A Multi-Pass Algorithm for Adjusting a Network Topology in Multipoint Communications

A Multi-Pass Algorithm for Adjusting a Network Topology in Multipoint Communications

Boris Peltsverger (Georgia Southwestern State University, USA), Svetlana Peltsverger (Southern Polytechnic State University, USA) and Michael Bartolacci (Penn State University - Berks, USA)
DOI: 10.4018/978-1-4666-0050-8.ch007
OnDemand PDF Download:


Multimedia traffic on the Internet has grown dramatically in the past few years. Web sites, such as YouTube and Hulu, offer entertainment and educational multimedia content that previously was only available through broadcast or cable television and on storage media, such as CD-ROMs and videotapes. Latency is a key issue in the delivery of online content, especially with respect to multicasting. The authors’ proposed approach seeks to reduce overall latency for multicast streams.
Chapter Preview

Current Approaches

Most conferencing tools in use utilize an Application Layer multicast where packets are replicated at the end hosts by sending identical packets over the same link. A Network Layer multicast is more efficient where it starts with a single packet from the source and then routers duplicate the packets only where necessary. In multiple unicasting, packets are generated by the source, thereby creating unnecessary traffic. These two approaches are illustrated by Figure 1.

Figure 1.

Multicast vs. multiple unicast messages

It is obvious that Network Layer multicasting is more efficient, it reduces traffic by simultaneously delivering a single stream to multiple users. To receive a particular multicast stream, hosts must join a multicast “group” by sending an Internet Group Management Protocol (IGMP) message to their local multicast router that will use Protocol-Independent Multicast (PIM) to build routing tree. Network infrastructures should be configured to route multicast packets.

Real-time Transport Protocol (RTP) that is commonly used to deliver audio and video over the Internet is an application layer protocol that assigns each media stream a separate unique RTP session ID, with its own RTP Control Protocol (RTCP) packets to report the quality. So it provides multicasting support, but again on application layer. That makes very important the cost of sending message over the “delivery tree.” RTCP is based on the periodic transmission of control packets to all participants in the session, using the same distribution mechanism as the data packets. It is to provide feedback on the quality of the data distribution.

Delay or latency is the amount of time it takes for a packet to reach its destination. There are four major types of delays:

  • Processing delay is the time it takes to process a packet by a network node (router, switch, proxy, etc.)

  • Queuing delay is the time a packet waits before it is processed by a node.

  • Transmission delay is the time it takes to send the packet’s bits though the link

  • Propagation delay is time it takes for the signal to propagate from hop to hop.

Complete Chapter List

Search this Book: