&%PAGE&

-  &%page&  -
Annex 1 to Doc. AVC-317R
Annex 1 to Doc. AVC-317R

Documents for the seventh meeting of the Experts Group  
1-10 July 1992, New Jersey and Rio de Janeiro


Normal Documents

	Note: Contributions with "*" have also been sent to MPEG for consideration at 
the joint sessions.  Some documents are also registered directly through the 
MPEG channel, which are not marked with * but have been considered at the 
joint sessions.  See AVC-318R. 

AVC-256R REPORT OF THE SIXTH MEETING OF THE EXPERTS GROUP FOR 
         ATM VIDEO CODING IN STOCKHOLM/HAIFA - PART I (CHAIRMAN)
AVC-257R REPORT OF THE SIXTH MEETING OF THE EXPERTS GROUP FOR 
         ATM VIDEO CODING IN STOCKHOLM/HAIFA - PART II 
         (CHAIRMAN)

Achievements and action points obtained at the sole sessions in Stockholm 
and the joint sessions in Haifa are recorded.

AVC-258  LIST OF REQUIREMENTS FOR MPEG-2 VIDEO 
         (REQUIREMENTS GROUP)

This document lists all the identified requirements for video part of the 
audiovisual coding, providing definition, application examples and comments, 
which were updated in Haifa.

AVC-259  GUIDE FOR THE VIDEO WORK (REQUIREMENTS GROUP)

This document extracts from the general requirements listing only those 
requirements that directly impact the development of video coding standard 
and the testing thereof, which were updated in Haifa.

AVC-260  TEST MODEL 1 (TEST MODEL EDITING COMMITTEE)

A comprehensive description of Test Model 1 (TM1) is given, incorporating 
coding schemes and various experiments.  This model is used in the course of 
research for comparison purposes.

AVC-261  THIRD PROGRESS REPORT (CHAIRMAN)

This is a progress report to Working Party XV/1, covering major 
achievements obtained in the three meetings (Yokosuka, Singapore, 
Stockholm/Haifa) toward defining Recommendation H.26X "video coding in the 
ATM environments", and particular items for consideration of WPXV/1.

AVC-262  LIAISON STATEMENTS SUBMITTED BY THE EXPERTS GROUP 
         (CHAIRMAN)

This is a collection of three liaison statements that the Experts Group sent; 
two to SGXVIII and one to CCIR Ad-Hoc Group on Digital Coding.

AVC-263  MEETING REPORT (WPXV/1)

This is a report of the Working Party XV/1 meeting held in Geneva during 7-
13 May 1992.  The following items particularly concerns the Experts Group;

    -	Very low bit rate video coding for PSTN and mobile 
telecommunications,
    -	SGXV endorsement of collaborative work between CCITT EG and MPEG,
    -	Request for inputs on likely applications of B-ISDN,
    -	Handling of MPEG "system" in the framework of AV Recommendations 
(H.200).

AVC-264  RULES FOR PRESENTATION OF CCITT|ISO/IEC COMMON TEXT 
         (CCITT)
AVC-265  SOME HELPFUL INFORMATION TO EDITORS OF CCITT/ISO COMMON 
         TEXT (CCITT SECRETARIAT)

These two documents provide "presentation rules" for documents which are 
intended to be both CCITT Recommendations and ISO or ISO/IEC International 
Standards.  These rules are called fictitiously Recommendation A.1000 and 
ISO/IEC 0001. 

AVC-266  NETWORK PLANNING FOR VIDEOTELEPHONY & VIDEOCONFERENCE 
         SERVICES (CCITT SGXII)

Current status of study is given for the effect of end-to-end delay 
(transmission and processing delay) on the communication quality.  Revised 
G.114 recommends 400ms as the upper limit of acceptable one way end-to-end 
delay, recognizing that this limit will be exceeded in some exceptional cases 
(including videotelephony over satellite circuits).  A summary of various 
evaluation results is also given which addresses the subject in audio only 
as well as audiovisual communications, indicating that highly interactive 
tasks require less than 400 ms pure delay (without echo) but that the 
effects of pure delay on speech quality appears to modestly increase as the 
delay is increased, hence that further study is necessary. 

