Mobile Context Inference Using Low-Cost Sensors

Evan Welbourne, Jonathan Lester, Anthony LaMarca, Gaetano Borriello: Mobile Context Inference Using Low-Cost Sensors. LoCA 2005: 254-263.

This paper reports on the fusion of location and non-location sensors data to leverage the synergy between them to enable a wider variety of high-level mobile context inference. The system senses location with a GSM cell phone and a Wi-Fi enabled mobile device (each running Place Lab), and collects additional sensor data using a sensor board that contains an 3-axis digital accelerometers, barometric pressure, analog visible light phototransistor, digital barometer/temperature, relative humidity, 2-axis digital compass, analog electric microphone. The authors show that like previous systems, their system can classify a mode of transit and extract significant places within a user’s daily movements. However, their can do it without the use of GPS (unlike other experiments) and can classify places based on the activity that occurs there.

In Section 3, the authors mention the “experience sampling method” (ESM). After checking Context-Aware Experience Sampling Tool, it appears that in ESM:

Subjects are asked to carry a beeper device that signals on a time-based protocol determined by the researcher. Each time the beeper activates, subjects fill out a survey that typically includes questions asking what the subject was doing and how the subject was feeling at the time of the alarm. With a sufficient number of subjects and samples, a statistical model of activities can be generated. ESM is less susceptible to subject recall errors than other self-report feedback elicitation methods.

Future work in the fuse of location and activity and in mobile context inference include:

• infer the mode of transit to select an appropriate motion model for a location particle filter and improve the location accuracy
• work with place and indoor activity (e.g. mobile place such as bus and train)
• Collect users daily life data and annoted them with the ESM. It could be used to study the effects of place and activity on interruptability and prompting.

Relation to my thesis: It is an example of the fusion of location and non-location sensors to frame the activity of users. I am involved in a project using ESM and might inspire from it for my second experiment.