&%PAGE&

-  &%page&  -
Annex 1 to Doc. AVC-65R
Annex 1 to Doc. AVC-65R

Documents for the second meeting of the Experts Group  

Normal Documents

AVC-22R  REPORT OF THE FIRST MEETING OF THE EXPERTS GROUP FOR ATM VIDEO
          CODING IN THE HAGUE (CHAIRMAN)

Achievements and action points obtained at the first meeting are recorded 
to facilitate our discussion at this meeting.

AVC-23  REPORT OF MPEG BERLIN MEETING (CHAIRMAN)

The following points of our concern are extracted from the MPEG meeting 
outcomes;

    -	joint sessions with CCITT Experts Group are to be held,
    -	MPEG Phase 2 work plan and milestones,
    -	technical requirements from the CCITT point of view are to be further 
clarified,
    -	test procedures for promising video coding algorithms.
 
AVC-24  EXCERPT FROM WPXVIII/8 MEETING REPORT (CHAIRMAN)

The following points of our concern are extracted from the WPXVIII/8 
meeting report;

    -	coordination of video coding aspects was initiated,
    -	IVS baseline document (AVC-25) will be used as a vehicle to achieve 
this coordination,
    -	some answers to our questions on the B-ISDN characteristics were 
provided with respect to network resource management and cell loss 
performance.

AVC-25  INTEGRATED VIDEO SERVICES - IVS - BASELINE DOCUMENT 
         (WORKING PARTY XVIII/8)

This living document is intended to provide a framework for ongoing studies 
on the video service aspects of B-ISDN in order to ensure consistency with 
B-ISDN among the works of different standardization bodies.  It is proposed 
that those groups have joint ownership of this document and that it be used 
as a vehicle to facilitate and promote discussion, liaison and agreements for 
IVS in B-ISDN.

AVC-26  REPORT OF THE 13TH MEETING IN BERLIN (MPEG)

This is reproduction of the complete report of the MPEG Berlin meeting held 
in December 1990.

AVC-27  EXCERPT FROM THE REPORT OF THE GENEVA MEETING 
         (WORKING PARTY XV/1)

The following points of our concern are extracted from the WPXV/1 meeting 
report;

    -	the first progress report of the Experts Group was approved,
    -	"terms of reference" of the group was slightly extended to allow 
direct interaction with other CCITT Groups,
    -	a comment was given on the balance between commonality and 
optimization of the video coding algorithm.

AVC-28  LIAISON STATEMENT TO CCITT STUDY GROUP XV (TG CMTT/2)

Common areas of interest between CCIR/CCITT TG CMTT/2 and Study Group XV 
are identified, leading to a practical collaboration method through the 
"Special Rapporteur for the second distribution of digital TV and HDTV."

AVC-29  COMMON PROGRESSIVE PICTURE FORMAT FOR HIGH QUALITY APPLICATIONS
         (BELGIUM, FRANCE, FRG, ITALY, NORWAY, THE NETHERLANDS, UK) 

It is proposed that a common worldwide picture format be adopted also for 
high quality video applications, which has the following parameter values;

     -	60 frames/second
     -	576/288/288 lines for Y/U/V
     -	720/360/360 pels for Y/U/V
     -	progressive.

AVC-30  IMPROVEMENTS OF HYBRID DCT CODING (NORWAY)

The following ideas have been tested for improving the coding performance 
of hybrid DCT coding schemes;

     -	optimized quantizers for transform coefficients
     -	less filtering for interpolating fractional pel positions
     -	adaptive use of four different two-dimensional VLCs having 1-4 bits of 
EOB
     -	prediction from two frames as alternative to interpolative coding 
     -	use of progressive images as input to the source coder

Simulation results are presented in SNR vs bitrate as well as in processed 
pictures.

AVC-31  ON BIT RATES IN A TWO-LAYER CODING SCHEME (RTT BELGIUM)

Experimental data are given to the loss of coding efficiency for a 2-layer 
coding against a 1-layer coding in terms of bit rate overhead vs stepsize 
ratio between the two layers.  VLC efficiency loss (in intra mode only) and 
impact of poor prediction due to low-accuracy loop (in inter/intra mode) are 
addressed. 

AVC-32  COMPATIBILITY METHODS FOR VIDEO CODING SYSTEMS 
         (THE NETHERLANDS, FRG, NORWAY, ITALY, UK, SWEDEN, FRANCE, BELGIUM)

Definitions are first given to "upward/downward compatibility" between 
different picture resolutions and "forward/backward compatibility" between 
different coding standards.  Then the following methods of achieving these 
compatibilities are defined and discussed; simulcasting, embedded bit stream, 
syntactic extension, switchable encoder and standard families.