AVC-267  BIT RATE STATISTICS OF A TV DISTRIBUTION CODEC 
         (RTT BELGACOM)

Using a hardware hybrid coder (without motion compensation), several VBR 
statistics are collected for TV programmes in ATM environments; cell inter-
arrival times, cell rate for unbuffered packetization, cell rate for slice 
buffered packetization, and bit rate for image buffered packetization.  It is 
suggested that data may be used for estimating statistical multiplex gain 
and determining the policing function for TV codecs.

AVC-268  MULTIMEDIA MULTIPLEX AND NEGOTIATION METHODS (JAPAN)

Three design examples of the interface between the video codec and the 
media control layer are discussed considering cell loss resilience and H.221 
multiplex requirements.  Negotiations for the communication mode in B-ISDN, 
including cases of uni-directional and non-symmetrical transmission, are also 
discussed from the view point of possible negotiation channels; outband, 
dedicated channel, inband.

AVC-269  SEGMENTATION TRICK TO IMPROVE STRUCTURED PACKING 
         EFFICIENCY OF CODED SIGNALS (JAPAN)

An improvement for structured packing method is presented with 
experimental results where truncation of transmission unit takes place if 
more than a certain number of cells are already filled and if the remaining 
capacity of the current cell is less than a certain value.  This method 
provides a better trade-off between transmission efficiency and the size of 
area damaged by a cell loss.  It is noted that the cell packing structure 
requiring cell boundary information must be standardized, while the 
segmentation algorithm is not a matter of standardization. 

AVC-270  CONSIDERATION OF LAN (JAPAN)

Difference between B-ISDN and LAN are reviewed with respect to network 
performance (delay, transmission error, clock).  Impact of LAN on H.32X 
terminals is also discussed.

AVC-271  MERIT OF VBR TRANSMISSION (JAPAN)

Simulation results are provided which indicate that VBR with sliding window 
policing of average rate can benefit a coding structure with periodical 
variation of the coded bit generation like MPEG1.  It is shown that the best 
benefit is obtained if the window size is exactly an integer multiple of 
coding structure parameter N. 

AVC-272  IMPROVEMENT OF DELAY UNDER AVERAGE BITRATE CONSTRAINT 
         (JAPAN)

Transmission delay is analyzed for CBR and VBR using an information 
generation model.  Sliding window and leaky bucket are compared as average 
rate UPC methods for VBR assuming the same mechanism is applied to the 
coder for preventive policing.  It is concluded that leaky bucket VBR can 
provide shorter delay than CBR, but sliding window VBR can not except for a 
particular case of periodical variation of information generation.

AVC-273  CLOCK RECOVERY FOR VIDEO (JAPAN)

It is first discussed whether video clock recovery be carried out in AAL or 
video codec, suggesting that the latter method can be more flexible for 
multimedia multiplexing.  Then, three alternative methods of clock recovery 
in the video codec are compared; buffer control, stuffing and frequency 
counting methods.  The frequency counting method is suggested from 
adaptability to VBR services.  It is also pointed out that the clock 
recovery should be studied for the case where a common network clock is 
not available at both the encoder and the decoder.

AVC-274  PICTURE QUALITY COMPARISON OF FORMAT CONVERSION THROUGH 
         / NOT THROUGH SCIF (JAPAN)

Conversion between CCIR 625/50 and 525/60 signals is experimented for three 
different schemes; two schemes use SCIF in the intermediate of conversion 
and one scheme uses direct conversion.  It is concluded that conversions 
with small delay give similar picture quality regardless of being through or 
not through SCIF. 

