FR2458958A1 - DIGITAL INFORMATION TRANSMISSION SYSTEM WITH DETECTION AND CORRECTION OF ERRORS - Google Patents

Abstract

DIGITAL DATA TRANSMISSION SYSTEM WITH ERROR DETECTION AND CORRECTION, EQUIPPED WITH MEANS OF CODING IN WORDS OF A MESSAGE, MEANS OF DECODING, MEANS OF SYNCHRONIZATION AND MEANS TO REPEAT EACH WORD. AN "OR-EXCLUSIVE" OPERATION IS EXECUTED BETWEEN EACH B1 WORD OR REPEAT OF A WORD WITH EACH C1-C12 PART OF A PREDETERMINED BINARY CONFIGURATION PRESENTING A HIGH AUTOCORRELATION COEFFICIENT. THE SAME OPERATION IS EXECUTED AT DECODING. WE COMPARE EACH WORD WITH ITS REPETITION UNTIL YOU FIND AN IDENTITY. THE CODING IS THEN ACCEPTED. IN CASE OF NON-IDENTITY, THE MESSAGE IS REJECTED. APPLICATION TO THE TRANSMISSION OF DIGITAL DATA, ESPECIALLY IN SYSTEMS GIVING RISE TO GROUP ERRORS.

Description

The present invention relates to a system for transmitting information

digital detection Eb

error correction.

  When transmitting digital data through transmission systems that give rise to errors in bit positions, it is known to employ a transmission system capable of detecting and correcting such errors. In some transmission systems, errors in bit positions occur substantially time-based and means have been developed, often referred to as codes, which are particularly

  desirable to detect and correct such a distribution

  errors. In other transmission systems,

  for example related to mobile radio systems, the probability of

  It is very big. errors on the binary positions occur in groups, often called

  "burst", which are often due to fading.

  The transmission system according to the present invention has particular advantages when used with transmission means in which the errors on the bit positions occur in particular.

in groups.

  Known transmission systems of the above kind further include means for synchronizing the transmitter and the receiver so that the receiver is informed of the start time of the message or a certain portion of that message. For this purpose, special sequences of bit positions may be used in which, of course, errors may also occur during transmission. It is known to provide means in the decoding system which is arranged to ignore a number of position errors.

  during the transmission of the characteristics

concordance.

  The most common error detection method

  nally used involves the transmission of

  Checks whose value depends on the values of the positions of the word to be protected by these control positions. Among these methods, the best known is the one in which parity control is used, but more complex polynomial controls are also known, which are derived from a polynomial and which are decoded according to

the same polynomial.

  To detect and correct errors, it is also known to use a word repetition procedure. Here, the decoding scheme consists in comparing the repeated transmissions of the same word, position by position, and the position that occurs most frequently is considered correctly received. The method is called "vote-based" error correction and has also been proposed to compare combinations of bit positions. The practical value of the different processes can be calculated on the one hand and measured in terms of message acceptance defined by the report, on the other hand.

  expressed in percent, between the number of messages correcting

  received and the number of messages transmitted.

  The present invention relates to a digital transmission system comprising coding and decoding means as well as synchronization means, and its purpose is to provide improved acceptance of

  particularly when using the means of

  mission in which errors frequently occur.

especially in groups.

  According to the invention, the digital information transmission system with detection and error correction comprises means for dividing an information stream into messages and each message into words, decoding means, synchronization means, and means to repeat each word at least once. It is characterized in that it comprises: means for executing an operation

  "OR-Exclusive" between each word or repetition

  words and each part of a predefined configuration

  completed binary positions with a high autocorrelation coefficient; means for executing an "OR Exclusive" operation on each repeated word or word and each part of the same bit position configuration during

decoding ___-

  means for comparing each word with the repetition of the same word until the comparison conveys an identity; means for accepting such an identity as a correctly received word and for rejecting the word in the event of non-identity; and - means to consider the recognized identity

  as a concordance information.

  Other features and advantages of the invention

  from the detailed description which will be

  to follow. In the accompanying drawings given by way of nonlimiting examples: FIG. 1 is a block diagram of a radio station provided with a transmission system according to the invention; FIGS. 2a and 2b represent messages of a 3a to 3f show a number of messages transmitted by the transmission system according to the invention, and FIG. 4 shows the result of measurement translated into curves of the acceptance of messages as a function of a number of messages. average HP signal for two digital transmission systems of known type and for two transmission systems

  digital device according to the invention.

  The radio station shown in FIG. 1 comprises an antenna 1 which, via a cable 2, is connected to an antenna switch 3 actuated by an electronic relay 4 capable of connecting the antenna 1 to a transmitter 5 or a receiver 6. A second relay 7, also electronically controlled, has contacts 8 capable of connecting the transmitter 5 to a modulator 9 or to an amplifier 10 comprising a microphone 11 and a contact 12 capable of connecting the receiver 6 to a demodulator 1e or an amplifier 14 comprising a

sound reproduction apparatus.