It is proposed that the definitions form a basis for further discussions on 
the compatibility issue.

AVC-33  SET UP OF CCIR 601 MULTI-PURPOSE CODING SCHEME 
         (PTT RESEARCH, THE NETHERLANDS)
The set up of a multi purpose coding scheme (MUPCOS) for CCIR 601 video 
signals is presented, where SIF signal for odd fields coded as MPEG-1 bit 
stream form a base layer.  The second layer transmits information for even 
fields as well as information for additional quantization accuracy of the 
SIF signal.  Simulation results are provided in the form of statistical data 
and processed pictures.

This scheme features compatibility with H.261 on hardware level and 
compatibility with MPEG-1 on video-multiplex level.  It is concluded that 
this compatibility has to be supported since the price for it is low.
 
AVC-34  FRAMEWORK FOR STANDARDISATION OF VIDEO SERVICES ON THE B-ISDN
         (AUSTRALIA)

This document identifies the long term objectives for video service 
provision on B-ISDN with its fundamental characteristics (cell based 
transport, very large capacity, great transport flexibility) considered and 
proposes a specific short term workplan to achieve these objectives, 
stressing the alignment with the B-ISDN developments.

It is also stressed to have close collaboration with SGXVIII for ensuring 
that the network developments accommodate the features required to achieve 
desired functionality for the video services.

AVC-35  ARCHITECTURE OF VIDEO SERVICE INTEGRATION ON THE B-ISDN
         (AUSTRALIA)

After analyzing what integration means and comparing several methods 
(negotiation, simulcast, layered) to achieve it on B-ISDN, "flexible layering" 
method is proposed as a framework, where as many layers as necessary, 
practical or economical for a given service are used.  It is also proposed to 
encourage other parties to adopt this framework.

AVC-36  PICTURE FORMATS SUPPORTED BY VIDEO CODECS ON THE B-ISDN
         (AUSTRALIA)

It is pointed out that an early decision of the picture formats which a 
coder will be capable of dealing with is essential for video services 
integration on B-ISDN.  The following resolution capabilities are suggested 
as a basis of this decision;

     -	videophone		��192x144 luminance, ��15Hz frame rate
     -	videoconference	��384x288 luminance, ��30Hz frame rate
     -	TV			��768x576 luminance, ��30Hz frame rate
     -	HDTV			��2048x1152 luminance, ��60Hz frame rate

AVC-37  MULTI-POINT VIDEO COMMUNICATION FOR B-ISDN (AUSTRALIA)

Early consideration of multipoint systems is proposed to exploit B-ISDN 
characteristics such as bandwidth-on-demand, multi-casting, selective 
routing, dynamic adaptation of bandwidth and time-sliced decoding and 
multimedia multiplexing.  It is proposed that these considerations should 
lead to development of appropriate signalling protocols by SGXI.

AVC-38  A COMPARISON STUDY OF VBR vs CBR FOR CONVERSATIONAL SERVICES 
         (AUSTRALIA) 

The advantages and new capabilities offered by VBR video transmission are 
first described.  Then, experimental results are given on comparison of 
required network resources between VBR and CBR, reporting that VBR has an 
overall performance gain and a significant statistical multiplex gain on a  
subjective quality basis.

It is also suggested that the VBR transmission should be supported by the 
new functionalities in signalling protocols.
 
AVC-39  VIRTUAL CHANNELS FOR MULTIMEDIA SUPPORT IN THE B-ISDN (AUSTRALIA)

It is proposed that early definition of arrangements for multimedia 
multiparty support should include the separation of multimedia connections 
into separate virtual channels.  This method is characterized by the 
separation of call and connection control, individual usage monitoring for 
virtual channels, service specific processing of the network, and easy 
support for layered coding.  A case study is given to the cross media delay 
for multiple virtual channels.

AVC-40  RESPONSE TO INTEGRATED VIDEO SERVICES - IVS - BASELINE DOCUMENT
         (AUSTRALIA)

It is recommended that the group formally endorse the IVS baseline document 
and establish a significant role in its evolution to coordinate between 
video service and network developments.  Some texts are proposed for 
updating the baseline document with respect to the video service 
interworking methods and the network issues.

