FR2766036A1 - Error correction technique for poor transmission condition asynchronous mode transfers - Google Patents

Abstract

The error correction technique has a transmitter (4) sending digital words (D1). The words are arranged into a matrix (6) made up of the information and redundant information (Rc) of a set size, determined by a second code detector. Following transmission across a network interface, a second matrix (7) is formed. The first field (1) is corrected column by column before transmission to the line by line code detector using the redundant matrix information.

Description

 The invention relates to a method for detecting and correcting errors introduced into digital information during its transmission between a transmitter and a receiver. The invention relates in particular to a method for detecting and correcting errors for digital information transmitted in an ATM network, an abbreviation for Anglo-Saxon for Asynchronous Transfer Mode. It effectively improves the quality of transmissions made in disturbed environments.

 Telecommunication systems studies are increasingly using the ATM transfer mode. This technique makes it possible to convey digital information of a variety and irregular nature and bit rate (for example voice, video, files). The information is transported through a network which can be of various topologies (for example mesh, ring, star, etc.). Each node of the network, called ATM switch, is connected to the adjacent node by a transmission route accepting all types of technologies (cable, microwave, optical fiber, etc.).

The ATM transfer mode implements a switching technique and an asynchronous transfer technique for the transmission of information over a network. These techniques are based on the principle of transmission by information packets of fixed length, called cells
ATM. This principle consists of segmenting the data from a transmitter into blocks of fixed length and adding a header to each block to form an ATM cell. ATM cells from different transmitters are multiplexed and transmitted asynchronously by the network routes to their destination. The transfer is asynchronous because the transmitted cells are identified only by their logical address and not by their time position or their rank within a periodic frame, as is the case in a synchronous transfer.

 Asynchronous transfer techniques are already adopted in military tactical networks operating in a disturbed environment. Disturbances generate errors within the digital information transmitted by network arteries. To combat these errors, an interleaving process is known.

 According to this method, a matrix structure is implemented.

It consists of two fields. A first field includes data, a second field includes data redundancy information.

The data to be transmitted by the transmitter are stored in a first matrix whose structure is identical to that previously described. The data from the transmitter is arranged by column, column after column to fill in the first field. As soon as a column in the first field is complete, redundancy information for the content of this column is stored in the second field in the extension of this column. Once the first matrix is complete, the digital information consisting of the data from the transmitter and the redundancy information is transmitted per line, to interleave the data and the redundancy information, line after line by routes from a network to a receiver. During reception, the received lines are stored in a second matrix, the structure of which is identical to that previously described. The storage is done by line, line after line.

When the second matrix is complete, a decryption system uses the content to correct the first field containing the data as a function of the errors detected thanks to the redundancy information.

 Thus for example, let be a matrix structure containing symbols whose first field comprises 24 rows of X columns and whose second field comprises 6 rows of X columns. The data to be transmitted by a transmitter are stored in the form of symbols in the first field of the first matrix. The redundancy information is obtained by using a Reed Solomon code. After transmission, two types of error can be distinguished.

 A first type of error modifies the digital information and corresponds to a bit error. In this case the symbol is wrong. This type of error is called an "error". The implementation of the preceding matrix structure makes it possible to detect and correct a maximum of three "errors" per column and for all the columns of 24 + 6 rows of the second matrix.

A second type of error causes the loss of digital information. This case corresponds to the deletion of a symbol, even a line for example. This type of error is called "erasure". According to the rules of the art, a determined mechanism for filling the second matrix is implemented during reception. It makes it possible to respect the framing of the digital information transmitted. This mechanism may include, in particular, a method of maintaining binary integrity. The mechanism can provide the position of the detected symbols lost (losses which correspond to "erasures") and replaced by stuffing symbols. Finally, upon receipt, the use of a Reed Solomon code, informed of the position of the "deletions", makes it possible to correct up to 6 "deletions" per column, for all the columns of 24 + 6 lines of the second matrix for the example considered.

