Efficiency of Public Wireless Sensors Applied to Spatial Crowd Monitoring in Buildings

Abstract

Contemporary world rest on wireless connections via networks developed rapidly. Majority of people using wireless networks i.e. WiFi at: work, home, and also when they are outside. A central issue in wireless networks is efficiency and stable connection to Internet. Additional challenge is to create and maintain reliable WiFi network with full open access, videlicet every user could connect with that network if only have a range. Exemplary aforesaid network is university network where users are very specific group. An innovative approach is to treat and use Access Points (APs) as wireless sensors network to crowd monitoring in buildings where such network is located. Thanks to this we receive accurate information about the number of users (people) without interfering with their sensitive data. The aim of this research is to use geostatistical methods to prepare spatial models and predict behaviour of crowd connected with wireless sensors on full of range considered area. The investigation has tended to focus on efficiency of wireless sensors belong to open WiFi network named PWR-WiFi located in building at the main campus of Wroc_law University of Science and Technology (WUST) in Poland. The data gathered for analysis are acquired from three consecutive years 2014–2016 to better show the dynamic growth of number of PWR-WiFi network users it means number of people crowded in building. Parameter, which most reflecting behaviour of PWR-WiFi network is the number of users, obtained from APs, that was investigated during research. Preliminary and structural analysis with approximation of variogram models were made and as the next step spatial prediction models of wireless sensors network efficiency were performed by Turning Bands geostatistical simulation method. Three models of spatial prediction were prepared for three subsequent years 2014, 2015 and 2016. Following, the results were compared with spatial prediction models created previously by ordinary kriging estimation method.