&%PAGE&

-  &%PAGE&  -
Document AVC-256R
CCITT SGXV                                            	Document AVC-256R
Working Party XV/1                        			Revision
Experts Group for ATM Video Coding				March 27, 1992

SOURCE : CHAIRMAN
TITLE  : REPORT OF THE SIXTH MEETING OF THE EXPERTS GROUP FOR
         ATM VIDEO CODING IN STOCKHOLM/HAIFA (March 18-27, 1992)
         - PART I
Purpose: Report

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

PART I  SOLE SESSIONS
Contents

			1. General
			2. Documentation
			3. Tape demonstration
			4. Reports of the meetings relevant to the Experts Group
			5. Picture format
			6. Network aspects
			7. Source coding
			8. Work plan
			9. Others


1. 	General

The sixth meeting of the Experts Group consisted of two parts; CCITT sole 
sessions in Stockholm and joint sessions with ISO/IEC JTC1/SC29/WG11 (MPEG) 
in Haifa.  The list of participants appears at the end of this report. 

The first part was held during 18-20 March 1992 at Najaden Hotel in Haninge,  
Sweden, at the kind invitation of Telia Research AB.  At the opening session, 
Dr. Oesten Maekitalo, Director of Telia Research, made a welcoming address 
on behalf of the hosting organization.  

At the end of the Stockholm sessions, Chairman thanked the hosting 
organization for the meeting facilities provided and the excellent 
secretarial support.  Chairman also thanked Mr. Katsuyuki Yamazaki, Special 
Rapporteur SWP XVIII/8-3, for his participation in this meeting and for his 
efforts to communicate between SGXVIII and this Experts Group.

The Experts Group also had a short closing session on March 27 in Haifa.


2. 	Documentation (TD2)

For this meeting, 52 AVC-numbered documents and 8 temporary documents were 
available as listed in Annex 1.


3.	Tape demonstration (TD4)

Several video tape demonstrations were given with D1 as detailed in Annex 2 
to present experimental results.  

Display of progressive scan materials at different scanning formats were 
first made available for the meeting, thanks to the effort of the hosting 
organization.  Experimental results were brought in by recording in either 
of Exabyte or D1 tape.


4. 	Reports of the meetings relevant to the Experts Group

4.1 SGXVIII meeting in Melbourne (AVC-178,207,208)

Mr. Biggar presented an overview of SGXVIII achievements focusing on the 
matters about which we sent liaison statements.  Correction was made on the 
QoS of the low priority cells (Section 3.6/AVC-178) that at the moment it is 
not guaranteed.

Mr. Yamazaki gave detailed content and background for each technical item 
contained in the SGXVIII liaison responses as introduction of our discussion.

The meeting appreciated this mutual attendance of representative experts 
between this group and WP XVIII/8, and recognized it quite valuable for the 
progress of each group's work.  Further opportunities should be sought in 
the future.
  
4.2 CMTT/2 SRG meeting in Singapore

Mr. Brusewitz outlined the current SRG activities of the Special 
Rapporteur's Group which is focusing on the compatible coding between HDTV 
and Conventional TV.


5. 	Picture format 

5.1 Status before this meeting

Chairman summarized the study progress up to this meeting as follows;

    -	Interim common understanding was obtained in August 1991 as recorded 
in Annex 3 to AVC-106R (Report of the Santa Clara meeting).
    -	We had agreed at that time to reach an agreement on the picture 
format issue at the earliest occasion in 1992 (Section 4.2.5/AVC-106R).

5.2 Format conversion (AVC-213,225,239,240,244; AVC-242)

Several format conversion experiments were presented as in Documents AVC-
213,239,240 with tape demonstration, which carried out conversion from the 
CCIR-601 format to the progressive SCIF as in AVC-29 or further back to the 
CCIR-601 format.  The issue is whether there are any practical conversion 
filters between CCIR-601 (525/60 and 625/50) and SCIF.  AVC-242 gave an 
evaluation of several interlace to progressive conversion methods in terms 
of camera cost, conversion hardware, possible picture quality and codec 
hardware.  In AVC-244, a trial to provide objective evaluation of format 
conversion was presented for the future work.

The following general comments were given;

    -	SNR comparison of conversion methods may not be sufficient, pictures 
should be observed.
    -	Existence of cameras with fast shutter and crispening post-process 
should be taken into account.  They may make conversion extremely 
difficult.
    -	An underlying question is whether the hypothesis stands here that "we 
are coding pictures but not pixels".  Interlace to progressive 
conversion and frame rate conversion are two challenging items.

