Leveraging platform-integrated sensors for people-centric sensing

Huy Phuong Tran

People-centric sensing, which focuses on sensing information about people, plays an important role in improving human life across different application domains such as healthcare, social interactions, and smart buildings. Current sensing platforms have two main limitations in supporting people-centric sensing, particularly in indoor environment because they often reply on wired and standalone wireless sensor platforms. The first limitation is the limited placement of these platforms, thus limiting overall sensing coverage. The second limitation is that these platforms are mostly static. This makes it difficult to monitor people, which are often mobile. The proliferation of personal computing devices (e.g., smart phones and laptops) creates great opportunities for people-centric sensing. These devices often have close proximity to users for long periods of time. Therefore, there are great opportunities for each of these devices to be used as a sensing platform. Our vision is that personal computing devices will be sensor-ready and can provide seamless sensing services across users, devices, and domains. The thesis statement of this work is that sensors integrated in personal computing devices can be leveraged to enhance people-centric sensing, particularly in indoor environment. There are several challenges in leveraging platform-integrated sensors (PISs) for peoplecentric sensing. First, accurate and sufficient information about users and their environment around them must be inferred from PISs data. Since, PISs are deployed at their hosts, PISs readings are affected by conditions in which the device is operating. Second, PISs must not affect the operation and user experience of the hosts. The devices are operated to handle tasks given by a user, not only for sensing. In our work, we investigate sensing quality and sensing efficiency of PISs. We found that user ambient conditions can be estimated from PISs data with high accuracy. Also, PISs and a sensing service have negligible impacts on a host device. In addition, we found that clothing type can affect the sensing quality of PISs at a smart phone, when the phone is in a user pocket. In our work, we show that it is possible to identify clothing types to improve the sensing quality of PISs.

Leveraging platform-integrated sensors for people-centric sensing

Files and links (1)

Abstract

Metrics

Details