AVC-275* SCENE CHANGE HANDLING IN LOW DELAY MODE (JAPAN)

Scene change which requires a large number of bits per picture and 
consequently additional buffering delay is discussed with TM experimental 
results, comparing a strategy which assigns almost constant number of bits 
per picture and a strategy which allows occasional picture dropping.  The 
second strategy needs definition of skipped pictures in the coder and the 
decoder.

AVC-276* COMPARISON OF PREDICTION METHODS OF THE LOW DELAY MODE 
         OF TM1 (JAPAN)

Three experimental results are reported for the TM1 low delay mode; the 
first experiment compares four frame-based prediction methods indicating 
FAMC gives the best SNR for M=1, the second experiment compares pure field 
predictions with frame-based ones indicating that "M=3 pure field" gives 
better SNR than "M=1 frame/field adaptive", the third experiment shows that 
half-pel MC referring to the reconstructed pictures gives slightly better 
SNR than the one referring to original pictures. 

AVC-277  SIMULATION RESULT ON COMPATIBILITY (JAPAN)

Simulation results for the two compatibility experiments are given; 
"prediction from the base layer" and "prediction for the prediction error".  
It is concluded that compatible type switching defined in the TM1 syntax is 
effective for the fist scheme and that the prediction from the base layer 
provides better performance than the prediction for the prediction error 
even after having been improved by remaking the prediction from the lower 
layer depending on the upper layer prediction mode.   

AVC-278* H.261 COMPATIBILITY REQUIREMENT (JAPAN)

After analyzing several H.26X/H.261 interworking scenarios (point-to-point, 
multipoint, multicast and database access), requirements for the H.26X 
decoder and coder are concluded as follows;

    -	The decoder in B-ISDN should be able to receive both of H.26X 
bitstream and H.261 stream, but one at a time, in all the situations 
including point-to-point, multipoint, multicast and database access. 
    -	H.26X encoder should be able to transmit either of H.26X bitstream or 
H.261 stream at a time for point-to-point communications or multipoint 
communications where a common operating mode is used or where MCU 
carries out transcoding.  It should also be able to transmit two bit 
streams simultaneously for multipoint communications where MCU does 
not provide transcoding or where a mesh connection is employed, 
multicast communications and in the database.

AVC-279* CELL-LOSS COMPENSATION SCHEME (JAPAN)

Structured packing (use of macroblock start pointer and absolute block 
address in the cell header), leaky prediction and concealment at the 
decoder are simulated for TM1 IP mode (M=1, N=150) and low delay mode, 
concluding that the scheme is effective, particularly for cell loss sensitive 
sequences.

AVC-280  SIMPLIFICATION OF FAMC (JAPAN)

Simplification of FAMC in motion compensation is experimented for address 
generation and interpolation to show that no or only slight degradation in 
SNR is incurred.  Simplification of motion estimation is also experimented 
for a two-step search comprising of frame-based ME and FAMC ME.  As a 
conclusion, simplified FAMC43 is proposed for prediction.

AVC-281  COMPARISON OF REQUIRED PROCESSING BETWEEN FRAME/FIELD 
         AND FAMC (JAPAN)

The number of operations per second is estimated for adaptive field/frame 
and simplified FAMC as in AVC-280 with respect to addition and 
multiplication.  FAMC requires 60% operations of the adaptive field/frame, 
but multiplication is involved and maximum memory bus load is doubled.

AVC-282  CODING/DECODING DELAY AND REQUIRED MEMORYS FOR THE 
         PREDICTION IN TM1 (JAPAN)

Necessary number of field memorys and coding/decoding delay are estimated 
for the pure field, the frame-based field structure and the frame-based 
frame structure.

AVC-283  HARDWARE VERIFICATION OF H.26X SPECIFICATIONS (JAPAN)

Study items to prepare for the hardware verification of H.26X are listed for 
discussion.  It is pointed out that the Experts Group need a policy to cope 
with those items surrounding the coding algorithm, which are not subjects of 
verification but required to test H.26X.