In addition to the format conversion results, progressive pictures obtained 
from film sources and their interlaced version were also presented (AVC-252).  
The following is the summary of impressions expressed after having observed 
those pictures;

    -	Loss of quality is noticeable, particularly for frame rate conversion.  
Broadcasters will not accept this degradation.
    -	Progressive scan display for progressive materials are very impressive 
compared to their interlace version.
    -	Displaying on a small CRT monitor or a large projection monitor gives 
different impression.

5.3 Coding at SCIF (AVC-239,241,252)

Simulation results for coding through SCIF were reported in AVC-239 and 
241, addressing its impacts on coding efficiency and prediction structure.  
It was found that practical conversion filters may lose 10-30% in coding 
efficiency and require multiframe prediction.

AVC-252 gave coded results for two progressive materials, pointing out that 
coding rate does not double due to progressive scan.

5.4 Decision policy (AVC-215,219,220,243,246,250; TD-5) 

Considerations were given for making a decision on the H.26X format issue in 
AVC-219,220,246,250 where a flexible resolution approach, clarification of 
SCIF aims, and square pixel approach were studied.  During the discussion, 
the following was pointed out;

    -	H.26X should be generic to be applicable to a wide range of 
applications, thus it should cover a range of formats.  What picture 
formats are to be used may be the issue of H.32X (audiovisual 
communication terminal for B-ISDN).
    -	What applications are really transported through B-ISDN should be 
carefully studied.  As extreme arguments, there would be no demand 
for videoconferencing at higher bit rates such as 5-10 Mbit/s, or 
there would be outside the terminal solutions for interworking 
between different television standards such as in broadcasting.

After some discussion, Mr. Morrison undertook to coordinate a drafting 
group to make a summary of discussion for considerations, conclusion at this 
meeting and necessary future actions.  The outcome is contained in Annex 3. 

There was a discussion on the linkage between format and algorithm, and the 
meeting confirmed that at this stage H.26X should include operation at 
CIF/QCIF as well.


6. 	Network aspects

6.1 AAL Type 1/2 updated specifications (AVC-208,210,214)

Mr. Yamazaki gave the meeting a tutorial presentation on the updated 
specifications of AAL Type 1 and possible candidates for AAL Type 2 
functionalities.

6.2 Review of AAL Type 1 specifications 

Video sampling clock recovery was reviewed in AVC-249 with reference to the 
SRTS method of AAL Type 1.  Based on the agreement that we provide a 
mechanism in H.26X which allows video sampling clock recovery (Section 
5.8/AVC-177R), the meeting considered whether this be achieved using AAL 
function or as part of video codec functions.  

This should be further studied considering VBR operation of the video codec 
and clock recovery for multiple sources.  It should be noted that some 
audio coding such as Compact Disc uses sampling frequency independent of 
the network's 8 kHz and quite jitter sensitive.  The following was also 
pointed out;

    -	At T/S reference points, 150 MHz network clock is available, but its 
submultiples may have to be generated inside the terminal.
    -	Some NT2 equipment, such as LAN and PBX, does not provide precise 
clock frequency nor are locked with B-ISDN clock frequency.  SRTS 
method is based on the availability of the identical clock at both 
ends.

An experimental result addressing the jitter specification was presented in 
AVC-230.  It was pointed out that the jitter specification uses a template 
for its frequency characteristics and that such specifications are found in 
G.810, G.823 and G.824.  It was also pointed out there may be such sampling 
clock jitter specifications in the existing standards, and that component 
and composite television signals may require different characteristics.

6.3 Required functions of AAL Type 2 (AVC-210,222,228; AVC-226)

Mr. Yamazaki clarified during the presentation of AVC-210 that SGXVIII 
intends to finalize the protocols for video transport in the 1994 
Recommendations, thus welcomes the input from this group on required 
functions.  It was also clarified in response to questions that CLP is 
determined on a per-VCC base as well as a per-cell basis, and that use of 
particular AAL Type is made known between the terminal and the network 
through designation of a bearer service (Type A, B, C or X). 

As to the necessity of partially filled cells, it was pointed out that there 
is a possibility to use stuffing bits in the video coding instead of using 
length indicator in the AAL.

