Traffic modelling and analysis of next generation networks.

Abstract

Wireless communication systems have demonstrated tremendous growth over the last decade, and this growth continues unabated worldwide. The networks have evolved from analogue based first generation systems to third generation systems and further. We are envisaging a Next Generation Network (NGN) that should deliver anything anywhere anytime, with full quality of service (QoS) guarantees. Delivering anything anywhere anytime is a challenge that is a focus for many researchers. Careful teletraffic design is required for this ambitious project to be realized. This research goes through the protocol choices, design factors, performance measures and the teletraffic analysis, necessary to make the project feasible. The first significant contribution of this thesis is the development of a Call Admission Control (CAC) model as a means of achieving QoS in the NGN’s. The proposed CAC model uses an expanded set of admission control parameters. The existing CAC schemes focus on one major QoS parameter for CAC; the Code Division Multiple Access (CDMA) based models focus on the signal to interference ratio (SIR) while the Asynchronous Transfer Mode (ATM) based models focus on delay. A key element of NGN’s is inter-working of many protocols and hence the need for a diverse set of admission control parameters. The developed CAC algorithm uses an expanded set of admission control parameters (SIR, delay, etc). The admission parameters can be generalized as broadly as the design engineer might require for a particular traffic class without rendering the analysis intractable. The second significant contribution of this thesis is the presentation of a complete teletraffic analytical model for an NGN. The NGN network features the following issues; firstly, NGN call admission control algorithm, with expanded admission control parameters; secondly, multiple traffic types, with their diverse demands; thirdly, the NGN protocol issues such as CDMA’s soft capacity and finally, scheduling on both the wired and wireless links. A full teletraffic analysis with all analytical challenges is presented. The analysis shows that an NGN teletraffic model with more traffic parameters performs better than a model with less traffic parameters. The third contribution of the thesis is the extension of the model to traffic arrivals that are not purely Markovian. This work presents a complete teletraffic analytical model with Batch Markovian Arrival (BMAP) traffic statistics unlike the conventional Markovian types. The Markovian traffic models are deployed for analytical simplicity at the expense of realistic traffic types. With CAC, the BMAP processes become non-homogeneous. The analysis of homogeneous BMAP process is extended to non-homogeneous processes for the teletraffic model in this thesis. This is done while incorporating all the features of the NGN network. A feasible analytical model for an NGN must combine factors from all the areas of the protocol stack. Most models only consider the physical layer issues such as SIR or the network layer issues such as packet delay. They either address call level issues or packet level issues on the network. The fourth contribution has been to incorporate the issues of the transport layer into the admission control algorithm. A complete teletraffic analysis of our network with the effects of the transport layer protocol, the Transmission Control Protocol (TCP), is performed. This is done over a wireless channel. The wireless link and the protocol are mathematically modeled, there-after, the protocols effect on network performance is thoroughly presented.