A Realistic Mobility Model and Its Application to a Reservation-based Call Admission Scheme for Ds-cdma Cellular Systems

iii ACKNOWLEDGEMENTS I would like to thank Cem Ersoy for his guidance throughout my PhD study, M. Ufuk Ça˘ glayan for being the model of an academician, Hakan Deliç for his comments on DS-CDMA systems, Abbas Yongaço ˘ glu for his short course on CDMA systems, Emre Çelebi for his valuable discussions and support in Linux, and Ian Anderson for the high spirits. I would also like to thank my wife Gülçin for her patience and help, my daughter Umay Ece for the joy, and my parents for all I have had in my life. Call blocking and call dropping are the reasons for outage in mobile wireless systems. Guard channels can be allocated for handoff calls in each cell to give higher precedence to handoff calls over new call attempts. Since new call attempts may be rejected although there are free guard channels, the decrease in the call dropping rate is achieved at the cost of increased blocking rate. Therefore, the number of guard channels is an important metric that effects system performance. In this thesis, we propose a call admission scheme that adjusts the number of guard channels dynamically. A reservation area is constructed for each active subscriber according to his speed, direction and recent mobility pattern. A reservation request, associated with a likelihood value, is sent to each candidate cell intersected by the reservation area. The number of channels to be reserved in each cell is obtained from the aggregation of the likelihood values in the received reservation requests. We have evaluated the proposed scheme against the classical scheme with fixed number of guard channels, and shown that the proposed scheme performs better in the sense that call dropping rate is reduced with lower cost. We also propose a realistic mobility model that captures human behaviors from real life such as moving-in-groups, conscious traveling, inertial behavior, and the non-pass-through feature of the physical structures in the terrain. The mobility patterns of the subscribers are determined according to a given real map composed of various types of physical structures. We have evaluated the proposed mobility model against the way point model, and shown that the choice of the mobility model results in a significant difference in system performance.