Delay-based Sub-grouping for NORM in Heterogeneous Networks

Chamil Premantha Weerasinghe Kulatunga, Godred Fairhurst

Research output: Contribution to conferenceOther

Abstract

The IETF Negative Acknowledgement (NACK) Oriented Reliable Multicast (NORM) protocol offers an attractive solution for small groups that do not have significant receiver heterogeneity. It avoids many of the problems associated with Asynchronous Layered Coding (ALC), such as complexity, Internet Group Management Protocol (IGMP) and routing overhead, lack of TCP friendliness and data duplication. NORM uses a combination of a representative and suppression based approach for reliability and congestion control to avoid multicast feedback implosion. The group's Maximum Round Trip Time (MRTT), maintained at the sender, becomes a metric in this scheme. A large MRTT makes the NORM services very unresponsive. NORM can also be deployed in a heterogeneous network environment where the RTT may differ considerably between sets of users in the network due to their differing propagation delay. An example is when some of the NORM receivers operate over a satellite network, driving the MRTT to a high value. In some satellite scenarios, e.g Digital Video Broadcast - Return Channel via Satellite (DVB-RCS), the feedback packets have to contest for return-path link access. This can lead to large variations in the RTT and results in a MRTT considerably higher than the average RTT. This paper investigates the situations where the NORM framework could be inefficient. By considering the impact on the MRTT, a new mechanism is proposed based on delay-based receiver grouping. This improves the performance of NORM in heterogeneous network environments, including those provided by DVB-RCS satellite networks.
Original languageEnglish
Pages131-135
Number of pages5
DOIs
Publication statusPublished - Sept 2007

Keywords

  • delay-based grouping
  • multicast congestion control
  • NORM
  • satellite networks

Fingerprint

Dive into the research topics of 'Delay-based Sub-grouping for NORM in Heterogeneous Networks'. Together they form a unique fingerprint.

Cite this