CEPT T N 31-02 E-1988 Multipoint International Videoconference System《多点国际视频会议系统》.pdf

《CEPT T N 31-02 E-1988 Multipoint International Videoconference System《多点国际视频会议系统》.pdf》由会员分享，可在线阅读，更多相关《CEPT T N 31-02 E-1988 Multipoint International Videoconference System《多点国际视频会议系统》.pdf（7页珍藏版）》请在麦多课文档分享上搜索。

1、 T/N 31-02 E Page 1 Recommendation T/N 31-02 (Edinburgh 1988) MULTIPOINT INTERNATIONAL VIDEOCONFERENCE SYSTEM Recommendation proposed by Working Group T/WG 14 “Network Aspects” (NA) Text of the Recommendation adopted by the “TelecommunicatioMs” Commission: “The European Conference of Postal and Tele

2、communications Administrations, considering that there is an increasing demand for videoconference services, and especially international videoconferencing ; such services consisting of basic audiovisual facilities and a variety of optional facilities according to Recommenda- tion T/N 31-01, recogn

3、ising the need for interconnectability between three or more videoconference rooms in different countries and in different O customers premises, recommends to the members of the CEPT that they adhere to the specification for a multipoint videoconference system as contained in the Annex to this Recom

4、mendation.” . Edition of February 28, 1989 F M 2326414 0006125 5 T/N 31-02 E Page 2 1, GENERAL REQUIREMENTS A Multipoint Control Unit (MCU) is a piece of equipment located in a node of the network (terrestrial or satellite) which receives several (maximum seven) 2 Mbit/s channels from access ports (

5、each access port corresponding either to a local or a remote codec, or to another MCU) and, according to a certain criterion, causes some of them called selected channels to be distributed towards the connected studios (see Figure 1 The basic functions of the MCU for a terrestrial or a satellite net

6、work are identical. The MCU shall have the capability: - to synchronise the incoming streams to a single 2048 kHz pilot clock, - to extract frame alignment from TSO in order to synchronise the different streams to the frame clock, to extract fiam parity, multiframe and supermultiframe alignment from

7、 TS2 in order to access in each incoming stream the codec-to-codec signalling channel, (T/N 3 1-02). - to process this signalling channel, - to process the sound channels in order to create an open sound system, in the case of an unencrypted - to decide image switching and dispatching according to a

8、 selection criterion (automatic or on request), - to signal in advance the switching decisions to the codecs such that degradation during and after the - to multiplex the selected video channels with the open sound channel and the effective data channels, - to distribute the reconstructed streams to

9、 the corresponding access ports. system, switching can be avoided, 2. SYNCHRONISATION OF BIT STREAMS 2.1. 2,2. Clock Synchronisation All incoming bit streams to the MCU must be derived from the same basic 2048 kbit/s clock. If no codec involved in a multipoint conference resides in a synchronous net

10、work, i. e. no signal is received with bit 8 in TSO of odd frames set to zero, then the MCU acts as a master clock source. Such an MCU should have a reference clock with a short term accuracy of 1 in 10E9, in order to avoid frame slips during a conference session. If one or more codecs are in synchr

11、onous networks Bit 8 =O then their clocks are taken as the master. In both cases the MCU sets on all outgoing channels Bit 8 in odd TSOs to zero. Frame Synchronisation The MCU has the following functions: i) Extract from TSO frame alignment and generate the frame clock. Frame parity should not be ex

12、tracted from TSO as it is not transparently transmitted through some networks. ii) Extract from TS2 multiframe and supermultiframe alignment and generate: frame parity, multiframe clock, supermultiframe clock. iii) Synchronise the bit streams at the PCM frame rate, such that switching can be perform

13、ed without interrupting the G.732 frame structure. 3. MCU AND CODEC USE OF TS2 ODD FOR MULTIPOINT CONFERENCE APPLICATIONS The bits are encoded according to CCITT Recommendation H.130 (Parts 1 and 2) and its subsequent additions. A majority decision of 5 out of 8 is used to provide resilience to chan

14、nel errors with respect to the signals in bits 3 and 4. Bits 1, 2, 6, 7, are transparently transmitted by the MCU. Bit 8 gives multiframe and supermultiframe alignment and recovery of frame parity. Bit 3 is for Coding Mode Identification. Bit 3.l.c indicates the facilities offered by the codec (set

15、to 1 if provided) and are fixed for each codec. The MCU should take account of these bits in order to set up a minimum operating mode for all the codecs involved in the conference. For each individual port on the MCU a logical AND is made between the incoming signals from all other ports. The result

16、ant signal is then used as the outgoing signal for that specific port, the rule being that an individual ports facilities bits should not be echoed back. 3.1. 3.2. 3.3. Edition of February 28, 19- W 2326414 000bL2b 7 I I T/N 31-02 E Page 3 Bit 3.1.0 Graphics Mode 1 Bit 3.1.1 High quality speech Bit