The meeting accepted several items listed in AVC-222, together with others 
identified in the multi-media/multi-layered video support, as possible 
requirements to the AAL Type 2 and agreed to communicate with SGXVIII on a 
"may be required" basis except the commonality item.  This exception is 
reflecting the SGXVIII's intention to make the protocols common as far as 
possible.  "Framing of video data" item was enhanced by pointing out the 
necessity of cell boundary indication at AAL-SAP.  

AVC-228 gave a stimulus to study the implications of LAN/MAN characteristics 
on the ATM video coding standard.  Since this is a new item, the members are 
requested to provide input.  There were comments that;

    -	802.6 like networks are considered as a possible NT2 configuration,
    -	the distance between gateway and the terminal will affect the problem,
    -	there will be difference between the case of "from the encoder in B-
ISDN to the decoder in LAN/MAN" and the case of the other way around,
    -	difference between LAN and MAN should be taken into account.

6.4 CLP (AVC-208,223)

In the presentation of AVC-208, it was clarified that one bit can be provided 
as part of AAL video transport service, which classifies the cell into two 
types.

The meeting agreed to the view contained in AVC-223 that CLP bit be used 
only for cell loss priority indication.

6.5 UPC/TD (AVC-208,227,247)

Mr. Yamazaki provided the SGXVIII results on the negotiation of peak cell 
rate for CLP=0,1 and Cell Delay Variation (CDV) definition points, presenting 
the three causes for CDV; multiplexing in AAL layer, multiplexing in ATM 
layer and GFC at ATM layer.

The meeting accepted the proposed items for clarification addressing CDV in 
AVC-227 and agreed to communicate with SGXVIII.

Document AVC-247 discussed monitoring points of VBR traffic, emphasizing the 
following fundamental problems;

    -	Difference of methods to monitor the traffic between the terminal and 
the network may cause cell discards which the terminal can not 
predict.
    -	If a UPC mechanism is not standardized, different UPC methods in 
different networks may further cause unpredictable cell discards.

Since these are quite serious for achieving VBR video coding, the issues 
should be kept alarming even if we sent SGXVIII the same comments before.  

AVC-247 also brought up the issue of using performance metrics for UPC 
performance evaluation.  As long as different UPC algorithms conform to the 
performance metrics bounds specified, no standardization of a UPC algorithm 
may be needed.  Its implication on the VBR coding should be addressed. 

6.6 Multimedia multiplex (Multi-component multiplex) and B-N interworking

	6.6.1 Multimedia multiplex for audiovisual services (AVC-211,224,226,248)

Document AVC-211 includes Draft Recommendation I.37y "Network capabilities 
to support multimedia services" which gives general conceptual guidance on 
the subject matter.  Chairman gave a comment that real time audiovisual 
services of our concern seems not covered in this document.

Document AVC-224 proposed to ask SGXVIII the possibility of providing B-ISDN 
to N-ISDN audiovisual service interworking which includes the H.221 protocol 
conversion inside the network, but the meeting felt this not practical and 
concluded not to include this question in the liaison document.

AVC-226 and 248 are addressing the study item of what multiplex method be 
appropriate for the H.32X terminal.  In AVC-226, a concept of switchable 

multiplex between H.221 based and cell oriented ones is provided.  As to the 
comments in AVC-248 on comparison of the multimedia multiplex methods, 
Japan was requested to take them into account in the future contribution.  
We need further study on this study item.

During the discussion, Mr. Morrison pointed out that the descriptions 
concerning multimedia multiplex method specified in the MPEG-1 system 
(Section 4.1/AVC-226) are not correct. 

	6.6.2 Support of multi-layered video (AVC-216,221)

These two documents address a similar topic; transport of multi-layered 
coded video signals over B-ISDN.  Taking into account that this topic has 
also some similarity with multimedia multiplexing discussed above, listed 
questions to the network specialists were integrated and sent to SGXVIII as 
part of the liaison document.

6.7 Cell loss

	6.7.1 network characteristics (AVC-208)

Mr. Yamazaki presented SGXVIII's response to our question concerning the 
network performance; cell error ratio and cell loss estimated from the G.82X 
specification.  It was noted that these values are based on the radio 
transmission systems and optical fiber systems are used in actual B-ISDN, 
thus much better performance is expected.  It was also noted, however, that 
the estimation in this liaison document does not include cell losses which 
may be caused by ATM nodes at the time of congestion, thus worse 
performance than this may be provided.

