance of technological solutions, we instead viewed them as opportunities to rethink solutions which eventually have to change over time.