Concept Design of Estimation Scheme for Social Graph Based on Occupancy Time in Space

Akari Ushiyama, Yukina Miwa, Shotaro Osaki, Yuki Yamada, Katsuhiro Naito

Taking care of students is effective work to avoid repeating a year or dropping out in education institutes. As taking care process, human relations among students is a related factor to find a care-required student. Additionally, dynamic changes in human relations require continuous recognition of students to taking care. As the recognition scheme of students, indoor positioning methods are candidate technology to track students with cheap wireless devices. Bluetooth, WiFi, and IEEE 802.1.5.4 are usually used for conventional service according to the decreasing costs of the devices. Bluetooth Low Energy (BLE) is the low energy standard of Bluetooth and is suitable for long operation with a battery. Since the authors have been proposed a concept design of an estimation scheme for social graph based on occupancy time in space, this paper extends the proposed position estimation scheme. It evaluates the estimation performance of the grid position as the fundamental location information for the social graph estimation. The proposed scheme uses a BLE beacon to track each student and recognizes a student share time in the same space. It also estimates social relations among students. We have developed a prototype system for tracking BLE beacon and estimating a location according to RSS values from the BLE beacon. In the evaluation, we have developed a special classifier to estimate a grid position according to the received RSS values. The classifier is trained by actual measurement data. The evaluation results show that the classifier can estimate a correct grid position with more than 90% by receiving one set of RSS values.

Full Text