Knowledge-based mobility support in wireless network

Abstract

This dissertation proposes a mobility support mechanism that simultaneously reduces handover blocking probability and maximizes the utilization of resources. The proposed mobility support mechanism admits new connections over considerations of both unused and reserved resources, and reserves resources for handover to one or more of the cells the mobile station follows. The proposed mobility support mechanism is based on the movement prediction of each mobile station. The handover blocking generally occurs when a new target cell has no available resource to support the handover. A simple method of providing resources for a handover is to reserve some resources at neighboring cells. However, incorrect resource reservation wastes resources, so it is very important to predict when and where the mobile station will move. In a specific mobile network environment, especially in an indoor environment, the mobility patterns of users reflect the routines of their lives: most mobile users have favorite routes and habitual movement patterns. The movement pattern of a mobile station may generally be composed of a connection initiating cell, one or more passageways, and a destination cell during connection time. Some information, such as sojourn time and the number of connections for each cell, can be kept in a mobile station as historical data. A cell has very large probability to be a passageway if it has many handover connections and little sojourn time during connection. In a contrary concept, if a cell has much sojourn time, the probability that the cell is in one of the places of residence is large. Consequently, if a passageway is predicted to be the next cell by handover, it is not suitable that resource reservation be made only at the very next cell. A proposed mobility support mechanism is a target-cell-based resource reservation mechanism. Resource reservation is made by using the above mentioned mechanism at CAC (Connection Admission Control) time. When a m...