FR2525050A1 - Repeater station for optical digital communication link - has two optical couplers and opto-electronic converters allowing electronic processing of transmission signal - Google Patents

Abstract

The repeater station includes two optical couplers (20,22) connected to the line (10). The first divides the incident signal into one component which is taken straight on by the output (28) and one which is diverted via a second line (26). This second component passes into an optoelectronic converter (40) which may either be applied to the input of an opposite type of converter (46) or applied to local processing circuitry (43). After passing through local circuitry the signal is applied to the second converter (46) and recombined (32) with the original undeflected optical component in the second coupler (22). The returned component by this stage has a delay w.r.t. the signal on which it is superimposed, and is also considerably stronger.

Description

Method and device for transmitting signals over a gated link of repeater stations
The invention relates to installations for transmitting remote signals, in particular optical signals. It relates more particularly to installations comprising a series of stations along a link, capable of receiving signals that reach them and retransmit them on this link after having regenerated or modified.

It is known that when transmitting signals on a link, and in particular light pulses on optical fibers, these signals are attenuated as and when they are propagated. Their weakening may become such that it is no longer possible to capture them in good conditions. thus, when the link is relatively long or when the signals are subject to additional attenuation factors, provision is made for regeneration stations spaced along this link.

Each station then comprises a receiving device adapted to pick up signals reaching it and a retransmitter device capable of reinjecting regenerated signals on this link with a sufficient amplitude for their transmission under satisfactory conditions to the next station.

Each station may also include a device for operating the received signals and its own signal injection means on the link. In the case of an optical link, such transceiver stations generally perform the conversion of the received optical signals into electrical signals. The latter, after reshaping, are applied to the excitation of a light source connected to the optical link.

Repeater installations are satisfactory as long as each of the repeaters in the installation is working properly. In case of failure of only one of these repeaters, however, the installation becomes totally inoperative due to the cascade connection of the different repeaters. This is the reason why it has been proposed to establish on the link a branch bypass of each station, intended to allow the passage of light signals directly downstream of the station in case of failure. For this purpose, it is possible, for example, as described in European Patent 00 2971, to equip this station with a fault detector and to place a shutter on the optical bypass ramp. This shutter, normally closed, is controlled by the fault detector to be placed in its open condition, or busy, when the station. signal regeneration is out of operation.So, the signals that reach a failed station are allowed to bypass it to the post immediately downstream. If the distance between the stations is suitably provided, the signals emitted by each station can be detected not only by the first station downstream from it but also by the second station when they are admitted bypassing the first station. An optical transmission chaste of good reliability is then obtained.

This type of installation, however, has the disadvantage of being dependent on the operation of a fault detector.

 I1 further requires a bypass shutter of each station. These shutters can cause additional attenuation of the signals to be transmitted; they also increase the cost of materials.

The present invention relates to a cascaded repeater transmission installation, in which there is a bypass ramp of each station which permanently allows the transmission of a portion of the energy of the signals received from a station upstream to a position downstream from this position.

In such an arrangement, the regenerated signals, injected on the link at the output of the station, are in general behind the signals that have passed directly on this branch link of this station. Indeed, if one takes no particular precautions, the time required for the regeneration of the signals is normally greater than the transit time of these same signals on the bypass ramp. The detection of the messages conveyed by the superimposed signals presents a difficulty. One possible solution to this problem is to reinject the regenerated signals on the link in synchronism with the signals admitted in bypass. This assumes, in the case of repeaters with electronic circuits for optical links, either the use of extremely fast circuits, or means for delaying the optical signal admitted bypassing the station. These solutions are expensive insofar as the former requires the use of high price components and where the second involves additional attenuation of the signals passing through the optical delay devices.

The aim of the invention is to overcome these drawbacks by relying on the observation that the signals regenerated in a station have a substantially greater amplitude or energy than the signals admitted as a bypass of this station and that this relationship is maintained downstream. along the link.

This difference in level is taken advantage of at the entrance of each station to rid the signals regenerated at the upstream station of the parasitic influence of the signals admitted in derivation thereof.

According to one embodiment, in the case where the signals are transmitted in digital form, in amplitude modulation by all or nothing, the peak level of the signals received by a station is detected and the only signals selected from among them are selected from among them. level is greater than a threshold set according to this peak level.