AVC-284  EXPERIMENT ON LOW DELAY MODE (JAPAN)

TM1 low delay mode is first compared with frame-based {N=15, M=1} in coding 
efficiency, concluding that there is no significant loss of coding efficiency 
by use of the low delay mode.  Then, it is pointed out that the current 
coding control method for the low delay mode causes visible moving of INTRA 
slices, proposing a modified coding control method to mitigate the effect.      

AVC-285  TM1 PURE FIELD CODING SIMULATION RESULTS 
         (PTT RESEARCH - NETHERLANDS)

Pure field coding {N=12, M=3} is compared with the adaptive frame/field 
coding.  It is concluded that about equal performance is obtained in picture 
quality.

AVC-286  TM1 PYRAMID CODING AND COMPATIBILITY VERSUS SIMULCAST 
         (PTT RESEARCH - NETHERLANDS)

A two layer compatible coding (1.5 Mbit/s for base layer and 2.5 Mbit/s for 
enhanced layer) as defined in Appendix G of TM1 is compared with simulcast 
(2.5 Mbit/s).  Prediction for the prediction error is used for odd numbered 
fields of CCIR-601 signals.  It is concluded that an obtained gain of 0.5 to 
0.7 dB in SNR and other factors (full compatibility, scalability and error 
resilience) favors this hierarchical structure.  

AVC-287  ATM CELL LOSS EXPERIMENTS WITH TM1 
         (PTT RESEARCH - NETHERLANDS)

A two layered coding using pure field based coding with upsampled MPEG1 
prediction error for odd numbered field prediction error is compared with 
one layered coding in cell loss resilience.  It is concluded that layered 
coding can be very useful for transmission over ATM-networks when the base 
layer bitstream can be sent over a guaranteed channel, pointing out that 
more study and information is necessary for multi-layered coding. 

AVC-288  ERROR SENSITIVITY OF THE TM1 SYNTAX 
         (PTT RESEARCH - NETHERLANDS)

Error sensitivity of TM1 syntax is discussed in resynchronization of 
bitstream, FEC, forced updating, layered coding, and picture slicing.  A 
specific proposal is made to add two bit identification fields after each 
extension start code.

AVC-289  PSTN VIDEO CODING ON 8 AND 16 KBIT/S USING A DOWN 
         SCALED H.261 (PTT RESEARCH - NETHERLANDS)

Simulated and hardware processed pictures are presented which have been 
obtained using H.261 or its subset (without MC) operating at QCIF, 8-16 
kbit/s.  It is concluded that a video bitstream conforming to H.261 can be a 
step towards interworking of PSTN videophone service with ISDN one.  It is 
pointed out that call set up duration, end-to-end delay, spatial and 
temporal resolution, and interworking between PSTN and ISDN are items of 
further study.

AVC-290  RACE 2072 MOBILE AUDIO VISUAL TERMINAL; PROJECT 
         INFORMATION (PTT RESEARCH - NETHERLANDS)

The objectives are to find a powerful audio and video coding algorithm for 
transmission of moving and still video in a mobile environment and to 
implement these algorithms on a demonstrator.  Technical approach, key 
issues, expected impact and participants are also presented.

AVC-291  LOW-DELAY CODING EXPERIMENT (BELLCORE - USA)

Three coding structures are compared in terms of SNR and buffering delay; 
A(N=15,M=3), B(N=15,M=1), C(N=150,M=1 with forced intra-slices).  Simulation 
results show that low delay mode defined in TM1 reduces the buffering delay 
by 2 to 3 frames, but at the same time it also degrades the picture quality 
by as much as 1.84 dB.  It is concluded that B and C performs very similarly, 
but C gives slightly larger buffering delay than B due to that it encodes 28 
more intra-slices at the first 15 frames.

