In this paper we propose an analytical model for jitter, wherein we implement the interrupted Poisson process (IPP) for incoming traffic. First, we obtain an analytical model for the jitter on one node with respect to the phase probabilities, traffic load, and tagged traffic share in the aggregate traffic flow. Then, we analyze N-node cases, and propose a model for end-to-end jitter. Our analysis leads to some fast-to-compute approximations that can be used for future network design or admission control. Finally, we validate our analytical results by comparing them with previous results for limit cases, as well as with event-driven simulations. We propose the use of our results as guidelines for jitter evaluation of real IP traffic.
In this paper we analyze jitter behavior for Interrupted Poisson process (IPP) as incoming traffic. We propose model for jitter on one node, as well as for end-to-end jitter with IPP traffic on network nodes. Our model describes jitter behavior with respect to phase probabilities, traffic load, and tagged traffic share in aggregate traffic flow. We conclude that jitter for IPP is heavily dependent on phase probabilities, and larger than jitter for Poisson process, as one limit case of IPP. End-to-end jitter always takes values between first node jitter value and sum of individual jitter values on each node. We offer more precise evaluation of these values, which also leads to some fast-to-compute approximations for limit cases of phase probabilities and traffic load. We propose our results as guidelines for jitter evaluation for real IP traffic.
In this paper we analyze delay and jitter in networks with traffic modelled as Interrupted Poisson process (IPP), and relatively small traffic loads. In one-node analysis, we estimate the probability for delay bounds violation, and we obtain expression for the jitter, with respect to phase probabilities, traffic load, and tagged traffic share in aggregate traffic flow. We also analyze delay and jitter for simple tandem network, where we propose a model for endto- end jitter. Our research shows how end-to end jitter is conditioned with phase probabilities of the incoming process, and we determine the nature of that correlation. Finally, we estimate the probability for end-to-end delay bounds violation, with respect to the jitter occurring on network nodes. Our propositions lead to some fast-to-compute approximations for the limit cases, which we propose as useful in evaluation of QoS constraints for real IP traffic
When using simulation in analysis networks with number of sources, one of problems is creating traffic sources matching real network data. Credibility of conclusions acquired by simulation tools are in great deal dependent of fidelity of simulation traffic source generators. This paper presents one simple approach for real traffic source reconstruction on the basis on limited real traffic data. Source model achieved here is almost in every aspect same as measured real traffic on large IP network; necessary steps and practical applications are shown here as well.
In network development problems can be encountered considering plans and projects that will provide satisfying QoS for different circumstances, different types of users, and different traffic flow profiles. It is necessary to develop satisfying mathematical model which will provide connections between QoS and network parameters. In this paper we will look into traffic with variable intensity which can effectively be described using Markov Modular Poisson Processes (MMPP). We will offer new analytical model, and using graphs we will show losses function depending on network parameters.
Ova stranica koristi kolačiće da bi vam pružila najbolje iskustvo
Saznaj više