17、3.1.3 Encryption Bit 3.1.4 System H Bit 3.1.5 Graphics Mode 2 Bit 3.1.6 Notes. i) MCUs not equipped to mix G.722 audio will set Bit 3.1.1 to zero. ii) The use of bit 3.1.3 for encryption is under study. Bits 3.1.2 and 3.1.7 Bit 3.1.2 Bit 3.1.7 O O 2 Mbit/s working only 1 O 1 Note. If the bit rate si

18、gnalled by bits 3.1.2 and 3.1.7 exceeds that available at the codecdigital interface then themeaning of the facilities bits is as follows: with codecs having a 1.5 Mbit/s serial interface - O O Never occurs 1 O Means 4 x 384 kbit/s working only O I Means 2 x 384 kbit/s working only 1 with codecs hav

19、ing a 2 Mbit/s serial interface, but effective rate of 768 kbit/s - O O Never occurs 1 O Never occurs O 1 Means 2 x 384 kbit/s working only 1 1 Means 2 x 384 kbit/s working only Bits 3.3 (colour transmission) and 3.5 (split-screen display) are transparently transmitted by the MCU. Bit 3.7 Fast Updat

20、e Request (FUR) When set to 1, the transmitter buffer occupancy is forced to decrease and stabilise to a state of less than 6K by preventing coded picture elements entering the buffer. Bit 3.9 Freeze Frame Request (FFR) Used to warn a decoder that its received signal may be interrupted after the sta

21、rt of the next supermultiframe for a period of no more than 2 seconds. On receipt of bit 3.9 set to 1 a decoder will normally “freeze” the contents of its frame store for 2 seconds or until a field start code is received with Bit A set to 1 see CCLTT Recommendation H. 120, part i. Both bits 3.7 and

22、3.9 should pass transparently across an MCU if set on an incoming signal: this is to allow for multipoint conferencing using distributed MCUs. Bit 3.1l.c indicates the power of the sound channel, integrated during 16 ms (period of the supermultiframe) and encoded with 8 bits. It is only used with en

23、crypted multipoint, otherwise it is set to zero. The MCU can use this bit to select the New and Previous Speakers channels (see paragraph 5,). Bit 3.13: Data Distribution When a codec receives this bit set to one, it must vacate in its transmission channel the same time slots which are vacated with

24、respect to the video signal in its receive channel and which are signalled by bits 4.1, 4.5, 4.7. The MCU uses this bit to ensure data continuity during a conference (see paragraph 9.). Bit 3.15: Loop Detection Bit 3.15 may be used by the MCU to detect whether one of its bidirectional 2 Mbit/s ports

25、 has been externally looped. It is necessary to monitor this condition since instability may result from such a configuration. The definition of bit 3.15 is as follows: Codecs set bit 3.15 to 1 in their outgoing paths. MCUs use a number of consecutive bit 3.15s to transmit a serial random bit stream

26、 of length n, repeatedly. If the reeeived bit sequence is the same as the transmitted random serial sequence then a loop has been detected. It should be noted that the received bit sequence may exhibit a phase delay compared with the transmitted sequence. The details of the random sequence need not

27、be rigidly specified as the sequence is oniy relevant when an individual MCU is in a looped configuration. However precautions must be taken to avoid false loop detection. This is likely to occur when two or more MCUs are connected together or when the transmission medium is subject to errors. A num

28、ber of recommendations are given below. Spare - set to zero O 1 1 2 Mbit/s and 4 x 384 kbit/s working only 2 Mbit/s and 2 x 384 kbit/s working only 2 Mbit/s and 4, 3, 2 x 384 kbit/s working 1 Means 4, 3, 2 x 384 kbit/s working only Edition of February 28, 1989 - -. . . . . . f L I 2326434 0006327 9

29、T/N 31-02 E Page 4 The length of the transmitted random sequence n should be sufficiently large to avoid duplication when two or more MCUs are connected together. It is suggested that the total length be in excess of 15 bits, thus the possibility of duplication is less than 1 in 65536. The sequence

30、transmission and detection mechanism should be sufficiently resilient to channel errors. This can be achieved in a number of ways, two simple methods are suggested here. First, considering the sequence as a number of individual bits, each bit may be transmitted for 8 consecutive bit 3.15s. The recei

31、ver takes a majority 5 from 8 vote as the received sequence bit. Thus to transmit a single sequence takes 8-kn bits. This is similar to the method adopted for bits 4.n. Alternatively, as the random sequence is transmitted repetitively, the decision as to whether the port is in the looped state or no

32、t is taken only when a number of sequences has been received. 3.4, Bit 4 is for Time Slot Allocation If the following bits are set to 1: Bit 4.1 Bit 4.3 Bit 4.5 Bit 4.7 Bit 4.1 1 Graphics transmission Bit 4.13 When a codec receives any of the bits 4.3/5/7 set to one, and bit 3.13 is set to 1 (see pa