AVC-41  STATISTICAL MULTIPLEX GAIN FOR VARIABLE BIT RATE VIDEO 
         (RTT BELGIUM)

It is argued that if both provides the same quality, VBR has always larger 
statistical multiplex gain than CBR since VBR is an "eroded" version of CBR.

Several factors impacting the statistical multiplex gain are listed up; codec 
adaptability, noise in source, source bit rate, link bit rate, change in image 
contents, integration periods.  It is concluded that there is no need to 
make a generic choice between CBR and VBR; CBR is a particular case of VBR. 

AVC-42  MULTIMEDIA MULTIPLEX IN ATM NETWORKS USED IN THE BELGIAN 
         BROADBAND EXPERIMENT (BELGIUM)

Multimedia multiplexing based on VCI multiplex is concluded as appropriate 
for B-ISDN because of the features; easy addition/deletion of service 
components during a call, service integration, information share among 
subscribers.  Requirements on differential delay are analyzed with respect 
to delay due to transmission and routing through the ATM network and that 
due to packetization, coding and AAL.  VP concept is concluded as not 
useful due to its quasi-static control.

A solution in the Belgian Broadband Experiment is given which is based on 
the service component concept to meet the differential delay requirement.

AVC-43  ATM NETWORK MODEL (SWEDEN, UK)

The following three ATM network models are presented which provide the 
congestion probability;

     -	Method of large deviation (as in the Annex 4 to AVC-22R)
     -	Gaussian (Normal Distribution) model
     -	Method of equivalent bursts

AVC-44  CELL LEVEL STATISTICS FOR A TWO-LAYER CODING SCHEME (UK)

Statistical data for information generation of a two-layer coding are 
presented, concluding that even for a videophone type sequence the source 
can be bursty, and that a relatively small buffer makes a significant 
difference to the shape of source. 

AVC-45  DESCRIPTION OF A COMPATIBLE CODING APPROACH FOR MPEG1 AND MPEG2
         (UK)

A compatible coding scheme are presented with coded results which codes 
CIF/SIF sized pictures and CCIR601 sized pictures.  CCIR601 picture is first 
down converted to the SIF and then coded by MPEG1 coder as the base layer, 
then the difference between the original CCIR601 signal and the upconverted 
version of the local decoded base layer signal is coded by MPEG2 codes as 
the second layer.  This scheme offers both forward and backward 
compatibilities between MPEG1 and MPEG2 (similarly H.261 and H.26X). 

AVC-46  CONSIDERATIONS ON PICTURE FORMAT IN ATM NETWORKS (JAPAN)

It is proposed that the following items should be further investigated and 
confirmed toward defining an SCIF (Super CIF);

     -	picture quality degradation due to format conversion
     -	loss of coding efficiency by use of SCIF
     -	delay due to format conversion.

Some preliminary investigation results are presented for these items.  
Comments are also given on multiple format approaches which may eventually 
be used and on the areas where SCIF be adopted.

AVC-47  CONSIDERATIONS ON STATISTICAL MULTIPLEX GAIN ACCORDING TO THE
         FIRST SIMPLIFIED NETWORK MODEL (JAPAN)

Statistical multiplex gain (SMG) curves for VBR coding are provided based on 
the first CCITT network model.  It is discussed that SMG becomes larger as 
network grows if we assume the link capacity increases accordingly, that the 
increase of SMG by allowing higher cell loss probability is rather limited 
and becomes even smaller as network grows, and that a service class which 
provides with reduced charge but with high cell loss may not be attractive.

It is stressed that subjective evaluation of the VBR effectiveness against  
CBR is important, particularly under the restriction of Usage Parameter 
Control of the network.    

AVC-48  VBR CODING UNDER USAGE PARAMETER CONTROL (JAPAN)

One of the possible methods for UPC is first described.  Then, the following 
study items are suggested as essential for the study of VBR coding under 
the UPC imposed by the network;
     -	averaging window size
     -	coding parameter control under UPC
     -	VBR framework which includes CBR as a particular case
     -	end-to-end delay.

AVC-49  SOME OBSERVATIONS ON VARIABLE BIT RATE CODED VIDEO SIGNALS 
         (JAPAN)

Observed average bit rates are provided for various sizes of sliding window; 
small (up to 0.1 frame period) and large (up to 300 frames period), 
concluding that instantaneous peak bit rate can be reduced significantly by 
a small buffer, and that some scenes show a large deviation of average rate 
but others not.  Some considerations are also given on to what cases VBR 
becomes effective.   

