Traffic Modeling and Simulation for NGN with Markov Reward Model and Neural Networks

The paper is devoted on traffic modeling and simulation. Under consideration is traffic with compression at media gateway nodes for Next Generation Networks (NGN). The NGN uses broadband transport technologies that enable QoS management, in which service-related functions are independent from underlying transport-related technologies. The transport layer provides connection between outer NGN elements, and elements located at the NGN servers, like databases and media gateways (MGW), which present interfaces between the NGN and other networks. Modelling and simulation of traffic with compression at media gateways is developed with Markov reward model using learning vector quantification. The NGN architecture is conceived to achieve independence of applications and services from basic switching and transport technologies. The bandwidth sharing policy with partial overlapped transmission link is considered. Calls arriving to the link that belong to VBR and ABR traffic classes, are presented as independent Poisson processes and Markov processes with constant intensity or random input stream, and exponential service delay time. Service delay time is defined according to MRM. Traffic compression is calculated using neural clustering and self-organizing maps. Numerical examples and simulation results are provided.