The meeting felt it safe that the video codec be resilient to this level of 
network performance; by means of e.g. including FEC to cope with bit errors 
as in H.261, and cell loss resilience techniques in the video source coding.

	6.7.2 cell loss resilience techniques (AVC-235,236,252; AVC-222)

There were presented experimental cell loss resilience techniques and 
survey of possible techniques as well as their selection criteria.  Mr. 
Haskell commented on the leaky predictor approach that channel hopping 
characteristics may be covered by leaky prediction instead of cyclic I-frame 
insertion and that trade-off between use of I-pictures and leaky prediction 
should be carefully studied in terms of coding efficiency.  Mr. Tabatabai 
commented on Figure 1/AVC-236 that trade-off between delay and 
"interleave+FEC" transmission efficiency  should also be carefully studied.

In AVC-252, layering of packetizers for two priority channels are 
experimented where the decoder ignores all the subsequent bits after a lost 
cell and used cell loss concealment techniques.  

	6.7.3 elements to be adopted in the standard

The meeting discussed for a while what elements be chosen for the standard 
or what should be studied toward that target.  The possible elements are as 
follows though they are not exhaustive;

    -	use of CLP bit (transmission coding)
    -	layering (source and/or transmission coding)
    -	leaky prediction (source coding)
    -	FEC/interleave (AAL)
    -	structured packing of video data into cells (transmission coding)
    -	concealment (outside the standard)

During the discussion, it was clarified that for layered video coding each 
layer can use different options of AAL on a per-VCC basis at the call set-
up.

Mr. Parke pointed out that the selection of error resilience techniques 
depends on such system considerations as the H.26X compatibility with H.261 
or other standards.  

The meeting concluded that further study is required. 

6.8 Reference terminal configuration (Figure 1/AVC-227)

The meeting recognized the importance of reference configuration of the 
audiovisual ATM terminal toward the following;

    -	identification of reference points and interface signals at those 
reference points 
    -	clarification of responsibility for specification 
    -	identification of missing elements for designing audiovisual 
communication terminals connected to B-ISDN

and appreciated Figure 1/AVC-227 as a useful input.  Members are requested 
to elaborate this.

6.9	Liaison to SGXVIII (TD-6)

As a summary of the discussion, the meeting agreed to send SGXVIII a liaison 
statement containing our responses to the request of comments and our new 
questions identified through the discussion of the week.  Mr. Biggar 
undertook to coordinate a drafting group for this purpose.  

The outcome of this drafting group is contained in Annex 4.


7.	Source coding

7.1 CBR and VBR (AVC-218,255; AVC-252)

An overview of the issue and checkpoints for choice were given in AVC-255.  
The questions raised in the illustration have been recognized as open in 
this group.  Contributions are awaited.

This contribution will be presented at the workshop during the joint 
sessions with MPEG in Haifa.

7.2 TMO experiments

	7.2.1 Specifications (AVC-212)

Mr. Schinkel introduced the preliminary working draft for Test Model as a 
starting point of the convergence phase work, which was edited by an adhoc 
group established at the Singapore meeting.  Stress was made on the 
following points;

    -	Cell loss experiment specification should be materialized as real 
experiments.
    -	H.261 compatibility should be experimented.  Currently the syntax 
allows it, but work needs to be done to achieve compatibility.  

	7.2.2 Performance and improvements

The following items were experimented mostly under the TM0 framework.

   1) Field vs frame coding (AVC-231,232)

There was a comment on AVC-232 that multiplication by the coefficients for 
field adjusted motion compensated prediction and dedicated motion detection 
may bring hardware complexity.  It was also clarified that by setting "MVfld" 
to zero progressive formats can be dealt with. 

   2) Half-pel MC (AVC-237,245)

Two comparison results were presented on whether the original picture or 
previous coded picture be used for final half-pel accuracy motion detection, 
but they differ in conclusion.  One possibility was thought to be whether 
the integer pixel position is recalculated in the final vector search of AVC-
237.  This point need be checked.

   3) Averaging two fields/frames for prediction (AVC-233,245,251)

Three different techniques imply that averaging two fields/frames for 
prediction is effective in improving coding efficiency.  Effectiveness of bi-
directional B-frames is in line with this fact.

   4) Quantization (AVC-245)

It was clarified that shifting the decision level is effective if variable 
length coding after quantization is also counted (lower level of 
coefficients generate less bits), though the decision level which gives 
minimum quantization error is in the middle of two adjacent reconstruction 
levels.  Mr. Haskell commented that this shift of decision level had been 
found effective, but it may be related to perceptive or temporal filtering 
effects.  

