&%PAGE&

-  &%page&  -
Annex 5 to Doc. AVC-177R
Annex 5 to Doc. AVC-177R

ADDENDUM TO LIAISON STATEMENT FROM SGXV EXPERTS GROUP TO SGXVIII 

Source : Experts Group
Purpose: Action

----------------

1.	Introduction

At its fourth meeting in Yokosuka, Japan, November 1991, the following issues 
were discussed by the CCITT SGXV Experts Group for ATM Video Coding, and 
recognized as items about which SGXVIII could provide information.  This 
addendum to the liaison statement prepared at the third meeting in Santa 
Clara, August 1991, seeks response from SGXVIII on the following issues.

2.	UPC and NPC

The choice of algorithm used to perform UPC and NPC functions has an impact 
on the video coding scheme, since compliance will affect the rate control 
strategy, particularly for VBR video.  Standardization of the UPC/NPC 
algorithm(s) is essential to ensure that terminal equipment can be used on 
any network and that inter-network (e.g, international) connections can be 
established.

The Experts Group wishes to advise SGXVIII of the impact that the UPC/NPC 
algorithm may have on the video coder, and seeks information on the 
program of standardization efforts in this area.

3.	Clock recovery

While simple synchronization strategies such as frame repeat or frame 
dropping are possible for lower quality video services, such methods are 
not acceptable if high quality is to be maintained.  Therefore, many video 
services will require a means of synchronizing encoder and decoder video 
clocks.

While the availability of a universal, synchronous, network based reference 
clock could provide a solution, the Experts Group has a concern that this 
may not always be available, particularly if network access is via a LAN and 
PBX.

In this latter case, an explicit means of providing clock synchronization is 
required.  The Experts Group notes, form the report of the SGXVIII  
Rapporteurs Meeting on AAL (Ottawa, 1-3 October 1991), that this is an area 
being studied by CCITT SGXVIII.  Both SFET and Time Stamping techniques are 
discussed in that report.  We believe that Time Stamping is more appropriate 
for VBR services than is SFET, and have previously expressed our strong 
desire for commonality of AAL for the support of VBR and CBR video.  We 
request that SGXVIII take this into account during its work in this area, and 
request early advice of progress.

4.	Quality of Service guarantees

The Experts Group understands that, as yet, no agreement has been reached 
within SGXVIII on whether a guaranteed QoS will be offered on Low Priority 
channels.  As the QoS (CLR, delay) has a major impact on the video coding 
algorithms appropriate for video service support, the Experts Group urges 
SGXVIII to give urgent attention to this issue and request early advice of 
progress.

The Experts Group also perceives a risk if no values of QoS are 
standardized.  If the same QoS cannot be provided on different networks or 
in different regions, this would compromise the viability of large terminal 
markets and communications spanning different regions.   

5.	ATM Header availability at receiving terminal

In discussion within the Experts Group and elsewhere, it has been suggested 
that certain bits within the ATM header (in particular, CLP and PT), if 
available to a receiving terminal, could provide useful information 
regarding the cell content and avoid the need for duplicate labeling at a 
higher layer.  The viability of using this technique is dependent on the ATM 
header being passed up from the network to the receiving terminal device.  
Can SGXVIII advise whether this will always be possible?

6.	Burstiness characteristics of cell loss

The length of a cell loss burst has a significant impact on the means of 
error protection/correction in a video coding system, and could even 
influence the basic coding architecture.  While we appreciate the SGXVIII's 
information on this matter provided by WPXVIII/6 in November 1990, a 
quantitative description of the expected cell loss burst behavior is desired 
at the earliest opportunity.  

The Experts Group notes, from the SGXVIII Rapporteurs Meeting on AAL 
(Ottawa, 1-3 October 1991), that a modulo 8 sequence number is considered 
adequate to detect error in a CBR, AAL Type 1 application, would this also 
be appropriate for a VBR video transmission using AAL Type 2?  Any 
information that would give a quantitative description of cell loss burst 
behavior would be welcome.  Are any measures of cell loss burst length 
likely to be included in the QoS parameters agreed between user and 
network?

7.	Multimedia multiplexing

Some advantages of multiplexing the components of a multimedia call on 
different VCs (i.e. cell-based multiplexing) have been recognized by the 
Experts Group.  To use this approach, however, requires determination of the 
media type by the receiving terminal at the time the connection/call is 
established.  What is the mechanism by which this media identification is 
envisioned to occur?

8.	Conclusion

In this addendum to the liaison statement to SGXVIII prepared at the August 
meeting in Santa Clara, the SGXV Experts Group for ATM Video Coding has 
provided information on the items it sees as important in progressing its 
video coding standardization activity.  Consideration of, and response to 
the issues raised above are requested.


END