International Telecommunication Union   ITU
Site Map Contact us Print Version
 Monday, November 14, 2005

Three new Recommendations related to IP Performance have been consented by ITU-T's Study Group 12.

G.1030 - provides a framework of tools to estimate end-to-end IP network performance for some user applications. User perception of application performance in packet networks is dependent on many factors, including network end-to-end performance, performance of terminals and other devices beyond the purview of the network operator. The application’s dependency on the communications network, and the user’s task and the extent of user interaction with the application need also to be taken into account.

All these factors are used to estimate end-to-end performance levels. At this stage, the framework includes a perceptual model for web browsing. Future versions will focus on multimedia conferencing and other applications. The Recommendation is designed to be helpful for people designing networks, enabling them to know what applications can be realistically supported. 

G.1040 - defines a new performance metric in IP networks for short transactions, such as trading of stocks, automated banking, and credit card point of sale transactions. The nature of such exchanges is that they need to be quick and reliable.

This Recommendation gives the ability for the network provider to either flag a problem based on their network measurements interpreted with this metric, or to say that – if a problem exists – it isn’t attributable to the network. The Recommendation allows the network service provider to see how much of the transaction time can be attributed to the network. The metric can also be useful in drawing up service level agreements.

G.1050 - addresses Network Model for Evaluating Multimedia Transmission Performance Over Internet Protocol. The need for such a model is driven by new challenges for multimedia applications in IP. Impairments that in typical data transfers are of little consequence may be much more serious in video or VoIP for example. The model is based on statistical models of a broad range of known deployed network configurations. This way a manufacturer of networking testing solutions can avoid speculation in configuring test scenarios.