   5) Low delay mode (AVC-233,252)

Two experiments are in line to show that use of B frames improve coding 
efficiency at the cost of delay time.  Improvements of low delay mode 
operation are encouraged.

   6) Embedded coding (AVC-234)

Use of base-layer for prediction is discussed in this document.  Difference 
in the down- and up-conversion filters affected the coded results in the 
first experiment but not in the second.  It was commented that this may be 
due to difference of the coding noise included in the base-layer pictures.  
Another comment was given that coding efficiency may be affected by use of 
temporal interpolation instead of field-repeating in the up-conversion of 
base-layer pictures.

   7) Adaptive VLC (AVC-245)

Three dimensional VLC using LEVEL, RUN and whether the coefficient is the 
last in a block or not as entry values was shown effective for both block 
scanning and frequency scanning.

   8) Low bit rate operation (AVC-238)

This document indicates a lower boundary of the current rate control 
method.  It was noted that this should not be confused with the low bit rate 
operability of the coming standard.  For example, spatial or temporal pre-
filtering may allow the satisfactory low bit operation.

7.3 Framework for H.26X (AVC-229; AVC-216,221; AVC-253)

Impacts of embedded coding were discussed in AVC-229.  The following 
comments were given;

    -	Being generic and guaranteeing connectability need a careful balance.
    -	Structure to switch embedded and simulcast might solve some problems.
    -	In LAN applications, storage capacity and throughput requirements of 
the server may need embedded coding.  Its usefulness depends, 
however, on the choice of complexity in the database or that in 
decoders.
    -	"Scalability" is relevant to this discussion, but distinction should be 
made between upward/downward compatibility and forward/backward 
compatibility. 
    -	Activities toward H.261 compatibility should be enhanced. 

The meeting recognized this as a stimulus for the future work, awaiting 
further input.

Document AVC-253 is a list of requirements for MPEG-2/H.26X.  As noted in the 
summary of picture format discussion, the Experts Group should clarify the 
probable applications and their implications for H.26X.

7.4 Position of Experts Group for the first Test Model definition 

	7.4.1 Work method (AVC-217; TD-7)

The meeting discussed how the source coding work should be organized in 
this group.  There was raised a question whether we should define our own 
Test Model as a subset of the MPEG TM.  

It is a common view of the group that the Test Model be joint CCITT/MPEG 
model and not that we use the MPEG model.  We should make efforts to 
improve TM features of our concern such as low delay, cell loss resilience, 
compatibility.  Other members from different fields may concentrate on 
another features.  TM improvements will be a collection of such efforts.

There was also expressed a concern on the procedure to improve Test Model.  
Currently, if an improvement is to be included in the Test Model, core 
experiments using the Test Model are required.  Since this seems to be a 
cyclic argument, necessary procedures should be clarified during the joint 
sessions.  

It was further commented that the syntax to be developed through the Test 
Model work should be reviewed before finalizing the standard from 
transmission efficiency and hardware complexity point of view and necessary 
modifications should be made.  

As to the low delay experiment, the meeting agreed to make a specific 
proposal as in Annex 7.

The meeting agreed to the view expressed in AVC-217 that Video and 
Requirements sessions be in sequence for matters relevant to MPEG-2/H.26X, 
recognizing that there may be some difficulty due to limited practical 
arrangements. 

	7.4.2 Documents

The following AVC-numbered documents are put forward for consideration of 
the joint sessions;

	Proposal   : Annex 5 to this report, AVC-217
	Discussion : AVC-229,236,251,252
	Information: AVC-231,233,234,235,254,255

Note: Those documents are not included whose contents are covered by the 
ones submitted through the MPEG channel.

	7.4.3 Representatives

The following is appointed as representative to the joint session;

	Requirements/Test	S. Okubo
	Video			G. Bjoentegaard
	System		B.G. Haskell
	Implementation	D.G. Morrison

	7.4.4 Speakers at the Haifa workshop

Chairman reported that on behalf of the Experts Group Mr. Barry G. Haskell 
(AT&T Bell Labs) and Mr. Peter List (DBP-Telekom) will kindly give talks on 
variable bit rate coding under UPC at the workshop in Haifa.


8. 	Work plan

8.1 Joint work with MPEG 

See Section 7.4.1 above.  Relation between H.26X and MPEG-2 in the final form 
should be further studied.

