Dear Mr. Li, experts,
it is quite discouraging that our questions remaining unanswered and the positive signs we were sending are ignored still.
We repeated our technical concerns for the UCLA proposal several times. Please be so kind to let everybody know what is needed to implement the proposal. So far provided documents are not sufficient.
The simulation scenario document exists for three months now. We were asking you several times to raise your objections for the provided simulation conditions. We were incorporating your favorite test values in the document. What are the reasons to defer the start of the tests still ?
The next meeting is approaching and results are needed as basis for the decision. For both of us it would be harmful to fight for details in front of the group again. We are soliciting you to co-operate - it is of mutual interest and the best for the standard.
Best Wishes Gero Bäse
----------------------------------------------------------- Gero Bäse Siemens AG Tel.: +49 89 636 53193 Corporate Technology Fax: +49 89 636 52393 Networks and Multimediacommunication CT IC 2 -----------------------------------------------------------
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ For help on this mail list, send "HELP ITU-SG16" in a message to listserv@mailbag.intel.com
Dear Mr. Baese, and Q.3 experts,
First, since I haven't got Mr. Baese's opinion on what we pointed out in the last email, I would like to repeat one more time and seeking a definite answer (i.e., agree or disagree with your reasons).
For the following in the testing condition documents "When the Annex K and/or Annex V is used, the length of the slice should be such to make packets fit in the length of segment to avoid segmentation of the IP packet (for fixed segment-loss-rate channels, see below)."
The reason for it:
If one want to evaluate if a protection scheme works, s/he need to show how much the performance improves between the cases "with" and "without" the proposed scheme. For a realistic measurement of the gain, the benchmark performance in the "without" case that one compared to needs to be the (near) optimal scenario. In other word, to demonstrate effectiveness of any scheme, one need to show that it can provide any gain over the *best* case the system has without it.
I think Mr. Baese agrees that use the right video slice length to avoid segmentation gives the (near) optimal performance for the case *without* any protection. Those annexes are in the mobile profile for a reason. That is the benchmark case that both of our proposals need to be compared to. This requirement is a fair condition and is not in favor of either proposal.
Mr. Baese wrote:
it is quite discouraging that our questions remaining unanswered and the positive signs we were sending are ignored still.
All the email correspondence between Mr. Baese and me are on the email reflector, so everyone knows what is going on.
It is strange to me that my explaination and reasoning for the testing condition gets no technical answers and is met with reactions from Mr. Baese such as above. I believe that we are here in ITU to discuss technical issues and to design the best technical solutions that works, not politics.
Mr. Baese, could you give me just your comments on my on the slice length (agree or disagree) with technical reasons?
Thanks,
Adam Li
---------- Adam H. Li Image Communication Lab (310) 825-5178 (Lab) University of California, Los Angeles (310) 825-7928 (Fax)
Folks,
At 04:47 PM 2/5/2001 -0800, Adam Li wrote:
Dear Mr. Baese, and Q.3 experts,
First, since I haven't got Mr. Baese's opinion on what we pointed out in the last email, I would like to repeat one more time and seeking a definite answer (i.e., agree or disagree with your reasons).
[...]
What is the point of even discussing conditions where entities (be it slices or partitions) are bigger than the MTU size? No reasonable implementation would do this.
For the following in the testing condition documents "When the Annex K and/or Annex V is used, the length of the slice should be such to make packets fit in the length of segment to avoid segmentation of the IP packet (for fixed segment-loss-rate channels, see below)."
I'm not sure whether this sentence is the best possible wording to describe that a packet should be smaller than the MTU size. But this is just a matter of wording anyway.
Stephan
P.s. Personally, I would be much in favor of letting Annex I die the peaceful death it deserves -- given the amount of support it receives.
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ For help on this mail list, send "HELP ITU-SG16" in a message to listserv@mailbag.intel.com
participants (3)
-
Adam Li -
Baese Gero -
Stephan Wenger