Today it is a basic requirement of mobile users to be able to connect to the Internet anywhere and anytime in order to make phone calls, download or upload data or to use multimedia applications.
Nevertheless the users moving with a certain speed into an arbitrary direction have to face to the physical fact that they have to change their subnetworks frequently during their movement as because their access points keep changing all the time. The numerous IP address changes can cause significant package loss and transmission delay in the ongoing communication sessions. The requirements of the mobile users and business interest of the service providers are common int he sense that these access changes should be seamless: avoiding package loss and maintaining quality of service without any transitional fluctuation and degradation in running communicataions.
Many researchers propose using the global positioning services provided by the commercial GPS recievers having acceptable accuracy and available for anybody today. The actual position of the mobile node can be located within few meters accuracy using GPS receiver even in every second. The actual magnitude and direction of the movement velocity can be defined. Based on the velocity vector the future position of the MN in the prediction window as well as the expected location and timing of the handovers can be predicted with different methods. Storing the data of the past handovers and appending them to GPS co-ordinates we can provide information about the exact values of parameters in the actual network coverages, and when the mobile node takes the same route again then we will know all the required details for the seamless handover in advance.
In my thesis work I summarize the different predictive mobility management methods based on the referring international articles. Several technologies, – like the stored roaming patterns, the signal strenght and quality, and actual GPS position based prediction – are studied.
I analyse the joint effects of the positioning error of GPS and rasterization in case of uniform and Gauss positioning error distribution. I study the error of the prediction vector on paths having varying curvatures at different velocities and prediction windows. In case of a certain geometrical model I simulate the movement of a mobile user towards a Wifi network. Estimation is made on the handover probability taking into account the effect of rasterization and the prediction vector.