Development and Evaluation of Population Monitoring System Assuming Outdoor Events

Keita ARAI  Hiroshi YAMAMOTO  Katsuyuki YAMAZAKI  

Publication
B - Abstracts of IEICE TRANSACTIONS on Communications (Japanese Edition)   Vol.J101-B   No.2   pp.80-89
Publication Date: 2018/02/01
Online ISSN: 1881-0209
DOI: 
Type of Manuscript: Special Section PAPER (Special Section on Student Research)
Category: 
Keyword: 
population observation,  sensor network,  sensor data mining,  Doppler sensor,  MQTT,  

Full Text(in Japanese): FreePDF(2.6MB)


Summary: 
It is important to manage the number of visitors in order to improve the event quality and to keep visitors' safety in an outdoor event. On the other hand, a low-cost doppler sensor becomes available these days to detect approaching and distancing of humans. Therefore, we propose a population monitoring system utilizing the doppler sensor. An administrator of the event and events can know the number of visitors in real time. In order to measure the congestion situation, we define Population Scale which is not only the number of people but also the amount We show that Population Scale can be calculated from sensor data by approximating with a Gompertz curve the relationship between Population Scale and the Doppler sensor's measurement. We extracted three feature quantities, the most crowded time, crowded period and the maximum Population Scale, by fitting by the EM algorithm from An experimental measurement. Although the the most crowded time and crowded period did not change, the maximum Population Scale was found different from day to day. There was a negative correlation between the Population Scale and the lowest temperature.