Sociality Analysis Of Users' Behavior In Wireless Networks

With the prevalence of mobile internet, the fast development of large scale data processing technology especially cloud computing technology, the rising of social network and changing human's communication style, it is forming a worldwide online social network in internet. Traditionally social network analysis based on internet gradually change from small-scale user relationship mining to large-scale user behavior analysis in network. Now users'social relation and behavior analysis based on internet focus on interesting mining of individual user and clustering users, aimed to bring more network scale and business profit by adding directly connection of related user, supplying more intelligent and personalized service and recommendation. Accompanied by popular and upgrading of mobile smart devices, the integration of mobile network and social network, endlessly emergence of network service and application based on Location and Social, human's life and communication have been changed deeply. Compared to social relations in wired internet, social relations in wireless network show some characteristics of environment sensitivity, high dimension, deep hidden relations and high demand of real-timing of service and application. We cannot apply methods of social relation analysis in wired network to wireless environment. It is of great significance to have research on social relation analysis of wireless network, mining social characteristics of wireless users, designing behavior concerned network protocol and services.A trace analysis framework of WLAN is proposed to mining and analyzing large scale WLAN users'behavior. We apply user profile model to represent user in WLAN and extract each user profile from data set. We define SoI, an index of social relationship between two users, to depict the closeness of social relation between two users. Considering the huge difference of visiting frequency of different AP locations, we bring in weight of location to modify the social similarity model. After modeling, an unsupervised hierarchical clustering algorithm is applied to cluster all users in wireless network based on weighted similarity. Though environment parameters are different, two universities both form many social groups with Pareto distribution of similarity and exponential distribution of group sizes. These findings are very important to the research of wireless network and social networks.