AVC-50  CONSIDERATIONS ON CELL LOSS IN ATM NETWORKS (JAPAN)

Requirements for cell loss ratio are first given.  Then, several methods are 
described which cope with not ignorable cell losses; particularly, cell 
interleaving which is applicable only to higher bit rate coding, 
packetization which should be used in any case for quick recovery from the 
damage and layered coding which needs evaluation of coding efficiency.
 
AVC-51  REQUIREMENTS OF THE HIGH QUALITY VIDEO CODING STANDARD H.26X -
         CONSIDERATION OF MULTIPOINT SYSTEMS, DISTRIBUTION SERVICES AND
         FUTURE ENHANCEMENT (JAPAN)

The following items are provided to discuss the requirements for high 
quality video coding standard H.26X;

     -	realization of multipoint systems, taking into account possibility that 
layered coding may be a solution,
     -	inclusion of an intraframe only video codec in the scope of the new 
standard, 
     -	consideration of the future enhancement by means of the syntax 
structure.

AVC-52  REVISION TO THE ANNEX 3 TO DOC. AVC-22R (JAPAN)

Some enhancements to the questions for SGXVIII are proposed which further 
clarify the points of our concern by adding recent findings in required cell 
loss performance, statistical multiplex gain, VBR considerations with respect 
to UPC, multimedia multiplexing, etc.   

AVC-53  INVESTIGATION OF MULTIMEDIA MULTIPLEX ON ATM NETWORKS (JAPAN)

In order to make discussion points clearer, it is suggested to first clarify 
the requirements on ATM networks from audiovisual terminal point of view.  
Required functions for an intermediate virtual "MUX" layer between "CS" 
layer and "User" layer are listed up for identifying those appropriate for 
networks.

A preliminary delay analysis is also presented for multiplexing in CS, 
pointing out that the size of CS-PDU relative to the bit rate is essential.
 
AVC-54  CLOCK RECOVERY FOR VARIABLE BIT RATE AUDIOVISUAL CODING (JAPAN)

Several clock recovery methods are discussed attempting to clarify what is 
needed for variable bit rate audiovisual coding and what is to be reflected 
in the AAL Type 2.  There are four cases; CS having no clock recovery 
function, clock recovery in one VC, clock recovery between multiple VCs 
from a terminal, clock recovery between multiple VCs from different 
terminals.
AVC-55  H.261 COMPATIBLE 2-LAYER VIDEO CODEC WITH HIGH CELL LOSS
         RESILIENCE (DAIMLER-BENZ RESEARCH, FRG)

Coding results are presented for a two-layer coding scheme where the first 
pass encodes the input original signal according to H.261 and the coded data 
packets are transmitted with high priority, while the second pass encodes 
the difference between the original signal and the locally decoded signal of 
the first pass with the identical H.261 coder and the resulting coded data 
packets are transmitted with low priority.  This scheme is compatible with 
H.261 and very resilient to cell loss in the second layer, up to 10E-1. 

AVC-56  CONSTRAINTS ON VARIABLE BIT-RATE VIDEO FOR ATM NETWORKS 
         (AT&T BELL LABS)

Conditions are analyzed with inequality equations which ensure that the 
video encoder and decoder buffers do not overflow and underflow when the 
channel can transmit a variable bit rate.  Based on this analysis, allowable 
variability in terms of number of bits per frame are given as a function of 
the network policing parameters.  The existing constant bit rate channel 
case is treated as part of this generalized analysis.  Finally there is 
described a means to select jointly the number of encoded bits for each 
video frame and the number of bits transmitted across the variable rate 
channel. 

AVC-57  PREPARATORY DRAFT PROPOSAL PACKAGE DESCRIPTION FOR MPEG PHASE 2
         (MPEG)

This is a collection of materials agreed or to be discussed for the MPEG 
Phase 2 work.  Overall plan, requirements and picture quality testing method 
for video coding candidates are described. 

AVC-58  SIMULATION RESULTS FROM A SUBBAND SPLITTING CODING SCHEME
         COMPATIBLE WITH MPEG1/H261 (CNET)

Coded results are presented for a coding scheme which encodes CCIR601 
pictures maintaining compatibility with H.261 and MPEG1 standards.  The input 
picture is first split into four subband sequences of SIF resolution by 
Hadamard transform matrixes.  The base layer for the LL component conforms 
to H.261/MPEG1, the enhancement layers transmit coded data for the 
remaining three subbands as well as additional data to the base layer.  In 
the current experiments, SM3 are used for all the four subbands.

