Predictive mobility management for wireless mobile networks

This work studies user movement behavior in the context of a wireless mobile network and investigates Predictive Mobility Management as a design philosophy supporting seamless mobile communication services. Predictive Mobility Management is shown to facilitate efficient mobility prediction that allows the network to pro-actively anticipate a moving user's service requests by service pre-connection, resource pre-arrangement and system pre-configuration. The value of this work is demonstrated through a systematic exploration of user movement behavior at different system views for the establishment of a comprehensive mobility model; through the theoretical development of low-complexity movement prediction algorithms that abstract mobility information from practically available observation data; and through intelligent applications of mobility prediction for system performance enhancement and add-on soft capacity in connection maintenance, location tracking, front-end system design, bandwidth allocation and call admission control. A variety of different techniques are employed in this work for obtaining the deployment of Predictive Mobility Management. Observed from a “micro” point of view, a mobile user is modeled as a linear dynamic system driven by time-varying forcing functions that simulate subjective moving intention from the user and objective random perturbations from the environment. Viewed at the network lever, a user's pseudo-random inter-cell trajectory is modeled as the result of a pattern-editing process that captures the regularities and temporary random tendencies present in the mobile user's daily movement behavior. An approach to user mobility prediction is derived through the modification of existing techniques in statistical signal processing and finite string comparison literature, which include extended, self-learning Kalman filtering technique for intra-cell mobility tracking, and approximated pattern matching technique for inter-cell movement pattern detection. Based on these, new algorithms and systems are developed to enable Predictive Mobility Management for various network issues. Two strategies are proposed for predictive connection and location management in wireless ATM systems to support multimedia applications with QoS guarantees. A mobility-adaptive infrared front-end system is developed to improve power efficiency and infrared link quality in line-of-sight environment. A predictive guard bandwidth scheme is proposed for efficient radio bandwidth allocation and call admission control. Detailed mathematical formulation and dynamic programming approaches are given for algorithm derivation. Analytical and numerical methods are employed to quantify the performance of the proposed systems and schemes in terms of well-defined measures such as probability of correct prediction, signal power degradation, handoff call blocking probability and Grade of Service (GOS). It is shown that the techniques we have applied and developed for this research can tackle situations that conventional mobility management strategies fail to handle, thus open a new avenue in mobility information acquisition, processing, and applications for efficient mobility management in wireless mobile networks.