Admission control in mobile cellular networks: design, performance evaluation and analysis
- Autores: Natalia Vassileva Vesselinova
- Directores de la Tesis: Yevgeni Koucheryavi (dir. tes.), Francisco Barceló Arroyo (dir. tes.)
- Lectura: En la Universitat Politècnica de Catalunya (UPC) ( España ) en 2012
- Idioma: inglés
- Tribunal Calificador de la Tesis: Boris Bellalta i Jiménez (presid.), Maria Carmen Domingo Aladrén (secret.), Jarmo Harju (voc.)
- Materias:
- Enlaces
- Tesis en acceso abierto en: TDX
- Resumen
Key to the success of the mobile telecommunication systems is provision of mobility and more specifically, continuity of communication sessions on the move. However, from engineering point of view the cobination of scarce radio resource capacity, radio channel randomness, cellular structure, and user mobility can lead to call interruptions. The latter are regarded as highly undesirable by mobile users. One essential part of the solution to this problem is to address resource insufficiency. To this end admission control algorithms are incorporated into the system. In this thesis we concentrate on admission control as a means of guaranteeing uninterrupted service to users with active calls on the move.The research work reported in the monograph can be briefly summarised as follows. We explored the extensive empirical, analytical, and simulation results concerning teletraffic random variables of mobile cellular systems from the perspective of admission control. We proposed a conceptually different from prior work admission control solution based on the scientific evidence about the statistical nature of system variables and on the main result of renewal theory namely, decision making based on estimated system behaviour. In particular, we proposed a new admission control metric that uses statistical estimates. We evaluated the performance of the devised admission control strategy, which we named MRT (Mean Remaining Time) after the admission control condition, using both analytical and simulation approaches. We mathematically modelled system performance for traditional exponential conditions through a Markov chain. To study the MRT performance for non-conventional teletraffic scenarios we developed a simulation pure performance model. We examined the MRT for conditions that matched measured data from real, live mobile cellular networks. The results show that the scheme can guarantee call continuity and that it achieves a continuous working interval in contrast to the discrete one of the common cut-off scheme, yet the MRT strategy meets the important practical requirement for simplicity. We proposed an approximation to the MRT strategy for the case when not all of the required statistical information is readily available. Next, we studied in-depth the implications of novel techniques introduced in advanced mobile telecommunication systems. In particular, we examined the effect of the adaptive modulation and coding (AMC) technique on system performance and consequently on admission control design in mobile WiMAX. The dynamic tuning to time-varying radio link conditions introduces new random variables that drastically change the traditional mobile cellular system model. In particular, cell capacity and call resource demands are not constant but random, determined by wireless link quality. We analytically modelled the radio channel randomness and the consequent non-deterministic resource demand for a streaming service with constant bit rate and strict delay requirements through a zone-based cell model. Furthermore, we examined system-level fairness, which metric had not been explored in previous studies on mobile WiMAX. Additionally, we studied the effect of AMC on system performance under the two basic admission control approaches proposed in the literature by incorporating them in the analytical model. The results show that the total new call blocking probability and forced call termination probability of a constant bit rate calls deteriorate when the radio channel conditions are quickly varying and the offered load is moderate to heavy. Furthermore, the results show important differences in blocking and dropping probabilities of calls belonging to the same service (voice) and call (either new or handoff) class but being served in different modulation and coding zones. The results also indicate that if the admission control is not adapted to the actual environment (non-deterministic) the system performance is considerably worsened.