It is concluded that more investigations are needed in order to definitively 
know the interest of such a simple method, in particular in terms of 
tradeoff between loss of efficiency and simplicity.

AVC-59  PICTURE FORMATS FOR HIGH QUALITY APPLICATIONS (CANADA)

The following two picture formats are proposed for a single format for all 
high quality applications within a region;

                              	525-line format	625-line format
Frame rate                    	30/sec         	25/sec
Interlace                     	progressive    	progressive
No. of pels (Y/Cr/Cb)         	720/360/360    	720/360/360
No. of significant pels/line  	704/352/352    	704/352/352
No. of lines/frame (Y/Cr/Cb)  	480/240/240     	576/288/288

It is also proposed to reach agreement among CCITT, CMTT and ISO to support 
these formats for high quality applications. 

AVC-60  PICTURE FORMAT FOR HIGH QUALITY INTERACTIVE VIDEO SERVICES
        (BELLCORE)

After having analyzed some fundamental requirements for selection of 
appropriate picture format(s), a common picture format with square pixel 
geometry is concluded as desirable to facilitate inter-region connections, 
to ease multipoint and bridging operations, and to realize integrated 
multimedia terminal.  The parameter values are 528 lines/picture and 704 
pels/line, 59.94 progressive pictures/sec for ATM teleconferencing.
  
AVC-61  STATISTICAL ANALYSIS OF VIDEO TELECONFERENCE TRAFFIC IN ATM
         NETWORKS (BELLCORE)

VBR coded data for a 30 minute sequence of video teleconferencing are 
analyzed to answer the questions of appropriate statistical model to 
represent coded data and appropriate video source model to be used for 
traffic studies.  The major conclusions are as follows;

    -	The number of cells per frame follows a gamma distribution, which is a 
stationary process in the absence of scene cuts and changes.
    -	For traffic studies, neither an autoregressive model of order 2 nor a 
two-state Markov chain model is good, though the former fits the data 
well in a statistical sense.
    -	A detailed Markov chain model with about 60 states is sufficiently 
accurate for use in traffic studies.

AVC-62  A SIMULATION STUDY OF VIDEO TELECONFERENCE TRAFFIC IN ATM
         NETWORKS (BELLCORE)

Using VBR coded data for a 30 minute sequence of video teleconferencing, 
cell loss and delay simulation results are provided for a case where 16 
sources are transmitted through a single stage multiplexer with buffer of 4 
ms and 5 ms delay.  It is concluded that different sources with identical 
statistical characteristics can experience widely different cell loss rates, 
and that this difference is due to the inherent periodicity of video traffic 
which can be alleviated by a simple buffer scheduling mechanism.

AVC-63  LIAISON STATEMENT TO CCITT WORKING PARTY XVIII/8 (TG CMTT/3)

TG CMTT/3 gives an update for "Annex 4 - Video service interworking" of the 
"IVS Baseline Document", indicating several interworking techniques and 
advantages/disadvantages of the layered coding.  Compatibility is also 
discussed in the light of "contribution" applications. 
  
AVC-64  ASPECTS ON PICTURE FORMAT FOR HIGH QUALITY APPLICATIONS (SWEDEN)

Several aspects are listed up which impact the decision of a new picture 
format; application, compatibility, parameters, equipment.  It is concluded 
that as much compatibility as possible shall exist among different 
applications, and that "distribution" and "future HDTV" aspects shall be 
given heavy weight in definition of a new picture format.  A principle for 
compatibility between various applications, ranging from videophone to HDTV, 
is also presented.
Temporary Documents
        
TD-1	Agenda for the second meeting in Paris (Chairman)
TD-2	Available documents (Chairman)
TD-3	Draft status report on ATM video coding standardization 
	(Experts Group)
TD-4	List of tape demonstrations (Chairman)
TD-5	Representatives to MPEG sub-groups (Chairman)
TD-6	Editors for "Status Report" (Chairman)
TD-7	Comparison of VBR/CBR coding - Framework for further study 
(Small group on VBR/CBR study)
TD-8	Action plan for ATM picture format (Small group on picture format)
TD-9	Draft liaison statement (Liaison Representative to CMTT/2)
TD-10	Liaison to SGXVIII from SGXV Experts Group on Video Coding for ATM 
Networks (Small group on network issues)
TD-11	Discussion results of the second meeting in Paris (Experts Group)
TD-12	Draft report of the second meeting of the Experts Group for ATM Video 
Coding in Paris (Chairman)


END