  Relays 4, 7 and 12, the transmitter 5, the receiver

  6, the modulator 9 and the demodulator 13 are connected

  by drivers to an input unit 16

  output of a microcomputer which essentially comprises a microprocessor 17. The microprocessor is connected in known manner to a timer 18, to. a ROM or

  PROM 19, a RAM 20 and a clock 21, respectively

  tively. The microcomputer can still be connected by a number of lines indicated under the general reference 22 to various control devices such as lamps, ringtones, display devices, push buttons, switches, keys,

printers or others.

  It is understood that the transceiver thus described can communicate binary messages as well as analog signals such as speech, and that this system is able to be switched between a digital transmission and

an analog transmission.

  It is stored in the ROM 19 information indicating how the message to

  transmit by the modulator 9 and the transmitter 5 to a receiver

  should be formulated and how

  message received from a remote transmitter must be decoded.

  The useful content of the messages is arbitrary but may for example relate to a selective call to a

  other receiver or may be information on a condition

  stored locally and routed through one of the lines 22 in analog or digital form. In Figure 2 is shown a message in a known format involving error detection but not error correction. The length of the message is

  52 binary positions of which the first twelve are used

  the next thirty-two positions representing data positions and the last eight positions being cyclic redundancy control (CRC) positions which can be calculated by the transmitter as well as the receiver based on thirty-two aforementioned positions. If the receiver recognizes the control positions as unrelated to the received data positions, this is interpreted as detecting a transmission error

and the message is rejected.

  Such formats are commonly used in many countries because of their simple structure, but due to the fact that a single binary position error in the message causes it to be rejected, the acceptance of messages is relatively bad as shown the

curve 41 of FIG.

  Another format which, in order to give a good comparison, includes thirty-two data positions and, next to that, is built according to principles.

  known to provide error detection as well as

error.

  This message format, shown in Figure 2b, contains one hundred and twenty four positions in total and begins

  by thirty-two positions arranged according to a configura-

  a predetermined high autocorrelation coefficient serving only as the synchronization configuration of the

  message. Error correction in the concordance part

  The result of this message is that it is only required that twenty-eight out of the total of thirty-two positions be correctly received in order to accept the configuration of synchronization.

chronization of the message.

  This criterion is an acceptable compromise between the tolerance of as many errors as possible and

  the rejection of a number of false synchronizations of messages

ge that would be disruptive.

  The synchronization section of the message is followed by a total of ninety-four positions among them thirty-two positions are data positiors, eight are CRC positions and the fifty-two

  remaining positions are redundancy positions.

  The error correction in this part of the message is performed by coding these thirty-two data positions and eight CRC positions according to a code called

  HAgelbarger describes B.W. Hagelbarger in Bell Systems Tech.

Journal, vol. 38, 1969, PP 969-984.

  This code has the ability to correct up to six consecutive position errors. The following nineteen positions must be correct and it is then able to correct the following six consecutive positional errors (maximum) again, and so on. The acceptance of messages in the format thus described is represented by

curve 42 in FIG. 4.

  Fig. 3a shows one embodiment of a message code having a length of ninety-six bit positions and including transmission means according to the present invention. References Bi, B2, B3 and B4

  refer to data bytes 1, 2, 3 and 4, respectively

  is lying; we will denote that a multiplet comprises eight binary positions. Here, a byte of data corresponds to a number "and it appears that the first word is transmitted followed by

  two repetitions after which the same configuration is repeated

  for the next three words of the message. Decoding is done by comparing each word with the repetition of the

  same word until the comparison shows an identity.

  We see that a multiplet in each of the four groups that each include three multiplets may include a

  error without the detection being disturbed, which constitutes

  kills the error correction. If more than one of the three bytes of a group are erroneous, then none of the comparisons show an identity, which causes the rejection of the message, this being the error detection. Byte voting provides error detection contrary to

  position-based vote that provides a single error correction

  only two bytes of at least three must be identical, whereas by voting at the binary position at least two out of three of the positions must be identical, which is of course always

the case.

  Voting at the byte level, in particular, has an advantage when the input signal to the decoder is random (noise). In this case, the vote at the binary position always results in an erroneous message, whereas, by the vote at the multiplet level, the

  individual bit positions in at least two multi-

  plets must be in agreement. It can happen, however

  that two multiplets in the same group are

  error in identical binary positions, which results in false detection. This eventuality occurs in any code. If greater security is desired than that obtained in the manner just described, the message may be extended by an additional multiplet, which serves as a cyclic redundancy check (CRC). This byte must then be repeated as the four bytes of data and must be decoded from the

same way.