 Thus, the interleaving of the data carried out by the interleaving process facilitates the corrections to be implemented. In fact, according to this method, any disturbance appearing on an artery is distributed throughout the digital information stored in the second matrix.

However, the error detection and correction capabilities of this interleaving process do not make it possible to respond satisfactorily to the various configurations of errors generated by the high disturbance rate affecting current military networks. Indeed
The increase in the detection and error correction capacities of this process necessitates increasing the size of the second field. Using the previous example, to detect and correct up to three "errors" per column, the redundancy information must be dimensioned with six lines. In the case of a military network, the rate of disturbance is high and leads, for example, to having to detect and correct up to twenty-four "errors". Under these conditions, the second field comprises forty-eight lines of redundancy information. Thus, the size of the second field quickly becomes of the same order of magnitude as the size of the first field or even greater. This is completely unacceptable from an efficiency point of view. In the case of the previous example, the method according to the invention must be able to detect and correct up to three totally erroneous columns and correct up to six erased rows; which is impossible with the known and previously described interleaving method, whatever the size of the redundancy field.

The object of the invention is to overcome the drawbacks mentioned and to respond to the problem posed. To this end, the invention relates to a method for detecting and correcting errors introduced into digital information, containing data from a transmitter intended for a receiver, during its transmission to the receiver by storage media. transmission subject to disturbances characterized in that it consists:
to define a first matrix of determined size formed by a first field of determined size and of a second field of determined size, to fill each column of the first matrix, column after column, by arranging in the form of symbols in the first field the data from the transmitter and by storing in the second field a first redundancy information, of determined size, determined by the application of a first detector and error correcting code on the content of the first field of the column
to transmit line by line the content of the first matrix, by associating with each transmitted line a second redundancy information, of determined size, determined by the application of a second detector and error correction code on the content of the line ,
to define a second matrix with the same structure as the first matrix,
to fill each line, of the second matrix, line after line, with a received line, and corrected by means of a second detector and error correcting code applied to a set formed by the received line and the second redundancy information associated
transmitting to the receiver the content of the first field of each column of the second matrix, column after column, by correcting the content of the first field before its transmission, by means of the first detector and error correcting code applied to the column.

 Thus, the method according to the invention defines a new matrix structure. This new structure comprises, in addition to the fields of the known matrix structure, a third field in which is stored a second redundancy information. The arrangement of the second redundancy field within the matrix structure has the effect of considerably increasing the capacity for detecting and correcting errors. Indeed, the second field makes it possible to carry out a first detection and correction of errors on the data but also on the first redundancy information. Thus the doubling of the redundancy information makes it possible to multiply by more than two, the capacity for detection and correction of errors. On the other hand, this significant gain in efficiency requires only very little additional memory space, both on transmission and on reception, and it requires only a slight increase in the throughput on the arteries of the network.

The invention will be better understood and its advantages and other characteristics will emerge during the following description presented by way of non-limiting illustration and made with reference to the appended figures which represent:
FIG. 1, a matrix structure defined by the method according to the invention,
- Figure 2, a diagram of the steps of the method according to the invention.

 In the various figures, the homologous elements are represented with the same reference.

The matrix structure defined by the method according to the invention consists of a matrix of determined size. According to an illustration, represented in FIG. 1, the matrix structure comprises L + RI lines of
C + R2 columns. This amounts to saying that the matrix structure has a size equal to (L + Rl) x (C + R2). L, R1, C, R2 are natural numbers. The matrix structure contains symbols. Depending on the application, a symbol can have a variable binary dimension, it can correspond to a bit, a byte, a 16-bit word or an ATM cell for example. A first field of size L rows x C columns contains data transmitted by a transmitter intended for a receiver. The data is stored in the form of symbols. A second field 2 of size R1 rows x C columns contains a first redundancy information Rc. The first redundancy information Rc is the result of a first detector and error correcting code applied to the symbols corresponding to the data transmitted by the transmitter. The second field 2 thus gives redundancy information for the first field 1. Thus, each column of the matrix structure is composed of a first field of L symbols and a second field of RI symbols. The first field contains symbols corresponding to data transmitted by the transmitter. The second field contains the symbols for redundancy information resulting from a first detector and error correction code applied to the symbols of the first field. The detector and error correcting codes are perfectly known to those skilled in the art. The error detection and correction code can be chosen according to the performance of the error detection and correction capacity to be achieved. The code chosen can be a Reed code
Solomon which has the particularity of being particularly effective within the framework of the process. A third field 3 of size (L + R?) Rows x R2 columns contains a second RL redundancy information. The second redundancy information RL is the result of a second detector and error correcting code applied to the first field 1 and to the second field 2 of the matrix structure. The second code chosen can also be a Reed Solomon code for the same reason as above.