AVC-292  FREQUENCY SCANNING AND ENTROPY CODING USING MUVLC 
         (HHI - FRG, RTT BELGACOM, SIEMENS GERMANY)

In this contribution, UVLC technique is mentioned, an improved version MUVLC 
concerning coding efficiency is described, a proposed syntax addition to 
TM1, C-programs for bitstream encoding and decoding a slice of macroblocks, 
and results on coding gain are given.

AVC-293  TM1 COMPATIBILITY EXPERIMENTS (UK)

Use of compatible prediction from the base layer is compared with simulcast 
in TM1 to conclude that it can improve the SNR picture quality.

AVC-294  A PROPOSAL FOR AAL TYPE 2 (BELGIUM, GERMANY, ITALY, 
         NETHERLANDS, NORWAY, SWEDEN, UK)

Requirements for AAL to support VBR video transport are listed; efficient 
usage of bits, low delay, extendibility, cell loss detection, bit error 
protection, compatibility issues, multipoint and multimedia considerations, 
end-to-end timing and multi-layer synchronization, and interleaving.  A 
generic AAL Type2 and three extensions thereof are proposed to stimulate 
discussion.  It is also pointed out that VBR will most benefit low bit rate 
video coding in picture quality and delay. 

AVC-295  ISSUES CONCERNING THE SUPPORT OF FLEXIBLE SPATIAL 
         RESOLUTIONS (AUSTRALIA)  

Necessary side information and encoder/decoder implementation are discussed 
for two possible solutions of the flexible spatial resolution approach; to 
encode the signal as being input and to pad the maximum picture size.  It is 
concluded that the first alternative is more appropriate from the view 
points of transmission efficiency and effective use of codec processing 
power.  It is also pointed out that a flexible width stripe be adopted to 
support the flexible spatial resolution approach. 

AVC-296* CELL LOSS CHARACTERISTICS FOR STATISTICALLY MULTIPLEXED 
         VIDEO SOURCES (AUSTRALIA)

Longer term characteristics of cell loss are analyzed for multiplexed VBR 
sources assuming a simplified time correlation model of of the video sources 
and a multiplexing process.  Average CLR, average congestion time, 
congestion time distribution and average CLR during congestion are 
estimated.  The results show that bursts of cell loss can last several frames 
up to tens of frames and that CLR can be of the order of 0.01 during the 
congestion. 

AVC-297* THE ATM ADAPTATION LAYER FOR VIDEO SERVICES IN THE 
         B-ISDN (AUSTRALIA)

This document gives an overview for the AAL for transport of video 
services.  Methods of multimedia multiplexing are described, presenting 
features of ATM layer multiplex.  Required functions of AAL Type 2 are 
listed with an example of specific fields.  It is also pointed out that cell 
loss experiments should incorporate some assumptions regarding AAL such as 
shown in this document.

AVC-298* ADAPTING MPEG1 VIDEO FOR ATM TRANSMISSION (AUSTRALIA)

Experimental results are provided for transporting MPEG1 coded video over 
B-ISDN.  Use of AAL error detecting functionality, dividing picture types 
into high and low cell loss priority, and varying the size of slice are 
addressed.

AVC-299  SCALABLE CODING ARCHITECTURE (AUSTRALIA)

A scalable encoder architecture is proposed which uses 4x4 DCT for the 
second layer, 2x2 DCT for the third layer and prediction for the prediction 
error.  This scheme incorporates multiple independent loops to give quality 
control of all layers and allowing the rate control to be applied to each 
layer.  In order to prevent loss of coding efficiency, independent motion 
vectors and coding modes for each layer are introduced.  It is proposed 
that the features discussed in this document be included in the scalability 
core experiments.

AVC-300* PROPOSAL FOR CELL LOSS CORE EXPERIMENTS ON LAYERED AND 
         NON-LAYERED CODERS (AUSTRALIA)

A proposal is made to define a cell loss core experiment by incorporating 
short term and long term cell loss characteristics, separation of high and 
low priority information, reference AAL, use of ATM cell loss priority, use 
of scalability syntax extension.  The aim is to compare non-layered and 
layered coding structures.