  In what has been described so far, we have not discussed the synchronization of the message, for the sake of simplification. The position configuration actually transmitted is not that of Figure 3a but that of Figure 3b. Note that the sign + circled means 'OU-Exclusive'.

  perform this operation. The multiplets C1-C12 represent

  tent twelve multiplets that have been predetermined as words

  concordance and were chosen from a set of

  binary ration having a high autocorrelation coefficient, i.e., the binary pattern represents a pseudo-random binary sequence including numbers

  pairs of O-positions and positions1 which may, for example,

  can be generated in a shift register with a

  or several feedback loops.

  In the decoder, the twelve received bytes are subjected to an'O Exclusive 'operation with the configuration

  chosen concordance of which are the initial words.

  This is due to the fact that a word submitted twice to an opera-

  If the decoder does not have the appropriate "binary" window in its memory, the "OR Exclusive" appears unchanged.

  is very weak that one of the comparisons shows an identical

  tity. This constitutes the synchronization of the message. Thus, the twelve multiplets C1-C12 structure the message to be decoded, that is to say that the decoder knows the beginning and the end. As a result, the synchronization of messages according to the invention has the effect of rendering useless a message matching configuration before each message as well as

show it in Figures 2a and 2b.

  By the choice of different configurations of C1-C12 positions each of which has a high coefficient of autocorrelation but are mutually orthogonal, the message code according to the invention can be used immediately for selective calls. So it is possible to call different units by simple change

  of the concordance configuration that is to say without using

  be positions of the message itself.

  It is known that in the case of data transmission from mobile radios, bit position errors are often presented as a number of consecutive errors. For this, it may be advantageous to choose, instead of the configuration shown in Figure 3a, the configuration shown in Figure 3c which can apply up to sixteen consecutive position errors occurring up to two times. Two groups of position errors each containing. nine positional errors and separated from each other by thirty-nine positions are

Corrected - with certainty.

  Another possibility may be to choose the sequence shown in Fig. 3d, which is capable of applying to twelve consecutive errors of positions occurring only once. Error groups

  twenty-five errors are cor-

  with certainty. The choice of one of the formats shown in Figures 3a, 3c or 3d, respectively, depends on

  the assessment of error correction.

  The increase in the number of consecutive errors that can be corrected is, in fact, less effective

  for position errors distributed evenly.

  In FIG. 4, the curve 43 shows the acceptance

  message of the code shown in FIG. 3c.

  Even better message acceptance can be achieved if message length is extended

  up to, for example, one hundred and twenty-eight binary positions.

  This allows each word to be transmitted four times as shown in Figure 3e. This structure applies up to sixteen consecutive bit positions that appear up to four times, these four groups of errors each containing up to nine errors and separated from each other by at least

  twenty three binary positions are corrected with certi-

  study. The words can be interchanged as shown in Figure 3f, so that what is repeated is a pair of words, and the comparison can then be made between the words themselves or between the word pairs. A message such as that of FIG. 3f withstands up to thirty-two consecutive binary position errors with respect to

  each half of the message, and groups of twenty-

  five consecutive position errors separated by thirty-

  nine positions are corrected with certainty. Accepting messages from a coded transmission in this manner is

  illustrated by curve 44 in FIG.

  When the comparison sequence executed during the decoding procedure is adapted to the statistical distribution of the bit position errors of the

  transmission in question, it is possible to improve the

  messages from this channel. Although it has been shown, in connection with FIG. 2, that the decoding is performed by means of a microcomputer, it is understood that the comparison procedure can be carried out in usual circuits with integrated circuit components by

example of the TTL or CMOS type.

Claims (7)

  1. Digital information transmission system   error detection and correction methods, comprising coding means for dividing a message information stream and each message into words, decoding means, synchronization means, and means for repeating each word at least once characterized in that it comprises:   means for executing an operation "OR   Excluding between each word of each word repetition and each part of a predetermined configuration of bit positions having a high autocorrelation coefficient;   means for executing an operation "OR   Exclusive "between each word or word repetition and each part of said pattern during decoding; means for comparing each word with the repetition of the same word until the comparison shows an identity; means for accepting a word; such identity as a correctly received word and to reject the word in the case of non-identity, and means to recognize identity as concordance information.   2. Transmission system according to the   cation 1, characterized in that it comprises means for repeating each word of the message a certain number of times   before transmitting the next word.   3. Transmission system according to the   cation 1, characterized in that it comprises means for structuring the message into sequences which are repeated a number of times, each of which consists of a or several words.   4. Transmission system in accordance with the   cation 1, characterized in that it comprises means for systematically comparing two words in a sequence of 1 2 repetition and to interrupt this comparison when one recognizes two identical words.   5. Transmission system according to the   cation 1, characterized in that it comprises means for recovering a message by executing a certain number of times,   number depending on the length of the message, comparisons   identity search sounds between two or more words.   6. Transmission system in accordance with the   cation 1, characterized in that it comprises means for   combine-this comparison of words repeated in a message   to reduce the number of comparisons required.   7. Transmission system according to the   cation 1, characterized in that it includes means for   reject a message as false if only one of the   reasons carried out systematically and repeatedly or a combination of such comparisons does not lead to an identity.-