Digital transmission system synchronizable on initialization sequences

0001The invention relates to a digital transmission system comprising at least one transmitter transmitting a signal comprising an initialization sequence and at least one receiver which is synchronized with the transmitter using correlation means which perform sequence recognition initialization.

0002It is a common problem to synchronize a signal receiver on a signal transmitter each placed at the ends of a transmission channel. Synchronization can be obtained by detecting a sequence sent at the initialization of the transmission.

0003The document "Frame Synchronization Techniques" RA SCHOLTZ, IEEE Transactions on Communications Vol. COM-28, No. 8, 1980, pages 1204-1213 describes a frame synchronization technique based on a correlation between an initialization sequence and a reference sequence. When the two sequences are identical, the correlation is maximum and a correlation signal is emitted, used to establish synchronization. However, such a technique is quickly limited when the transmission channel generates distortions.

0004This situation occurs in digital transmission systems, for example in telephony, for which at one end of a channel is a transmitter / receiver which operates as a transmitter while at the other end there is another transmitter / receiver which operates as a receiver, and vice versa. In such a bidirectional operating mode, it is necessary to synchronize the two transmitters / receivers with one another during an initialization period. To perform this synchronization, synchronization signals emitted by the transmitter and identified by the receiver are used. To overcome the synchronization difficulties caused by the transmission channels, the synchronization signals are initialization sequences. Synchronization of the receiver can be obtained by exploiting the repetition of the signals transmitted by the transmitter.

0005However, it is necessary to identify with precision the moment when the receiver has recognized the initialization sequence, poor precision causing synchronization faults and / or excessive complexity of the reception equipment.

0006The object of the invention is therefore to determine with great precision said moment, even in the presence of distortions brought by the channel or in the presence of noise.

0007This object is achieved with a system for which, the initialization sequence comprising a preamble followed by a reference sequence, the correlation means are preceded by equalization means which equalize the signal received on the preamble, the means of correlation operating the correlation on the reference sequence.

0008Thus advantageously, the correlation means can operate with perfectly equalized signals on the preamble of the initialization sequence, which makes it possible to detect with great precision the appearance of the reference sequence.

0009In the case of a signal modulated from states of a constellation, the preamble is formed of a first sequence of states. This sequence is preferably repetitive to reduce the material resources and accelerate the equalization.

0010To increase the performance of the correlation, the reference sequence then comprises a first part which extends the sequence of states of the preamble, this first part being juxtaposed with a second part formed by another sequence of states. This makes it possible to precisely detect the appearance of the transition located at the juxtaposition of the first part and the second part of the reference sequence and to synchronize with high precision even with a low signal-to-noise ratio.

0011These different aspects of the invention and others will be apparent and elucidated from the embodiments described below.

0012The invention will be better understood using the following figures given by way of nonlimiting examples which represent:<ul id="ul0001" list-style="none" compact="compact"><li>Figure 1: a representation of an MAQ 4 constellation.</li><li>Figure 2: a diagram of a transmission system according to the invention.</li><li>Figure 3: a diagram of an equalizer according to the invention.</li><li>Figure 4: a diagram of the correlation means according to the invention.</li><li>Figure 5: a graph showing the variations of the output signal of the correlation means.</li></ul>

0013By way of example, the description which follows relates to an MAQ 4 modulation, represented in FIG. 1, used to synchronize a transmission modem conforming to the standards of the V series defined by the International Telegraph and Telephone Consultative Committee (CCITT) , standards published by the International Telecommunications Union, Geneva (1991). However, the invention can be used to synchronize other systems, with or without standards, using other initialization sequences.

0014Consider, for example, the case where there is a transmitter / receiver at each end of the channel to constitute a bidirectional link.

0015FIG. 2 represents a digital transmission system 10, which comprises a first signal transmitter / receiver 12A placed at one end of a CHA channel, and a second transmitter / receiver 12B placed at the other end. The signals to be transmitted, for example for applications in the field of telephony, are coded using the constellation MAQ 4 represented in FIG. 1. It is a constellation with four states: A (-6 , -2): B (2, -6): C (6.2): ​​D (-2.6).

0016According to the invention each transmitter / receiver comprises CORR correlation means, 30A, 30B, preceded by EQUAL equalization means 14A, 14B which correct the distortions of the signals due to their transmission. Each transmitter / receiver also comprises processing means PROC 16A, 16B operating on the equalized signals and timing means RHY 18A, 18B which supply the timing signals necessary for the transmission system. The equalization means 14A, 14B serve for synchronization. The processing means also have their own equalizers operating on the data containing the information to be transmitted. According to the aforementioned standard, the initialization sequence is formed of a first sequence repeating signals modulated by states A and B of the constellation according to the sequence ABABAB ... taking place over 256 clock-symbol periods of elementary duration T. At the end of the 256 periods, the first sequence is replaced by the second sequence repeating signals modulated by states C and D of the constellation according to the sequence CDCDCD ... taking place over 16 clock-symbol periods.

0017The synchronization of the first and the second transmitter / receiver one on the other will consist in detecting the moment of appearance of the transition between the sequence of states AB and the sequence of states CD and in synchronizing the timing means .

0018The equalization according to the invention is carried out on, for example, 200 symbols of the ABABAB suite. These 200 symbols constitute the preamble within the meaning of the invention. At the end of this preamble, the equalization on the preamble is stopped and the correlation can be performed on the remaining AB symbols and on the CD symbols. The reference sequence will then include the symbols placed on either side of the transition, that is to say AB symbols and CD symbols. With a deterministic sequence ABABAB ... the equalization tends to calculate the equalization coefficients specific to this deterministic sequence. It is not necessarily suitable for CD symbols. -Nevertheless, this equalization on the AB symbols is decisive for reinforcing the correlation peak which appears at the time of the transition.