The memorized value of the peak level of the signals received on the station has a time constant on the rise and another on the descent, chosen according to the repetition period of the transmitted pulses and the possible sequences of successive zeros (no signal ). This level is used to produce, after a suitable attenuation, a reference signal which is used as a threshold for the analysis and processing of the signals received by the station. Pulses whose amplitude is greater than the threshold level thus fixed are considered to carry a binary information element whereas any portion of signal of amplitude below this threshold is considered as carrying the element. opposite binary information.

A threshold thus determined is effective for making this selection, whether the upstream station is in operation or not,
In particular, if the regeneration circuit of the latter ceases to operate, the peak level of the signals received is no longer influenced by the regenerated signals but is determined by the only weakened signals allowed in bypass of the upstream station. The threshold of analysis of the received signals is lowered according to the new value of the level of peak so as to continue the analysis with a greater sensitivity.

Depending on the degree of attenuation of the signals admitted in shunt of each station, which depends on the distance between the stations and the percentage of signal directed towards this shunt, it is possible or not to provide additional provisions making it possible to check whether the level of crest detected corresponds to the sum of the peak levels of the regenerated signals and the signals admitted in shunt of the upstream station or at the peak level of the regenerated signals only (in the case where the pulses of the two types of signals are interlaced in time, without overlapping ) so as to fix, with a better accuracy, the detection threshold from this peak level, so better to protect against noise.

In the case where the signals are transmitted in an analog form by amplitude modulation, the invention furthermore provides for the elimination of the signals received by a station from "forward echoes" or precursor signals admitted as a bypass of an upstream station, by based on the fact that one can know the amplitude of the signals admitted in derivation and their time offset with respect to the regenerated signals to keep only them.

For this purpose, it is possible, in particular, to delay the received signals at a first instant and to subtract therefrom, after a suitable attenuation, the signals received at a second instant, offset with respect to the first instant of a time interval corresponding to the delay existing between the signals regenerated at the upstream station and the signals having shifted thereto.

The invention is aimed not only at the process just mentioned but also at an installation equipped with stations with a bypass strap and making it possible to detect the signals received on the basis of the criteria that have just been specified.

In particular, it relates to receiving stations comprising means for clearing the regenerated signals coming from an upstream station of the signals admitted in shunt thereof, as well. when these signals are transmitted in digital form only when they are in analog form.

The explanations and the nonlimiting description of the following exemplary embodiments are given with reference to the appended drawings, in which FIG. 1 represents a signal transmission link along which repeater stations are distributed. FIG. 2 schematically illustrates the constitution of such a repeater and its connection to the transmission link FIG. 3 illustrates an embodiment of a repeater equipped with a digital signal receiver according to the invention FIG. 4, composed of FIGS. 4A to 4D , is a signal diagram over time, illustrating the explanations of the operation of FIG. 3 FIG. 5 shows the circuits of a receiver according to the invention for analog signals FIG. 6, formed from FIGS. 6A to 6D, is a FIG. 7 is a circuit diagram for the analysis of the peak levels of FIGS. received signals FIG. 8 represents histograms 8A and 8B illustrating the analysis of the peak levels; and, Figure 9 illustrates an embodiment of a repeater station embodying the principles of the invention for bidirectional operation of the transmission link.

FIG. 1 shows an optical fiber link 10 which makes it possible to establish communication between two terminals 12 and 14. A plurality of stations 161 to 165 are interposed on the link 10. Each of these stations 16 is able to detect the signals transmitted in at least one direction determined on this link. I1 makes it possible to re-transmit on the latter, in the same direction, regenerated signals corresponding to the signals received. I1 can also be used to inject directly into the link information generated in association with this station, being superimposed on those passing through the branch line.

Repeater stations such as 161 to 165 are useful on links of relatively long length to compensate for attenuation of optical signals under the effect of various losses between terminals 12 and 14 by regenerating them at intervals which may be more or less regular.

According to the embodiments, the signals transmitted by the channel 10 may be digital, and consist of light pulses modulated between a zero level and a non-zero level.

They can also be analog in the form of a continuously variable light flux over time.

