FR2703859A1 - Method of dynamic management of the correction capacity of a layer for matching to the ATM - Google Patents

Abstract

The method consists in interlacing Reed-Solomon coded information over a variable depth depending on the application before transmitting it in packets on an ATM network, and, on reception, carrying out a de-interlacing, being dynamically and automatically synchronised whatever the interlacing depth, before the Reed-Solomon decoding which corrects the errors and the erasures generated by transmission. Applications: high-definition television and compatibles, high-fidelity sound and compatibles, video telephony, multiplexing of all these signals (for example MPEG 1 and 2).

Description

Dynamic management method of the capacity of
correction of an ATM adaptation layer
The present invention relates to a method for dynamically managing the correction capability of an AAL adaptation layer called AAL for interactive and distributive audiovisual services, ATM being the abbreviation of "Asynchronous Transfer Mode" and AAL the abbreviation of "ATM Adaptation
Layer "in the English language The ATM technology consists of the transmission of useful data in the form of cells consisting of five bytes of header and forty eight bytes of information.

 Recommendations CCITT I.321, I.361, I. 362 and I.363 define the different layers of the reference model, their functions and their fields of application.

 The CCITT Recommendation I.362 defines four classes of services characterized by their rates, their connection modes and the temporal relationship levels between transmitters and receivers.

 CCITT Recommendation I.363 defines four types of ATM adaptation that can be used in each class of service.

 Audiovisual transmissions are characterized by very different information rates, ranging from 64 Kbitls for low-quality videophone, 256 to 512 Kbitls for HiFi sound, 2 MbiUs for videoconferencing, up to 8 Mbitls and 34 Mbitls for the distribution of standard television and high definition television. The real-time constraints inherent in the transmission of this type of signal prohibit the retransmission of the information when, in the bit stream received, errors are detected.

 In CCITT Recommendation I.363, an optional method for correction of bit errors and cell losses is proposed for type 1 AAL. Based on the use of a Reed-Solomon code (128,124) on the body of Galois GF (256) still noted GF (28) and an interleaver of order 47 this method is only applicable to distributed video services.

One of the peculiarities of this technique is that the processing time of the information is all the longer as the rate of service treated is low. This interleaving time is approximately equal to the filling time of the interleaving memory:
Tcoder = 128x47x81bit
Técéceur = 128 x 47 x 8 / debit
This characteristic proves to be a major disadvantage in the particular case of low bit rate and / or interactive applications such as high fidelity sound or video telephony for example, for which the end-to-end processing time must be as short as possible so that the quality of service is good.

 The proposals made to date at this level of the telecommunications system, to remedy this problem, are limited to reducing the size of the interleaving matrix by using a Reed-Solomon code different from that already recommended (for example RS (32). , 30) on GF (256) coupled to an interleaver of order 47: at identical rate, the processing time is here divided by four).

These proposals have the following major disadvantages:
- incompatibility between different Reed-Solomon codes:
RS (128,124) * 47 and RS (32,30) * 47 for example.

 - RS codes on GF (256) reduced therefore not used to the maximum of their yield.

 - hardware realization of VLSI components and even electronic cards offering the features of the AAL type 1 for different types of services (video, audio, videophone ...) very complex since only the height of the interleaving matrix is constant while its length and especially its RS code on GF (256) are modified.

 The aim of the invention is to overcome these defects by proposing the use of a single Reed-Solomon code on GF (256), whatever the type of application envisaged, coupled to an interleaving memory. of variable size chosen at the encoder and an automatic interleaving synchronizer at the decoder.

For this purpose, the subject of the invention is a method for interleaving data from distributed and interactive and coded audiovisual services.
Reed-Solomon to protect against bit errors and cell losses in an asynchronous time network characterized in that it consists in retaining the same Reed-Solomon code regardless of the application envisaged, to interleave these coded data RS on a variable depth fixed by the application before transmission in packets or cells on the network, to deinterlace the information contained in the packets or cells, without resorting to indications "SAR-PDU" other than those described in the recommendation 1.363 of
CCITT to synchronize the mechanism and correct by code
Reed-Solomon the errors and erasures in the returned symbols.