 Thus, each line of the matrix structure is composed of a first field of C symbols and a second field of R2 symbols. The first field contains symbols of the data if it is a line among the first L lines, otherwise it contains symbols of the first redundancy information Rc. The second field contains redundancy information RL result of a second detector and error correction code applied to the content of the first field of the line considered.

 During a transmission of digital information between a transmitter and a receiver by arteries of a network subjected to disturbances such as a military network, the method according to the invention is implemented to fight effectively against disturbances.

 FIG. 2 illustrates different stages of the method according to the invention.

A transmitter 4 transmits data to a receiver 5. The data is transmitted by routes of a network not shown.

 In a first step, the data transmitted D1 by the transmitter 4 are stored in the form of symbols in a first matrix 6. The first matrix 6 is formed of the first field 1 and of the second field 2 of the matrix structure previously defined opposite Figure 1.

Thus each column C6 of the first matrix 6 comprises a first field formed by L symbols and a second field formed by RI symbols.

The symbols of the transmitted data D1 are arranged in the first field 1 according to a classification by column. In parallel with the storage of the data, redundancy information Rc is stored in the second field 2.

As soon as the first field of a column C61 is full, redundancy information Rc formed by RI symbols, calculated in the manner previously explained with reference to FIG. 1, is stored in the second field of column C61.

 The first step is completed when the two fields 1, 2 of the first matrix are full.

In a second step, the digital information stored in the first matrix 6 is transmitted by the arteries, not shown, of the network. The content D2 of the first matrix 6 is transmitted by line, line after line. Given the column filling mode of the first matrix 6, carried out during the first step, the line transmission of the content D2 of the first matrix 6 interleaves the data stored in the first matrix 6. Each line L6j among the L + R1 lines of the first matrix 6 is transmitted with a second redundancy information RL which is associated with it. The second redundancy information
RL is calculated in the manner previously explained with reference to FIG. 1. The calculation is carried out at the start of the transmission of the line L6j by the arteries of the network. This makes it possible to benefit from the gain in efficiency of detection and correction of errors of the new matrix structure by very little increasing the available memory size. Indeed, the additional memory required corresponds to the size of the redundancy information
RL of a line, ie R2 symbols. This characteristic obviously limits the increase in flow on the arteries. The second step is completed when all the content D2 of the first matrix 6 has been transmitted.

 In a third step, the digital information received D3 from the network routes is processed as and when it is received.

The digital information received D3 is in the form of lines received L'6j each accompanied by their redundancy information RL.

The result of the processing is stored by line, line after line, in a second matrix 7. The processing of the digital information received D3 consists in correcting the digital information by applying, to the assembly formed by a received line L ' j and its redundancy information RL, the second detector and error correcting code. The processing gives a result of size C symbols. This result is stored in the second matrix 7 according to a row arrangement. Thus, the size and structure of the second matrix 7 are exactly the same as those of the first matrix 6. Thus, on the reception side, the additional memory required is the same as on the transmission side.

 The third step is completed when the two fields 1, 2 of the second matrix 7 are full.

 In a fourth step, part of the digital information contained in the second matrix 7 is transmitted by column, column after column, to the receiver 5.

