&%PAGE&
- &%page& -
Annex 6 to Doc. AVC-106R
Annex 6 to Doc. AVC-106R
LIAISON STATEMENT TO SGXVIII
Questions: 3,4/XV; 2,13,22/XVIII
SOURCE : EXPERTS GROUP FOR ATM VIDEO CODING IN SGXV
TITLE : LIAISON STATEMENT TO SGXVIII
PURPOSE: FOR ACTION
-----------------------
1.Introduction
The third meeting of the CCITT SGXV Experts group for ATM Video Coding
Experts was held in Santa Clara, 14-23 August 1991, to progress studies on
video coding for services on the B-ISDN.
The Experts Group sent some comments to SGXVIII to evaluate the
Integrated Video Services (IVS) Baseline Document after our second meeting.
We are glad to know that these comments have been adopted in IVS. To
achieve the best video service in B-ISDN we wish to send some additional
comments detailed in پک2 below.
The Experts Group also wishes to send some additional questions and
requirements which are detailed below in پک3.
2. Comments on IVS Baseline Document
2.1 Cell loss ratio (page 16, 2nd paragraph, 2nd and 3rd sentences)
Please revise the sentences as follows:
Current sentences;
"For example, a high quality videoconference connection... This may be..."
New sentences;
"Table 1 (reproduction of Annex 1 to AVC-96 with two bottom rows deleted)
provides some network performance requirements obtained from some example
service quality figures. The table concentrates on bit error and cell loss
error correction techniques. Layered coding concealment techniques are
however under consideration and lead to different figures."
Additional note to Table 1;
* These values are calculated under the assumption that cell losses are
isolated. If cell losses tend to occur successively, another cell loss ratio
and another cell loss correction technique may be required.
* We assumed that one cell loss always causes a picture degradation. The
visual perception of the picture, however, may be acceptable even if cell
loss concealment technique is not used. Therefore there is a possibility
that these requirements will be relaxed.
2.2 A2.6 Service bitrate (page 11, 2nd paragraph)
Usage parameter control and network parameter control, CLP bit.
"When the cell tagging option is exercised, non compliant CLP=0 cells may be
overwritten to CLP=1".
We are concerned that the network can modify the CLP bit. Some layered
coding techniques intend to use CLP bit for layer indicator. For such a
case, changing the CLP bit may cause more problems than discarding the cell.
This fact has already been mentioned in the last liaison statement from
SGXV;
Use of CLP bit (page 16)
"The CLP bit is seen as a useful mechanism to provide protection against
cell loss by controlling that information which might be lost. It is crucial
that, after a cell is labeled 'high priority' by a terminal device, this is not
changed by the network."
2.3 QOS related to Cell Loss Priority (bottom of page 12)
QOS related to Cell Loss Priority
"The CLP indicator in the cell header may be set by the user or the service
provider. In the case of video services, the CLP bit is set by the layered
coding provider..."
The wording should be clarified; e.g. who is "the layered coding provider"?
Is he a user?
2.4 List of H.26X requirements
We have identified such functional requirements as attached for the high
quality video coding standard H.26X, which we propose to be included in the
IVS Baseline Document.
3. Questions and requirements
3.1 Cell loss ratio
If cell loss ratio is rather low, we believe that the channel coding alone,
either at AAL or higher layer, can cope with low cell loss ratio based on
acceptable delay requirement. However if the cell loss ratio is extremely
high, some technique for cell loss resilience is required also for video
source coding.
On the other hand, we intend to select a video source coding algorithm at
the beginning of 1992. Therefore after that day, it becomes difficult to
implement additional cell loss resilience technique for source coding
algorithm.
Therefore we are eager to know the likely value of cell loss ratio as soon
as possible. For this purpose we sent the liaison two times. However, the
answer from SGXVIII is not clear yet. IVS Baseline Document only says that
the requirement from SGXV is that value and does not mention whether this
requirement is achievable or not from a network standpoint of view.
3.2 CLP bit
3.2.1 Negotiation for two priority flows
The liaison statement from SGXVIII at Geneva meeting, 11-28 June 1991 (annex
2, last paragraph) said as follows;
"There will be negotiation for both priority flows."
It is clear for CBR. However the following two types of negotiation can be
considered (Figure 1);
Case 1: Negotiations are done for both flow separately.
In this case low priority flow cannot use the erosion area of high
priority flow.
Case 2 : Negotiations are done for high priority flow and sum of both
priority flows.
In this case low priority flow is not restricted as Case 1 and can
use the erosion area of high priority flow.
The question is which is the likely solution.
3.2.2 Merit of using CLP bit
The question is what is the merit of using CLP bit. Layered coding is a
suitable technique to use both priority classes. However whether we adopt
this technique or not depends on its expected merit. What degree of
network resource saving can be obtained by using low priority cells?
This question is also related to the selection of video source coding
technique. Therefore quick response is required.
3.3 Usage Parameter Control for peak cell rate
The liaison statement from SGXVIII at Geneva meeting, 11-28 June 1991
annex 2 said as follows;
"A maximum Cell Delay Variation will be allocated to the Customer Equipment
(CEQ) between the ATM SAP and the interface at the T Reference Point.