Other characteristics and advantages of the invention will appear below in the light of the following description made with reference to the appended drawings which represent:
FIG. 1 the different steps of the process according to the invention in the form of a flowchart,
- Figure 2 a block diagram of the interleaving mechanism for implementing the method to the transmitter according to Figure 1.

 FIG. 3 a block diagram of the interleaving mechanism allowing the implementation of the method to the receiver according to FIG.

 - Figure 4 an ATM cell segmentation format, for a type 1 adaptation according to CCITT recommendation I.363.

 - Figure 5 an organization diagram of an interleaver for implementing the method according to Figure 1 for a first example of application.

 - Figure 6 a scheme of organization of a different interleaving device for carrying out the method according to Figure 1 for another application example.

 - Figure 7 an organization diagram of a different interleaving device for carrying out the method according to Figure 1 for a third type of application.

 FIG. 8 is a detailed diagram of the ATM cell segmentation format allowing the implementation of the different interleaving devices in the same and single product.

 FIGS. 9, 10, 11, 12, 13, the performances, in terms of residual error rate after correction for given bit error and cell loss rates, of the various combinations given as realistic examples and this for different flow rates.

The method of protection against bit errors and cell losses
ATM according to the invention is illustrated by steps 1 to 12 of Figure 1. It consists of step 1 to convert each symbol or byte representative of the source signal (audio, video, video telephony, multiplex MPEG type 1 or 2. ..) in a Reed-Solomon code word as described for example on page 207 of G. Cullmann's free entitled "Detector codes and error correctors" published by DUNOD 1967 in the collection "Initiation to the novelties of the science "or in the manner described pages 304-308 of the book of
Bernard Sklar entitled "DIGITAL COMMUNICATIONS Fondamentals and
Applications published by Prentice Hall 1988. The coded information obtained form words of codes C (N, K, D) or RS (N, K, D) of Hamming distance D, of length N in number of symbols and containing
K useful information.

This information is interleaved in step 2 by means of an interleaving device arranged according to the block diagram of FIG. 2, further described, in the form of examples in FIGS. 5, 6 and 7. According to this device, the words of N symbols are stored horizontally in the interleaving memory with a capacity equal to:
C1 = N * L where L is the interleaving depth, also called interlace order, and is a function of the intended application or the correction capacity necessary to provide a good quality of service. The "SAR-PDU payload" information fields of the ATM cells are obtained by vertically reading the interleaving memory as described in the examples of FIGS. 5, 6 and 7, where the numbers set to superscript indicate the cell number of the cell. belonging in the matrix:
32 represents byte # 3 of cell # 2.

The sequence number calculation which is carried out in step 3 and their insertion leads to a cell segmentation (SAR-PDU) according to a format shown in FIG. 4 and detailed in FIG. 8 in accordance with the recommendation CCITT I.363 . The ATM cell therefore comprises 40 header bits "Header" and 384 bits of "SAR-PDU" of which:
- 4 SN sequence number bits and 4 SNP SN field protection bits against errors forming the "SAR-PDU header"
- 376 bits of information forming the "SAR-PDU payload".

 The whole forms a cell of 5 + 48 = 53 bytes. The SN field itself is divided into two parts (Figure 8), one consisting of the 3 least significant bits for a modulo 8 numbering, the other, consisting of the most significant bit, also called CSI, allowing the insertion an indication of start of interleaving matrix for example.

 In reception the unpacking of the cells is carried out in step 6. The cells are restored in the SAR-PDU format of FIG. 4 and a correction of the sequence numbers takes place in step 7 followed in step 8, by replacing the lost cells detected by checking the sequence number integrity, by "ghost" cells in the deinterleaver, and by destroying the inserted inserted cells.

