A novel collaborative approach for location prediction in mobile networks

摘要

Human location prediction has been a matter of concern for several years due to its many applications. It has become more important nowadays because of prevalence of mobile devices which have adequate tools for inferring location. Different approaches for making this prediction could be divided into three categories, based on the movement history they use. These include history of mobile user himself, history of all mobile users in a place, and history of only related mobile users. Besides the problem of limiting shared data to only required data, preserving privacy is the matter of concern for persuading mobile users to share their data. In this paper we have proposed a new method in which the amount of the shared data is decreased to a minimum, and only the data which will improve the partner's prediction will be shared. Our method preserves privacy by blurring the shared data up to different degrees. The experimental results show that regardless of amount of blurring, as long as the user movement is not lost because of blurring, the accuracy of prediction will be improved about 7 %.