Please show us the meaning of "a maximum cell delay variation". When we
assume a system configuration as in Figure 2, the minimum inter-arrival time
will change because of multiplex of multiple VCs at terminal adapter and NT2
as shown in Fig.3. Therefore, we cannot control peak cell rate at the T
reference point. We have two questions;
Question 1: What is the definition for peak cell rate at the S reference
point?
Question 2: What technique does SGXVIII recommend for multiplexing in the
adapter to keep the peak cell rate at the S reference point?
3.4 Requirements for AAL Type 1
We are concerned about the circuit emulation mode for existing standard
audiovisual terminals. Some tentative requirements have been identified and
are listed below. Further requirements may be identified as the work of the
Experts Group progress.
3.4.1 Interleaving
CMTT suggested to SG XVIII that the CS layer should be capable of
interleaving. Considering that the delay produced by the interleaving
processing depends on transmission rate, the interleaving function should
be optional, not mandatory.
3.4.2 Cell loss notification
Since not all erroneous information can be corrected by AAL, cell loss
notification is indispensable for the decoder to lessen the damage to the
reconstructed picture.
3.4.3 8 kHz timing
When conventional terminals are used, 8 kHz timing is necessary to
synchronize the first bit of each octet between the sender and the receiver.
CMTT also requires 8 kHz timing on behalf of conventional codecs, so SG XV
also requires 8 kHz timing for the same reason.
3.5 AAL Type 2 SAR
The major user of AAL type 2 will be video services. Therefore we agree
that we have responsibility for providing the major inputs to SG XVIII
leading to the definition of AAL type 2.
Given the necessity to support a wide range of video services with different
requirements, rates, etc., it appears that the fields given as examples in
I.363 AAL Type 2 example may be restrictive. For example, an AAL Type 2
with minimum functionality may offer the flexibility to accommodate a wide
and diverse range of video services, but the minimum functionality is now
under discussion.
Commonality of the SAR-Sublayer for CBR & VBR video
It is the common view held by the Experts Group that issues in relation to
SAR functionality to support VBR video using AAL Type 2 apply equally to
the support of CBR video. Therefore, in line with the Experts Group's
desire to contribute to the development of requirements of the AAL Type 2
for the support of VBR video, AAL Type 2 will be developed within the
Experts Group to support both VBR and CBR video services.
3.6 Network interworking
1) When network provides interworking function between N-ISDN and B-
ISDN by network gateway as shown in Fig.4 (reproduction of Fig.1/AVC-91),
what types of AAL are required for B-ISDN side terminal?
2) How does B-ISDN intend to provide transparent N-ISDN circuit
emulation especially those for cell loss sensitive and time delay critical
services such as visual telephone using H.261, H.221 etc.?
3) What slip rate is expected when providing N-ISDN circuit emulation
between N-ISDNs via B-ISDN?
4) The Experts Group is considering the conversion of N-ISDN user
multiplexed signals to B-ISDN VC multiplexed signals either in Terminal
Adapters (TA) or B-ISDN/N-ISDN Inter-Working Units (IWU). Current user
multiplex structures (e.g. H.221) can reconfigure their internal rate
allocation in the order of 20msec. The Experts Group requires information
on the possibility of;
a. Associating a group of virtual channels with different QOS
requirements with a a single resource allocation, and
b. Resource allocation renegotiation in the order of 20msec.
5) What N-ISDN bit rates will be supported by the B-ISDN circuit
emulation mode?
3.7 Signaling for multimedia synchronization
One obvious consequence of mixing media on a single virtual path is that the
Quality of Service (QOS) required must correspond to that of the most
sensitive service, and for many applications this may not be cost effective
solution (i.e. mixing loss sensitive data and delay sensitive video traffic)
and multiple virtual paths may be required. Multiplexing all services onto
a single VP results in zero cross media delay at the cost of a potential
mismatch of QOS. Separate VPs ensure a QOS matched exactly to the media.
We are considering to use one VC for each medium for each service, in other
words, each service component (multiplex in ATM layer). For multimedia
communications, a cross media maximum differential delay should be
guaranteed. The typical case is the lip-sync between video and audio. For
this requirement it seems that multimedia call should be marked/indicated.
Signalling for call establishment and for the addition and deletion of media
components must therefore be capable of indicating that particular services
are associated for the purpose of synchronization. Further study is
requested for signaling and control to handle the cross media delay to be
minimum.
3.8 Technique to support low bitrate information
We intend to use multiplex in ATM layer as mentioned above. However there
exists very low bitrate information such as pointing or telewriting. What
technique does SGXVIII recommend to transmit that kind of low bitrate
information (e.g. 300 bit/s) in multimedia connections?
4. Conclusion
This document is including some comments to IVS Baseline Document and has
raised some important questions and requests concerning the network issues.
The SGXV Experts Group for ATM Video Coding intends to continue close
relation to SGXVIII for the development of video services.
END