Deinterleaving of the data then takes place in step 9 by means of a symmetrical deinterleaving device of the device of step 2 using a capacity memory:
C2 = L * N
Naturally, the deinterleaving operation of step 9 must be synchronized with the deinterlace matrix start indications as well as with the insertion of the ghost packets and / or the destruction of the inserted packets of step 8 so that the mechanism which consists in the vertical writing of the "SAR-PDU payload" fields of the received cells followed by the horizontal reading of the reconstructed Reed-Solomon code words is effective.

 These codes are then analyzed in step 10 to perform the error and erase corrections.

The method of the invention which has just been described applies to very different configurations of interleaving matrices and Reed-Solomon codes, provided that the number of bytes contained in the matrix is a multiple of 47 to have a whole number of fields "SAR-PDU payload" in a matrix, whatever the depth of this one, this to simplify the management of the system. It is important to note that this method can equally well be applied to different interleaving methods, for example based on horizontal writes of the codewords.
RS and diagonal readings of the tables thus made.

The following examples are based on matrices with a rectangular structure (horizontal and vertical scanning) in order to simplify the demonstration:
- RS code (47,44) coupled to matrices of depth 24 and 12
- RS code (188,172) coupled to depth matrices 24, 12, 6 and -3.

 However for maximum efficiency it is preferable to set a Reed-Solomon code as small as possible for all applications and to "play" only on the size of the interleaving matrix to adapt to the various applications supported by AAL type 1. To do this one will take the precaution of choosing a Reed-Solomon code word length on GF (256) multiple of 47 (in bytes) as close as possible to 255 (28-1). length N of the Reed-Solomon codeword which can therefore take, for a code on GF (256) the following values: N1 = 47, N2 = 94, N3 = 141, N4 = 188, N5 = 235 is chosen equal to N5 which will offer the best performance, we note that 235 = 5x47.

Examples are presented in FIGS. 5, 6 and 7 which make it possible to understand the originality of the interlacing device. In the case of ATM networks, the length R of the "SAR-PDU payload" field of the cell
ATM is equal to 47 bytes. N length of the code word RS and L depth of the interleaving matrix can therefore be chosen respectively equal to 235 and 24 (Figure 5), or 235 and 12 (Figure 6), or 235 and 6 (Figure 7) (all other value of L is acceptable, for example L = 5 ...). K, the number of useful octets in the Reed-Solomon codeword and hence D, Hamming distance of the Reed-Solomon code do not alter the interleaving principle. The choice of D will depend on several parameters such as the redundancy accepted to protect the service, the characteristics of the transmission system and the expected performance of the protection mechanism.

 For the sake of simplification of the description of the method, a realistic example is now described. The chosen parameters can obviously be modified.

 The redundancy rate generally considered acceptable for performing error correction in the area of digital telecommunications is between 5% and 10%. This is all the more acceptable as the information compression methods currently being developed make it possible to envisage useful rates of between 30 Mbitls and 40 Mbitls for compressed high-definition television signals, rates that are not as critical for broadband asynchronous network that could be the 140 Mbit / s announced in recent years.

The Reed-Solomon code chosen for this example is therefore a
RS (235,219,17), which adds a 7.3% redundancy to useful information.

It can be interleaved at different depths depending on the applications as shown in the table below where the values of R are given as examples:

<tb> Depth <SEP> Number <SEP> of <SEP> cells <SEP> Range <SEP> of <SEP> flow rates <SEP> Number <SEP> of <SEP> cells <SEP> Delay <SEP> max
<tb><SEP> correctable
<tb><SEP> R = 24 <SEP> n = 120 <SEP> 10 <SEP> Mb / s ~ 140 <SEP> Mbps <SEP> 8/120 <SEP> 4.5 <SEP> ms
<tb><SEP> R = 12 <SEP> n <SEP> = <SEP> 60 <SEP> 4 <SEP> Mb / s ~ 10 <SEP> Mb / s <SEP> 4/60 <SEP> 5, 7 <SEP> ms
<tb><SEP> R <SEP> = <SEP> 6 <SEP> n <SEP> = <SEP> 30 <SEP> 0.5 <SEP> Mb / s ~ 4 <SEP> Mbls <SEP> 2 / 30 <SEP> 22.8 <SEP> ms
<tb><SEP> R <SEP> = <SEP> 3 <SEP> n <SEP> = 15 <SEP> 128 <SEP> Kb / s ~ 0.5 <SEP> Mb / s <SEP> 1115 <SEP > 44.1 <SEP> ms
<Tb>
If for the highest rates the processing delay remains negligible (<10 ms), it becomes significant at 128 KbiSs but still significantly less than the processing time of the compression algorithms used (between 150 ms and 300 ms).

 Figures 9, 10, 11, 12 and 13 show the performance of the various examples given above for bit rates respectively of 40, 10, 4, 0.5 Mbitls and 128 Kb / s. The corresponding graphs show on the ordinate the residual error rate as a function of the incoming error rate, for RS code interleavings (235, 219) on respectively 24 lines (curve A), 12 lines (curve B), 6 lines (curve C) and 3 lines (curve D).

It is clear that if reducing the size of the interleaving matrix decreases the performance of the correction system for a given service rate, this is not the case when switching from one flow category to another . The defect of this correction technique is that it behaves moderately in the presence of consecutive losses of cells, but this event is extremely rare for low rate services. If unfortunately, on a particular transmission, this happened too often then it would be enough to move to the larger size matrix to increase the correction capacity. This would increase the processing time at the same time.

Claims (8)

 A method for interleaving data from distributed and interactive and coded Reed-Solomon audiovisual services to protect them against bit errors and specific cell losses of asynchronous time networks, characterized in that it consists in keeping the same Reed code Solomon, whatever the application envisaged, to interleave (2) these coded data RS on a variable depth fixed by the application before their transmission in packets or cells on the network, to deinterlace (9) the information contained in packets or cells, using the parameters described in Recommendation I.363 of the CCITT (CSI bit) to dynamically synchronize the mechanism and to correct (10) by means of Reed-Solomon code the errors and deletions in the returned symbols.  2. Method according to claim 1, characterized in that the interleaving (2) consists in memorizing line by line the information coded in Reed-Solomon words of N symbols in a memory and in reading this memory column by column or diagonal by diagonal to form the fields "SAR-PDU payload" ATM cells this taking the precaution of choosing N multiple of the size of these fields "SAR-PDU payload".  3. Method according to any one of claims 1 and 2 characterized in that the deinterlacing (9) is to store the fields "SAR-PDU payload" column by column or diagonally diagonally in a memory and read this line memory by line to reform the Reed-Solomon codewords.  4. Method according to any one of the preceding claims, characterized in that the size of the interleaving and deinterleaving memories is variable for a given Reed-Solomon code word and depends on the application and / or the transmission and / or cost conditions. of the developed system.  5. Method according to any one of the preceding claims, characterized in that the Reed-Solomon codes are RS (N, K, D) codes on GF (2n) of N-code length multiple of the size of a field. SAR PDU payload "and preferably for maximum efficiency as close as possible to 2n but remaining lower.  6. Method according to any one of the preceding claims, characterized in that it consists in using the erasure correction capacity (probable errors already localized).  7. Method according to any one of the preceding claims, characterized in that each symbol has the dimension of a word of n bits if the error correction code is defined on GF (2n). In the case of ATM, n = 8.  8. Method according to any one of the preceding claims, characterized in that it consists in inserting ghost cells in a deinterleaving array by marking each symbol so that it is interpreted as an erasure by the Reed-Solomon decoder and this in synchronism with the dynamic variation of the size of the interlacing depth.