FIG. 2 shows a typical embodiment of a transceiver station 16 suitable for use with a link or optical bus such as 10. The station 16 is connected to this link by two optical couplers 20 and 22 respectively. for the input and output signals. Each of the couplers 20 and 22 is for example a T coupler of known construction. The input coupler 20 divides the energy of the received signals on its input 25 between a channel 26 at the optical input of the station 16 and a channel 28.

The coupler 22 includes an output 31 on the optical link 10 and on which are combined the signals received from a channel output at the output 1b and the output channel 28 of the coupler 20 and the signal path. The input 33 of the coupler 22 is connected by an optical conductor 35 in shunt of the station 16, forming a shunt through which a fraction of the energy reaching the input of the coupler 20 on the link 10 is transmitted directly to the output 31 of the coupler 22 on this same link 10.

The energy not transmitted on the branch 35 is applied in the station 16 to an opto-electronic converter 40 which converts the optical signals received on the channel 26 into electrical signals present at its output 42. This is connected, by a on the other hand, by a link 43 to a device (not shown) operating these signals inside the station and, on the other hand, to a terminal 44 of a switch 45
connected to the input of an opto-electronic transmitter 46. The
switch 45 makes it possible to connect the input 47 of the transmitter 46
at terminal 44 to transmit the signals from the
receiver 40 directly to the input of the transmitter 46, or
a link 48 from the linked operating device
16. The transmitter 46 delivers on its output 49
optical signals which are transmitted to the input 32 of the coupler
output 22.

Whatever the position of the switch 45, the signals present at the output 31 of the coupler 22 are formed by the
the signals passing directly through the
bypass 35 along the track 10 without going through the post
16 and signals that are injected on that path by the
output coupler 22. When the transmitter 46 operates in
pure repeater, that is to say when its input 47 is connected
directly at the output of the receiver 40 by the switch 45, the operating time of the electronic circuits
40 and 46 generally creates a time lag between trains
light pulses transmitted directly by the bypass 35 and those which are injected at the input 32 of the coupler
22.The latter normally represent the information that
it is sought to transmit along the link 10; the
signals transmitted on the bypass 35 have a ~ level minus
higher than that of the regenerated signals. They represent a noise that we want to get rid of. Moreover, these if
Bypassed signals may be used in case of failure of station 16 to excite the station immediately downstream. For this purpose the distances are suitably selected
between stations 16, so that the attenuation of the signals transmitted bypassing a station is not too strong
when they reach the next position. Thus, the assembly presents. The interest to allow to tolerate the breakdown of the
the electronics of a substation 16 leaving the system
to reconfigure itself.

When a station such as 162 (Figure 1) operates normally,
it transmits on the link 10, towards the station 163,
signals corresponding to a regenerated version of the signals that it receives on its input coming from the upstream station 161.

At the same time, these signals coming from the station 161 are transmitted downstream of the station 162 bypassing the latter.

The station 163 thus receives two versions of the same signal which are normally shifted in time as a function of the delay imposed by the electronics of the station 162 to the regenerated signals with respect to the signals which have passed directly along the optical link 10 and 35. However, because of the optical transmission losses, the signals having traveled the path between the stations 161 and 163 without regeneration, by 10 and 35, are substantially more attenuated than the signals regenerated at the station 162. Consequently, in the case of signals digital, it is expected to detect the signals received at the station 163 by means of a threshold detector whose level is set so as to eliminate the signals transmitted bypassing the station 162 while collecting those which have been regenerated in the latter The detection threshold is made variable so as to adapt not only to possible variations in the regenerated signal level at the output of the station. but also to allow, in the event of failure of the latter, the detection of attenuated signals transmitted bypass directly from the station 16. Accordingly, it is intended to adjust the level of the signal detection threshold according to the peak level of these signals.

The detection threshold is located approximately equal distance between the received level of the regenerated signal and that of the signal which is not. The noise margin of the signal received at 163 is therefore lower in the case where the attenuation between 16 and 162 is low. If it is known, for example, that the amplitude of the light pulses transmitted in shunt of the station 162 is more than half less than the amplitude of the light pulses emitted by the station 162, and for example attenuated by -8 dB with respect to the latter, the detection threshold at station 163 is fixed at -2.5 dB. In this case, when the station 162 operates in direct transmission or in regeneration, only the signals of amplitude greater than -2.5 dB of the peak level detected at the station 163 will be received as corresponding to a high level. The light pulses exits from the station 161 and admitted in derivation of the station 162 are not "seen" by the reception station 163 and are received as low levels. If the station 162 fails, only the station 163 receive the signals emitted by the station 161 and having passed through the bypass of the station 162. The peak level of these signals is very significantly lower than the level of the signals received in this station. post before the breakdown. This new peak level is detected at station 163 which accordingly adjusts the threshold level for the detection of significant pulses in the received signals so as to enable the message transmitted along line 10 to be read despite the failure.

