Behaviroal sensing in homes
Understanding users' behavior at home is central to behavioral, social and mental-health studies. For example:
- Social researchers are interested in how much time parents spend with their children at homes.
- Medical professionals would like to know how caregivers interact with a patient.
- Doctors want to know if a patient has signs of depression or anxiety by understanding how she/he spends time at home.
Current solutions rely on self-reporting (e.g., questionnaires and diaries). However, studies have shown that self-reporting is often inaccurate and subjective, and the large overhead makes users stop reporting in long-term studies. On the other extreme, putting cameras in homes to record everything all the time is accurate but privacy-invasive.
Marko
Marko is a tool for enabling behavioral sensing using radio signals. It transmits a low power wireless signal and analyses its reflections from the environment. To answer behavioral related questions, we create an abstraction for behavioral sensing with 3 elements: where, when & how long, and who. The system is designed with three components:
- Extract short user trajectories from radio reflections (to answer where & when)
- Tag trajectories with user identities (to answer who)
- Scale to new users and new homes automatically (for real-world deployments)
It provides a solution that is accurate, low overhead, and non-invasive.