Correlation Analysis and Clustering of Wi-Fi Users in Campus

Nowadays, Internet technologies and location based servies have been developing rapidly.Meanwhile,more and more universities begin to use wireless network for teaching and management.For an increasing number of unversity Wi-Fi data of location information,analysing users' trajectory features and doing clustering research is of much importance.Through research on Wi-Fi users in unversity,then corrlelation infomation and trajectory similarity can be figured out,also Wi-Fi users can be clustered into different groups.In this sense,underestanding the behavior of Wi-Fi users in a unversity has great significance to the manager of unversity administrations and relevent secure department. This paper explores the patterns and clustering results of a unversity Wi-Fi data of location.With the help of Wi-Fi sensory nodes distributed in different places of the unversity,we have collected over 19 million users' location records of more than 100 places in a unversity.We first did some research in order to find out unversity Wi-Fi users' group pattern through our dataset.Secondly,in this paper,we use an correlation algorithm based on frequent tree to implement correlation of specific user's mac.Based on the result of correlation algorithm,we use LCSS algotithm to find out the relation between a specific mac's associated macs and their trajectory similarity.At last,on the basis of the personnel distribution frequency matrix,we use a bottom-up hierarchical clustering algorithm based on cosine distance to implement clustering work,and the final clustering result is satisfactory.