FIG. 3 illustrates the constitution of a digital signal repeater station 50 connected between the input and output couplers 22 placed on the optical link 10 and united, as in FIG. 2, by a bypass path 35.

The signals at the output 26 of the coupler 20 are converted into electrical signals by an opto-electronic converter 52. These signals are amplified in an amplifier 54 and then applied to a low-pass filter 55 (time constant of the order of magnitude the duration of the pulse) whose role is to eliminate the high frequency spurious fluctuations that may affect each of the pulses collected at the output of the converter 52. A high-pass pre-emphasis filter 56 is connected to the output of the filter the output 57 of this filter 55 is applied, on the one hand, to an input 58 of a comparator 60 and, on the other hand, to an input 59 of a low-pass filter 62, of greater inertia than the filter 55, sized to reach a stable output level corresponding to the peak value of the pulses received at the end of a train of five successive pulses, for example. The usefulness of the filter 62 is to eliminate the H.F noise residues, and the medium frequency noise that the low-pass filter 55 passes, so as not to excessively distort the peak detection. An amplifier located at the output 63 of the filter 62, can lower the output impedance to make it compatible with the following components; this output is transmitted to a peak detector 65.The latter conventionally comprises a rectifier such as diode 64 and a resistance circuit 68 capacitor 66 connected in parallel between the output of the diode 64 and the ground M. The output of the attenuator 71 is connected to a second input 70 of the comparator 60. The output 72 of this comparator may be coupled by a switch 74 to either an operating circuit associated with the station 50 by a disk 75, or directly to the input 76 of an opto-electronic converter 78 adapted to retransmit regenerated optical pulses at its output 79 on the input 32 of the output coupling T 22.

The resistor 68 is strong so that the discharge time constant of the resistor circuit 68 / decoder / ssoit very high while its load is faster and determined by the time constant of the filter 62. The output 69 of the peak detector 65, is connected to an attenuator 71 constituted for example by a voltage divider. In this example, assuming that the gain of the circuit between the output 57 of the shaping filter 56 and the input of the attenuator 71 is 1 (O dB), as well as between the output 57 and the input 58, one chooses the gain of the attenuator equal to -2.5 dB according to the previous example, and close to -3 dB in the general case where the derivative signal is significantly lower than the other.

FIG. 4A shows a pulse train A1,
A2 ... modulated in amplitude by all or nothing as a function of time from an initial moment. It is assumed that these pulses are a regenerated signal emitted by a station similar to the station 50 and placed upstream of it on the link 10.

FIG. 4B shows a train of pulses B1, B2... Admitted in a derivation of this upstream station on the link 10. These pulses follow one another in a sequence similar to that of the pulses of FIG. 4A, but in advance on it and with a much lower amplitude than the train of Figure 4A which has been regenerated by the upstream station. This regeneration resulted in a delay in the retransmission of the corresponding optical pulses on the link 10 with respect to the pulses of FIG. 4B.

The signals present at the input of the coupler 20 consist of the superposition of the signals of FIGS. 4A and 4B.

In FIG. 4C, the output signal of the attenuator 71 is represented. It comprises a series of ramps C1, C2, C3 corresponding to the successive pulses A1, 2, A3, etc., if we neglect the 4B pulses because of their very low level in this example compared to that of the pulses of Figure 4A.

At the end of the fifth pulse A5 (instant t1) the output level of the filter 62 is stable. The level of the peak detector 65 is maintained at a corresponding value and the output signal of the attenuator 71 is therefore maintained at a stable value S, during the arrival of the following pulses A6, A7, lower than their value. A pulse D6 or D7 (FIG. 4D) therefore appears at the output 72 of the comparator 60 for each pulse A6 or A7 on its input.