 Each column C71 of the second matrix 7 is processed before being transmitted to the receiver 5. The processing of a column C7i consists in correcting the digital information contained in column C71 by applying the first detector and error correcting code to column C71. Only the result of the processing, of size L symbols, is transmitted to the receiver. Column by column transmission allows the deinterlacing of the data contained in the first field of the second matrix.

 Thus, the capacity for detecting and correcting errors has been notably increased thanks to the specialization of the two redundancy fields. The second redundancy field allows correction on a line of erroneous symbols. The first redundancy field allows correction on a column of "deletions" and / or residual "errors".

The succession of the process steps as described above is the most effective because:
- the correction of the isolated "errors" carried out initially by means of the second redundancy information RL makes it possible to exploit to the maximum the capacity of correction of the "erasures", correction which is carried out in a second time by means of the first redundancy information Rc,
the second RL redundancy information protects the first Rc redundancy information,
- the processing linked to the second RL redundancy information is done in real time and requires very little increase in memory.

In a second embodiment of the method, it is carried out in the order:
-the storage by line, line after line, in a first matrix of the information transmitted by the transmitter,
the determination of the first redundancy information Rc, calculated on a column, column after column, after the first matrix is full,
the transmission of the content of the first matrix as described previously in the second step,
storage in a second matrix as described previously in the third step,
the transmission by line, line after line, to the receiver of the content of the second matrix after having corrected the first field thanks to the exploitation of the second field by using the first detector and error correcting code.

The choice of a REED SOLOMON code for the first detector and error correction code as well as for the second detector and error correction code is the most effective choice. REED codes
SOLOMON allow you to detect and correct symbols. Depending on the implementation and expected performance, a choice of symbol size is made.

Under these conditions, the second redundancy information RL makes it possible to detect and correct for each information received (R2 / 2) symbols from among (C + R2) symbols received. The first redundancy information Rc makes it possible to correct up to:
- E erroneous symbols among the (L + Rî) symbols of each column of the second matrix, with E <(RI (R1) / 2.

- P detected symbols lost by the filling mechanism of the second matrix among (L + Rî) symbols, with as relation between
E and P:
2E + P <RI and E <(R1) / 2 and P <R1.

 In addition, the simultaneous use in reception of a process allowing the precise detection of the locations of lost symbols, makes it possible to double the correction capacity of the REED SOLOMON code.

 Depending on the performance to be achieved, other choices can be made for the detector and error correcting codes. In particular, a less powerful code is generally suitable for calculating the second RL redundancy information. Thus, it is for example possible to choose another cyclic code, such as a Hamming code or a BCH code.

Claims (2)

 1. Method for detecting and correcting errors introduced into digital information, containing data from a transmitter intended for a receiver, during its transmission to the receiver by transmission media subject to disturbances, characterized in that 'it consists:  to define a first matrix (6) of determined size (L + R1 rows x C columns) formed of a first field (1) of determined size (L rows x C columns) and of a second field (2) of size determined (R1 rows x C columns), to fill each column (C6;) of the first matrix (6), column after column, by arranging in the form of symbols in the first field (1) the data (Dl) from l transmitter (4) and by storing in the second field (2) a first redundancy information (Rc), of determined size (R1), determined by the application of a first code detector and corrector of errors on the content the first field of the column (C6i),  to transmit line by line the content (D2) of the first matrix (6), by associating with each line (L6;) transmitted a second redundancy information (RL), of determined size (R2), determined by the application d '' a second code detecting and correcting errors on the content of the line (Lij),  defining a second matrix (7) with the same structure as the first matrix (6),  filling each line (L7), of the second matrix (7), line after line, with a received line (L'si), and corrected by means of a second detector and error correction code applied to a formed assembly the received line (L'6;) and the second associated redundancy information (RL),  to transmit to the receiver (5) the content of the first field (1) of each column (C7j) of the second matrix (7), column after column, by correcting the content of the first field (1) before its transmission, by means of the first detector and error correction code applied to the column (C7;).  2. Method according to claim 1, characterized in that the first and second detector and error correcting codes are REED SOLOMON codes.