Session chairs:

Francois Le Faucheur (CISCO), France

Georgios Karagiannis (University of Twente), The Netherlands

QoS is a key topic for the research community, but key questions remain:

• Why is there a disconnect between QoS in the research community and in the industry?

• Are the researchers selecting irrelevant problems to solve? Are the "solutions" not convincing?

• What are the structural issues for deployment?

• Do industry and research agree on the topics for study and criteria for success?

• How can research influence commercial solutions and deployment and how can industry influence research?
  
  

This session invites short paper submissions (up to 5-6 double-spaced single-column pages with font sizes of 11 or larger, including all figures and references, see the Instructions for Authors page) providing elements of answer to these questions, in particular with respect to the following QoS aspects:

• Inter-Provider QoS: what is needed from technical, operational and commercial viewpoint?

• QoS in mobile environments: are mobile solutions and QoS solutions converging or diverging?

• Security threats and solutions in end-to-end QoS signaling. What are the fundamental interactions of QoS with security? Clearly you need to secure access to QoS services in order to control access and bill, but what about the flip side? Can many security problems be recast as QoS problems and what advantages might be obtained by doing so?

• Interactions between precedence-based QoS (with resource preemption) and network overload e.g. in the case of disasters or other impulse events? Is there danger of signaling melt-down or massive under-utilization of resources due to thrashing? What can be done to mitigate the situation if this is viewed as a real practical problem?

• How do the statistics of the traffic matrix affect the viability of different QoS mechanisms and signaling? We know that at the extremes you need either per-microflow QoS (where a single flow uses a significant fraction of a link bandwidth), or coarse class-based QoS (where the law of large numbers holds). What about the transition region? Can it be detected? Are there useful hybrid mechanisms? Can they be applied adaptively? Are there useful QoS protocols that can be used to signal such hybrid mechanisms?

Note that the objective of this session is not to discuss the low level details of a particular QoS problem or solution, but rather to provide views/analysis of QoS problem space and QoS solutions and how those are being, or should be, addressed by the research, the standardization and the industry communities.