The level S is much higher than that of the attenuated pulses of FIG. 4B so that when one of these pulses such as B7 is interspersed at least partly between two pulses A6 and A7, the corresponding signal on the input 58 comparator 60 is less than the threshold level S reached by the output of the attenuator 71 on the input 70 of this comparator. Signals such as B7 are therefore eliminated as noise by this comparator and can not produce at the output 72 thereof a signal corresponding to the received information and allowing the emission of regenerated pulses.

In case of failure of the station located upstream of station 50, the pulses of FIG. 4A disappear. Only the pulses of FIG. 4B remain on the input 25 of the coupler 20.

The peak level 82 of these pulses is much lower than the peak level 81 of the pulses A1, A2, etc. Consequently, from the moment t2 when the upstream station stops operating, the output signal of the detector peak 65 begins to decrease causing with it the decay of the signal 83 at the output of the attenuator 71. This decay occurs for example at a speed about 1000 times lower than the rise in amplitude of the peak detector, if one wants to tolerate, in normal operation, a lack of pulses for a hundred elementary durations. Consequently, after a sufficiently long time, the output signal of the peak detector 65 stabilizes again, but at the level 82 of the pulses B. At this level 82 there corresponds a threshold level 84 for the signal 83 at the output of the attenuator 71. From the moment when this level is reached in steady state (time t3) each pulse Bn received on the input 58 of the comparator 60, causes the appearance on its output 72 of an accompanying signal Dn crossing the threshold level 84. Accordingly, despite the failure of the upstream station, the detection can continue.

 It is also possible to use the level of the output signal of the attenuator 71 to perform a failure detection of the station located upstream of the station 50 on the link 10, compared with a predetermined fixed threshold level.

This indication may be used to avoid taking into consideration during processing the signals corresponding to the transient period t2-t3 created by the failure.

The system just described can operate for a monodirectional or bidirectional optical link.

In the second case, it is possible to operate with a light having the same spectral characteristics in both directions, for example a white light by using Y couplers whose base connects at 26 and 32 and each branch of which feeds or is powered by 52 or 79 of each lane.

It is also possible to operate with signals having different spectral characteristics in each direction.

In FIG. 9, a first diffraction grating or a prism or other spatial wavelength separator 100 is placed on the link 10 making it possible to deflect a portion of the energy reaching it at the wavelength in a first direction in which an optical link 102 directs the transmitted light to an optional filter 103, which can be used to reduce inter-channel crosstalk. The rays transmitted by this filter are picked up by a receiver 105 whose output can be coupled to a transceiver 106 with bandpass filter around Xf to reinject the signals at the same wavelength A 1 on the optical link 10 by the intermediate of a second spatial spacer of lengths blonde 110 suitably oriented.The separators 100 and 110 are transparent to the signals transmitted on the bypass ramp 35 which allows the wavelength 7X1 1 to shunt the transceiver station 105, 106. A second path is provided which comprises an optical link 112 connected to the separator 110, enabling the recovery in this link 112 of signals that are predicted on the opposite side of the onden link on the link 10. The link 112 is equipped with a filter 113 passing only the rays of wavelength 2 and whose output is connected through a receiver 115 to a transponder 116 allowing reinjection ct optical signals to the probe length 2 on an optical channel 118. The latter leads to the separator 100 allowing the retransmission on the channel 10 of the signals of length? z2 in the opposite direction of the long-term signals
2 donde x1, previously mentioned. The signals at the two wavelengths each transit in their respective direction on the bypass pathway 35.

An extension to a larger number of wavelengths is obviously possible, with various transmission directions.

The system which has just been described for an optical link may be generalized to other types of link, and in particular to electrical links in situations where the signals transiting on such links are likely to undergo significant attenuations requiring, in particular, use of repeaters. In addition, it goes without saying that the components and circuits mentioned here by way of example for performing the regeneration of the optical signals can be replaced by any other device for performing the same functions and specifically to adapt the detection threshold of a receiver at the peak level of the signals it receives.

According to another aspect of the method described here, it is possible to detect non-digital but analog signals in each station. We consider a specific sequence of signals emitted by the station 161 (FIG 1). These signals are picked up by the station 162 and normally regenerated for their transmission to the station 163. The latter receives, together with the regenerated sequence, a sequence of weakened signals transmitted bypassing the receiver 162. In order to clear, in the station 163, the signals regenerated in the station 162 of these weakened signals which are the regenerated signal precursors subsequently received by the station 163, the signals received at the station 163 are delayed at a first instant and the received signals are subtracted at a second instant after having attenuated them by a suitable factor.