33、ragraph 3.3.), is also vacates the corresponding time slots in its transmitted stream and sets to one the corresponding bits 4.b in its transmission channel. Bit 4.1 is transmitted transparently by the MCU as the MCU cannot switch half time slots, i. e. the MCU takes no action. Bits 4.9 and 4.15 are

34、 used for bit rate signalling: Bit 4.9 Bit 4.15 O O 2 Mbit/s operation 1 O 4 x 384 kbit/s 1 1 3 x 384 kbit/s O I 2 x 384 kbit/s At 5 x 384 kbit/s At 4 x 384 kbit/s At 3 x 384 kbit/s At 2 x 384 kbit/s The MCU should take account of bits 4.9 and 4.15 in order to set up a minimum operating mode for all

35、 the codecs involved in the conference. For each individual port, Bits 4.9 and 4.15 from every other port on the MCU are analysed to determine which is the lowest requested bit rate permitted by the facilities bits 3.1.2 and 3.1.7. The code for this bit rate is then used as the outgoing signal in Bi

36、ts 4.9 ands 4.15 for that specific port. Again the rule is that an individual ports bit rate facilities bits should not be echoed back. To avoid a lock up situation, the codec should be return its received bits 4.9 and 4.15 its transmit path, but should generate them independently. TS2 even is not u

37、sed for video TS16 is not used for video TS17 is not used for video TS18 is not used for video Use of error correcting code Time slots 1-15 and 17-31 Active Time slots 1-15 and 17-25 Active Time slots 1-9 and 17-25 Active Time slots 1-6 and 17-22 Active 3.5. Bit 5 carries a 4 kbit/s Message Channel

38、This bit is used to convey an asynchronous message channel at 4 kbit/s for signalling between the room and the MCU or between rooms or between MCUs. This channel is implemented according to the protocol defined in “Message Structure layers 1-6 for use in Audiovisual Communication” as currently speci

39、fied by NA3/HP4. Facilities using the message channel will be defined in Doc “Layer 7 Facilities for Videoconference”. 4. AUDIO PROCESSING Each terminal connnected to an MCU must receive a mix of the audio from all other terminals. The audio signals should be summed at the MCU without normalisation,

40、 i.e. unity gain on each channel. The introduction of dynamic mixing for the suppression of ambient noise may be included but speakers would still enjoy unity gain. Note. Not applicable with encrypted multipoint. Edition of February 28,- 2326434 0006128 O i T/N 31-02 E Page 5 5. SWITCHING DECISION C

41、RITERIA The criteria for switching to some extent depends on the philosophy of the multiconference service in each administration. Any solution, automatic or manual, can be inplemented without altering the basic arrange- ment of the MCU. The minimum working mode or automatic mode is as follows: the

42、MCU, by comparison of the incoming sound channels or, in case of encrypted sound channels, by the means of the sound power bit (bit 3.11 in TS2 odd), selects the loudest speaker (called New Speaker or NS). A second channel selected by the MCU, being the previous loudest speaker (called Previous Spea

43、ker or PS). The NS is sent the PS channel and the other rooms are sent the NS channel. This mode is always used when the multiconference is established. Details of the switching criterion with respect to sound levels, hangover time etc. are under study. Five manual overrides are currently identified

44、: (a) The system remains automatic but one location is considered to be the chairman of the conference. Participants are able to transmit a “request for the floor” to either the chairman or all rooms. At an appropriate time, the chairman orally gives the floor to the requesting conferee who, as he b

45、egins to speak is automatically selected as the NS. (b) One location (e. g. NS or chairman or VIP) is able to choose the ailocation of the second selected channel (normally the PS channel) by transmitting a request to the MCU. (c) Each location has the choice between the channels, which can be made

46、available by the MCU connected to that location without affecting the displays of other locations. (d) Complete manual chairman control with no voice detection. (e) Manual forcing, where one location may force the MCU to regard his port as the NS. This override is known as Visualisation forcing. It

47、may be used in one of two cases: i) where a Chairman or VIP wishes to be seen without interruption, ii) where a terminal is using a graphics camera, but is not equipped with a graphics capable codec. Only the automatic mode does not require the use of the message channel in bit 5. Modes (a), (b), (c

48、), (d) imply the use of the message channel and extra control equipment (push-buttons, lights, signalling and data connections to the codec .) at the conference room. Mode (e) normally makes use of the message channel, but an interim solution is available on a National basis (see section 7.1 .). 6.

49、MCU PROCEDURE FOR SOURCE SWITCHING Once the switching decision is taken, either by monitoring the audio levels or by the message channel, the MCU must prepare the connected codecs and operate as following: i) it sends a FFR (bit 3.9) to all the codecs which will be affected by the switch, via the selected transmission channels connected to them, ii) it performs image switching whilst maintaining the basic G.732 frame structure continuity in the selected channel(s), iii) it waits for- at least 32 ms to allow sync recovery in all decoders, iv) it sends a FUR (bit 3.7) to the