0019Thus, when the following sequence CDCD ... occurs the correlation means CORR, 30A, 30B can accurately detect the moment of appearance of the transition between the states A, B and the states C, D of the sequence of initialization.

0020Figure 3 shows a preferred embodiment of an equalizer according to the invention. It includes a shift register formed of cells 20₀ to 20₈, the shift being carried out at a rate T / 3, that is to say three times higher than that of the symbol clock T of the input signals x (not). The output of every second cell, that is to say cells 20₀, 20₂, 20₄, 20₆, 20₈ enters a multiplier 22₀ to 22₄ which also receives a weighting coefficient C0, C1, C2, C3, C4. The outputs of the five multipliers are added to a summator Σ 25 which delivers the equalized signal y (n) which enters the correlation means shown in FIG. 4.

0021The signal y (n) continuously enters a shift register 32 formed of elementary cells R₀, R₁, ... R_N-1, R_NOT, where N + 1 is the length of the reference sequence. The outputs of each cell S₀, S₁, ... S_N-1, S_NOT enter into multiplication means 32₀, 32₁ .... 32_N-1, 32_NOT which receive respectively on an input an output S₀, S₁ ... or S_NOT and on another entry a conjugate state of the constellation forming the reference sequence. The distribution of the reference sequence is made in such a way that the multiplication means 32₀ receives the conjugate of the last state of the reference sequence, that the multiplication means 32_NOT receives the conjugate of the first state of the reference sequence and that, the intermediate multiplication means each receive, in order, a conjugate of the intermediate states of the reference sequence. All outputs 34₀, 34₁, ..., 34_N-1, 34_NOT multiplication means are accumulated in an adder 35 which delivers an output signal z (n) according to the equation:<maths id="math0001" num=""><math display="block"><mrow><mtext mathvariant="italic">z</mtext><mtext> (</mtext><mtext mathvariant="italic">not</mtext><mtext>) = </mtext><apply><sum /><lowlimit><mtext mathvariant="italic">i</mtext><mtext>=0</mtext></lowlimit><uplimit><mtext mathvariant="italic">i</mtext><mtext>=</mtext><mtext mathvariant="italic">NOT</mtext></uplimit><mrow><mtext mathvariant="italic">y</mtext><mtext>(</mtext><mtext mathvariant="italic">not</mtext><mtext>-</mtext><mtext mathvariant="italic">i</mtext><mtext>) .</mtext><mtext mathvariant="italic">r</mtext><mtext>*(</mtext><mtext mathvariant="italic">i</mtext><mtext>)</mtext></mrow></apply></mrow></math><img file="EP0702467A1_D0001.tif" /></maths>    The reference sequence can for example be the sequence:<maths id="math0002" num=""><math display="block"><mrow><mtext>r * (i) = A * B * A * B * A * B * A * B * C * D * C * D * C * D * C * D *</mtext></mrow></math><img file="EP0702467A1_D0002.tif" /></maths> which includes 16 conjugate states formed by two equal groups of 8 conjugate states distributed on either side of the transition.

0022The reference sequence can be short to limit the material means necessary for the implementation of the correlation.

0023When the succession of incident sequences circulates in the cells R₀ ... R_NOT, there occurs a moment when the result of the multiplication of the sequence being, at this instant, in the cells R₀ ... R_NOT by the reference sequence described above provides a maximum output signal z (n). The instant of the transition can thus be determined with great precision.

0024FIG. 5 is a graph showing the variations of the output signal z (n) in the case of a signal to noise ratio of 6 decibels. This signal remains erratic (representing noise) until the transition between the states AB and the states CD is correctly located in the cells R₀ ... R_NOT. The output signal z (n) then has a peak which is used by the timing means 18 as a recognition signal REC1, REC2 to respectively trigger the synchronization of the two transmitters / receivers.

0025It is observed that the signal z (n) is a discrete series of correlation values. Around the maximum correlation, the timing means 18A, 18B will identify the exact position of the central peak which corresponds to the maximum correlation. For this, we compare, two by two, consecutive values ​​... z (n + k-1), z (n + k), z (n + k + 1) ... where n is a current index . The maximum value of z makes it possible to determine the sample, for example the sample (n + k) where the maximum appears. Knowing the composition of the initialization sequence, the approximate position k of the maximum of z is known in advance, which makes it possible to stop the course of the correlation when the transition has passed with certainty (delay td). By delaying the data to be processed correspondingly (delay equal to td), in the processing means 16A; 16B, it is thus possible to synchronize the timing means and the processing means at the time of the transition.

0026The maximum of correlation is particularly narrow which makes it possible to locate it with ± 1 sample, which considerably simplifies the hardware complexity of the system. However, taking into account the noise, the central peak can have a higher additive noise than a lateral peak which, for its part, can be the true maximum. For this, it is possible to use, in the processing means, several processing equalizers wedged respectively on the central peak and on several lateral peaks. With an accuracy of ± 1 sample, the invention makes it possible to greatly reduce this multiplicity of downstream processing equalizers.

0027The invention has been described with means operating the correlation by carrying out multiplications of the incident sequences by a reference sequence formed of conjugate states using multipliers. It is nevertheless possible to operate the correlation measurements differently, for example using a digital signal processor programmed to perform the correlation.