AVC-301  FLEXIBLE ENCODER DEFINED PREDICTIONS (NTR - NORWAY)

A methodology to unify many prediction modes is proposed whereby the 
encoder is able to specify and down load predictors.  It is characterized as 
follows;

   1)	The predictors are not fixed in the standard but may be optimized for 
each application.
   2)	The method gives room for almost all the known predictors like frame 
based, field based and FAMC.
   3)	It gives room for defining new predictions also after the standard is 
fixed.
   4)	The predictors may be down loaded on the sequence level (or picture 
level).
   5)	Only one vector is used for each macroblock for P-frames/fields.
   6)	The implementation complexity is comparable with traditional half-
pixel motion prediction. 

AVC-302  SIMULATION WITH FIELD CODING AND M=1 FOR LOW DELAY 
         (NTR - NORWAY)

Simulation results are given for the TM1 low delay mode (field coding) in 
terms of SNR and buffering delay.  The definition of prediction follows the 
"encoder defined prediction" method proposed in AVC-301.  
 
AVC-303  SIMULATIONS WITH VERY LOW BITRATES - (8-16) KB/S 
         (NORWAY)

Simulation results are presented for a hybrid DCT scheme operating at 8-16 
kbit/s and 32 kbit/s with the picture format QCIF.  The average picture 
frequency of 8-10 Hz is reported.

AVC-304  VERY LOW BITRATE VIDEOTELEPHONY STANDARDIZATION 
         (BELGIUM, FRANCE, GERMANY, ITALY, THE NETHERLANDS, 
          NORWAY, SWEDEN, UK)

It is claimed that videotelephony service on PSTN and mobile network be 
initiated to cover all areas of service description, infrastructure, systems, 
terminals and call control.  It is also claimed that video coding algorithm 
should interwork over various networks.  Information on relevant European 
activities is given. 

AVC-305  VERY LOW BIT RATE "H.261 LIKE" VIDEO CODING SIMULATION 
         (CNET - FRANCE)

Using a picture format of 1/9CIF and H.261 coding scheme with modifications 
in GOB structure (removal of GOB) and VLC (3-D VLC), simulations were 
carried out at 8.4 and 16.8 kbit/s.  It is concluded that only 5% improvement 
in coding efficiency is obtained by those modifications and that the picture 
quality is not acceptable if the size of display is greater than 3-4 inches.

AVC-306  TM1 PYRAMID CODING FOR THE SCALABILITY REQUIREMENT
         (PTT RESEARCH - NETHERLANDS)

A proposal is made for a compatible scalable codec structure which uses 
prediction for the prediction error, pointing out that the structure can be 
extended for CTV/HDTV compatibility and that it can form a starting point 
to reach a common solution for the compatibility and scalability 
requirements.

AVC-307* VBR MPEG BIT-RATE CHARACTERISTICS (DAVID SARNOFF - USA)

This document provides detailed statistics for VBR operation of MPEG1 coder, 
which have been collected through simulation of more than 3000 frames of 
video;

   1)	frame-by-frame bit rate, probability density function and 
autocorrelation plots for each test sequence,
   2)	SNR versus average and peak rates,
   3)	effect of different peak rate control levels on the image quality,
   4)	comparison of CBR versus VBR image quality for the same average bit 
rate.

Cell loss resilience and its impact on VBR mode are also discussed.

AVC-308* ERROR CONCEALMENT FOR MPEG VIDEO OVER ATM 
         (DAVID SARNOFF -USA)

Simulation results for cell loss concealment of MPEG1 coded video are 
reported where CRC in the transport-level detects cell errors and identify 
the spatio-temporal position of MPEG slices, replacing errored DCT 
coefficients with EOB for low priority errors, replacing each errored 
macroblock with an estimate from the spatio-temporally surrounding 
macroblocks for high-priority errors (or for one-layer transmission).  It is 
concluded that a reasonable quality of restructured pictures can be 
obtained on one-tier transmission media with packet/cell loss rate as high 
as 0.1%.