8.2 Coordination with CMTT/2 SRG

Further actions toward the joint work may be necessary, but the meeting 
could not find time to discuss this matter.

8.3 Consideration of relevant CCIR works

Since the impact of digital broadcasting on the B-ISDN communication 
terminals will be great, the meeting felt it necessary for the Experts Group 
to liaise with CCIR groups and invite to use the outcome of the joint 
MPEG/CCITT EG standard development.  As an initial step, the meeting agreed 
to send a progress report of this group to a relevant CCIR Bologna meeting 
on coding in May 1992.

8.4 Participation in the SGXVIII meeting

Volunteers are solicited who can attend the SGXVIII meeting held during 9-19 
June 1992 in Geneva to liaise between this Experts Group and SGXVIII as 
mentioned in Section 4.1 above.

8.5 Milestones

Though "outline recommendation" for H.26X had been intended around this 
time (see Annex 5 to AVC-22R) for an input to WPXV/1 meeting in May 1992, the 
meeting concluded that we have not sufficient materials and that submitting 
a progress report would be more appropriate.  This conclusion was also 
affected by the fact that Working Party meetings are held regularly without 
being constrained by the study period change at the end of this year. 


9.	Others 

9.1 Intellectual property (TD-3)

The practice of the patent application disclosure in the previous 
Specialists Group was reviewed, which collected such information before the 
applied patents appear in the public domain on a "if everybody does so" 
basis.  Since some difficulty is felt this time because of the joint work with 
MPEG, the meeting agreed to raise to start the collection of patent 
information relevant to the Test Model for consideration of MPEG/Chairmen 
meeting.

9.2 Document handling

The Experts Group agreed to distribute written contributions 7 calendar 
days in advance of the meeting.  Contributions received after that deadline 
may not be considered by the meeting.

9.3 Future meeting plan (sole sessions)

    -	July 1-3, 1992 in New Jersey before the Rio de Janeiro joint sessions 
with MPEG
    -	September 1992 in USA, joint sessions with MPEG
    -	October 28-30, 1992 in Ipswich (?) before the London joint sessions
    -	March 1993 in Australia, sole and joint sessions


END

*   *   *

Annexes

    Annex 1	Documentation
    Annex 2	List of tape demonstration
    Annex 3	Summary of discussion on the picture format issue
    Annex 4	Liaison statement to SGXVIII
    Annex 5	Proposal on low delay mode experiments
Participants of the sixth meeting of
Experts Group for ATM Video Coding 
(18-27 March 1992, Stockholm and Haifa)


										S H 

FRG		Mr. P. List			DBP Telekom			  X
		Mr. F. May			Daimler-Benz			X	CM
		Mr. G. Zedler		DBP Telekom			X X	CM

Australia	Mr. M. Biggar		AOTC Labs			X X	CM
		Mr. G. Smith		AUSSAT				X X

Belgium	Mr. O. Poncin		RTT Belgium			X X	CM
		Mr. B. Voeten		Bell Telephone		X

Canada     	(advised that no one can attend) 

USA		Mr. B.G. Haskell		AT&T Bell Labs		X X
		Mr. N. Randall		DIS				X X	(CM)
		Mr. A. Tabatabai		Bellcore			X	CM
		Mr. F. Tobagi		Starlight Networks		X X

France 	Mr. J. Guichard		CNET				X X	CM

Italy		Ms. L. Conte		CSELT				X X	CM

Japan  	Mr. S. Okubo		NTT				X X	Chairman
		Mr. K. Sakai		Fujitsu 			X X
		Mr. Y. Takishima		KDD				X 	(CM)
		Mr. T. Tanaka		NTT				X X	CM
		Mr. K. Yamazaki		KDD				X
		Mr. T  Yukitake		Matsushita			X X

Norway		Mr. G. Bjoentegaard 	NTA				X X
		Mr. H. Sandgrind		NTA				X	CM

Netherlands	Mr. D.A. Schinkel		PTT Research   		X	CM

UK      	Mr. I. Parke		BT				X X
		Mr. D.G. Morrison		BT				X X	CM

Sweden		Mr. H. Brusewitz		Telia Research		X X	CM
		Ms. C. Verreth		Telia Research		X X

Korea		Mr. J-H. Jeon		Korea Telecom		X X	Observer
		Mr. J-Y. Nam		ETRI				X X	Observer

							  CM: Coordinating Member
							(CM): Substitute for CM