Mobility prediction based on collective movement behaviors in public WLANs

Predicting the location of a mobile user in public WLANs such as city or campus WLANs has received increasing interest over the past decade. Public WLAN enable mobile users use their portable devices to access Internet applications from where they want even outside their normal work environment and still remain connected to the Internet whilst on the move. These portable devices such as IP phones, smart-phones, iPads, iPods and tablets are light enough to walk-and-talk, so these users often leave their devices on most of the time for accessing Internet applications, for example Voice over IP (VoIP) or realtime multimedia data transmission, e.g., streaming audio and video. Due to public WLANs support such a large number of portable devices and their dynamic relocation, the matters of location management and network resource allocation become more and more serious. Mobility prediction which may accommodate the network with future location information of all mobile users has played a crucial role in the accurate estimation of network resource demands at future time. Therefore, several research works have focused on mobility prediction toward more efficient network resource management in public WLANs. This paper survey various mobility prediction approaches and find out that most of them have some deficiencies. Motivated by the limitations of the previous approaches, this paper introduces a novel mobility prediction approach which utilizes mobility rules discovered from multiple similar users. Due to combination of clustering and sequential pattern mining, two proposed models are not only able to deal with the lack of information on personal profile but also able to avoid noise of random movements in mobile users' profiles. We have conducted experiments to investigate the efficiency of the proposed approach as well as to study how the order of combination effects on the prediction accuracy.