AVC-309  VIDEO TRANSMISSION OVER A RADIO LINK USING H.261 AND 
         DECT (BT - UK)

This document describes an adaptation method of H.261 video codec to a 
radio link.  The improved error resilience comprises of FEC, interleaving of 
the data stream and repeating corrupted blocks.  For both of video and 
radio parts, experimental hardware can be demonstrated.

AVC-310  MODIFICATIONS TO TM1 TO SUIT BROADCAST APPLICATIONS 
         (AT&T -USA)

Two proposals are made to adapt TM1 to the "tune-in" situation in broadcast 
applications; inclusion of quantizer matrices in the picture header, and use 
of leaky prediction where leak factor is transmitted for every P picture in 
the form of (1-2-n).

AVC-311  IVS ACTIVITIES (SWP XVIII/8-3)

This document provides a draft agenda for the IVS Technical Session to be 
held during 26-27 October before the Experts Group meeting.  The objectives 
are to promote further mutual understanding and to achieve an optimized 
association between networks and coding techniques.

AVC-312  LIAISON STATEMENTS (SGXVIII)

This collection of liaison statements contain the following items of the 
Experts Group's concern; possible functions for video signal support of AAL 
Types 1 and 2, support of H.320 terminals in B-ISDN, traffic control and 
usage parameter control issues.  The second item requires specific answers 
from the Experts Group.

AVC-313  NON-COMPATIBLE VS SIMULCAST VS COMPATIBLE - EXPERIMENT 
         G.3 (BELLCORE - USA)

Simulation results are compared between non-compatible at 4 Mbit/s (A), 
simulcast at 2 Mbit/s (B), compatible at 4 Mbit/s with upscaled prediction 
from HHR base layer at 2 Mbit/s (C), and HHR at 2 Mbit/s (D).  It is concluded 
that A is preferred over B, perceptual quality of C is not far off from A, 
some prediction improvements are necessary for P and B pictures, and 
perceptual difference between B and D is greater than it appears on the SNR 
values.

AVC-314  STATISTICAL ANALYSIS OF VIDEO TELECONFERENCE TRAFFIC - 
         III (BELLCORE - USA)

Analyzing a 32,000 frame traffic data produced by a H.261 codec with motion 
compensation, it is confirmed that the number of cells per frame for video 
conferences follows a gamma distribution.  It is also shown that the density 
function fits with that of a Markov chain.

AVC-315* TIMING RECOVERY FOR VARIABLE BIT-RATE VIDEO ON ATM 
         NETWORKS (AT&T - USA)

For the MPEG system (multimedia multiplex), several methods are described to 
recover timing in video systems where the transmission delay jitter may be 
substantial as in packetized networks.  It is also pointed out that for high 
quality real time audio and video display,  the stability specification of 
MPEG may not be stringent enough, requiring additional time base correction.

AVC-316  RECOMMENDATIONS FOR CELL LOSS EXPERIMENTS 
         (DAVID SARNOFF - USA)

Some core experiments are suggested, taking into account codec mode, ATM 
transport mode, ATM AAL features, ATM network scenarios, and decoder 
concealment features.



Temporary Documents

TD-1	Agenda (Chairman)
TD-2	Available documents (Chairman)
TD-3	List of tape demonstrations (Chairman)
TD-4	Generic prediction in AVC-301 (Chairman)
TD-5*	Report of small group meeting on low delay mode (Mr. Bjoentegaard)
TD-6*	Report of small group meeting on cell loss resilience (Mr. Biggar)
TD-7*	Report of small group meeting on H.261 compatibility (Mr. Parke)
TD-8	Draft meeting report for the sole sessions (Chairman)


END