IIT Delhi has unveiled AndroCon, a system capable of inferring user activities and indoor layouts through GPS data analysis, highlighting significant privacy implications. Developed by MTech student Soham Nag and Prof. Smruti R. Sarangi, AndroCon can determine user surroundings and activities with high accuracy.
GPS Data Analysis Breakthrough
AndroCon leverages variations in GPS signals, such as changes in signal strength and frequency shifts, to deduce a user's environment. Nine low-level GPS parameters, including Doppler shift and signal power, allow the system to identify whether a user is indoors or outdoors, or engaged in specific activities like traveling or standing.
The study, conducted over a year and covering 40,000 sq km, demonstrated up to 99% accuracy in identifying surroundings and over 87% accuracy in recognizing human activities. This accuracy is achieved without reliance on the phone's camera or microphone.
Privately Mapping Indoor Spaces
Furthermore, AndroCon can generate precise indoor maps, identifying rooms, staircases, and elevators with an error margin of less than four meters. The integration of classical signal processing and machine learning transforms noisy GPS data into detailed, actionable insights that were previously unachievable.
Privacy Concerns and Implications
The ability of AndroCon to extract such extensive information through fine location permissions in Android apps poses substantial privacy risks for users. Any app with these permissions could potentially gather sensitive details without explicit user consent, a concern highlighted by the researchers. This study is published in the ACM Transactions on Sensor Networks.
