Blocking probabilities in multiservice networks with unicast and multicast connections

Abstract

This paper is concerned with blocking probabilities in loss networks where both multicast and unicast connections are present. Each connection specifies a route and a bandwidth requirement. Multicast connections are point-tomultipont and allow transmission from one source to several receivers, any of which can join the connection at any time. Users that subscribe to the same data stream share one multicast connection so that no additional bandwidth is required. We present a Markov process model of a single network link. We also develop a convolution algorithm that finds the blocking probabilities on a single link and a reduced load approximation for estimating end-to-end blocking probabilities in a network. Keywords-Blocking probability, loss network, multicast, reduced load approximation I.INTRODUCTION Multicast is an efficient communication paradigm for transmitting the same data to multiple receivers. Instead of sending an identical data stream, or message, to each receiver, the source sends only a single stream, regardless of how many receivers have requested it. This stream is then replicated by multicast-enabled network equipment, but only to branches of the tree where receivers that requested the stream are located. Allowing bandwidth savings, this technique is useful for distributing software updates, live audio or video streams or for applications such as nearreal-time stock tickers.Since multicast has become an integral part of many networks (e.g., ATM, IP) during the last few years, the problem of evaluating performance of multicast networks and networks with both multicast and unicast traffic is certainly worth attention. Chan and Geraniotis [1] have considered a multicast network with several sources that stream layered video, they have given explicit expressions for end-to-end blocking probabilities and have developed a reduced load approximation to estimate those probabilities. Rykov and Samouylov [2] have proposed a model of a multicast network as a reversible Markov process. Gaidamaka and Samouylov in [3] have developed a Markov process model of a single-link multicast network and a computational algorithm that finds blocking probabilities. An exact algorithm that computes blocking probabilities in a single-source multicast network (tree topologies) was proposed by Virtamo et al. in [4]. They have extended their model of multicast network to include unicast background traffic (see also [5]). Boussetta and Belyot [6] have generalized the Kaufman-Roberts recursion [7], [8] to handle a single link network with unicast and multicast connections.

Approximating Performance Measures of a Triple Play Loss Network Model

“…Figure 1 shows behavior of the approximated measures in dependence on the utilization factor. Similarly to numerical experiment [13], blocking probabilities fluctuate periodically. After a qualitative analysis of the graphs in Fig.…”

Decreasing the Call Blocking Probability of Broadband IPTV Services in Stationary and Peak-hour Scenarios

“…An exact algorithm of calculating the CBP on a single link was developed by J. Karvo et al [2], which was followed by its extensions to calculate the E2E CBP for the multi-layer multicast traffic [3]. Samouylov and Yarkina [4] further considered the CBP evaluation in multiservice networks with both multicast and unicast connections. Nevertheless, none of the aforementioned models is dedicated to the domain of IPTV service.…”

“…With respect to CBP evaluation, earlier publications [2][3][4][5] have already provided analytical methods to evaluate the CBP for a single link, as well as for an entire network with tree topology. To the same end, our previous work [6] [7] also developed link state-vector-based simulation models..…”