This can be achieved for example by means of a suppressor circuit as shown in FIG. 5 which integrates into the receiver 40 of FIG. 2. After conversion into electrical signals, the analog signals received in the receiver on FIG. a line 120 are sampled and admitted to the input 121 of an analog memory, such as
CCD (charge transfer device) 122; Or other. This device 122 comprises a plurality of successive cells within which the signals are transferred in series between its input 121 and its output 123 at the rate of a clock H. The rate of this clock and / or the number of cells of the device 122 are selected so as to establish a certain delay between the signals present at the output 123 with respect to the signals received on its input 121 and to have a sufficiently high sampling frequency so as not to lose information as to the shape of the analog signal.

It is also expected to connect to the input line 120 in parallel with the CCD device 122, an attenuator 125 whose output is connected to an input 126 of a differential amplifier 129 whose other input 129 is connected to the output 123 of the CCD device. The output 130 of the amplifier 128 delivers an analog signal which comprises the regenerated information in the station immediately upstream on the optical channel, cleared of the noise constituted by the optical signals having shunted from this upstream station on the link. 10.

In Figure 6, there is shown an analog signal received at 163 and having passed through the shunt 162 (curve 152).

The curve 154 represents the same signal received with a delay due to its passage through the repeater 162. The curve 150 is the sum of 152 and 154 and corresponds to the signal actually seen. For the sake of clarity, these analog signals are represented by a series of steps; the delay due to regeneration has been taken here equal to 2 steps = 2, fool.

The signal 160 corresponds to the CCD output applied to the input 129 of the differential amplifier 128; it corresponds to 150, delayed by 2bt.

The signal 162 is attenuated by the ratio between the main signal and its precursor echo 152. It therefore has the same amplitude as the "precursor echo" with the second-order distortions. By subtracting this reconstituted echo 162 from the composite signal that has lagged late 160 with respect to this reconstituted echo, we recreate relative provisions of 160 and 162 identical to those of 150 and 152. By difference we thus obtain 165 at the output of the differential amplifier, image very close to 154 which is the desired signal (at an offset of 26t near).

This method can be extended to a larger number of delay cells to improve filtering. Similarly, the delay can be adjusted dynamically by the signal processing device.

In the case of operating on digital signals which are relatively attenuated along the link between successive stations, it may happen that the signals admitted in derivation of one of these stations are of a non-negligible level. the peak level of the regenerated signals, for determining the peak level of the signal combination reaching the downstream station. The peak level of these signals may in fact be constituted either by the peak level of the regenerated signals alone, or by the sum of the peak levels of the regenerated signals and the signals transmitted by shunting depending on whether the pulses of the latter signal fall or not in intervals, or spaces, between pulses of the regenerated signal.

 It may then be important, to properly set the detection or analysis threshold of pulses received in a station along the transmission path, to know which of these two situations applies, or to take action. to prevent this knowledge from being necessary. In this respect, a first solution consists in weakening the signals admitted bypassing an upstream station sufficiently with respect to the level of the signals regenerated in this station, so that the value of the peak level of the weakened signals does not raise any difficulty as to setting the threshold of analysis of these signals in the post downstream.

Another solution to solve the problem of setting the analysis threshold in each of the stations is to arrange for the difference in propagation time between the optical branch path and the regeneration channel to be set in such a way that the two paths have their signals superimposed entirely or not at all.

Thus, for example, it can be arranged that the delay between the shunt-transmitted pulses and the corresponding regenerated pulses is substantially equal to an integer multiple of the pulse period forming the digital signals to ensure a superposition of some regenerated pulses and weakened. Conversely, if the duration of each pulse is less than half the period of these pulses (duty cycle less than 50%), it can also be arranged so that the delay mentioned above is such that a regenerated pulse falls always between two weak impulses transmitted bypass a half-period (to a whole number of periods).

When, because of the variable characteristics of the transmitted pulses, and in particular of their periodicity, of their duty cycle and of the drift time of the retransmitters, it is not possible to know in advance if the peak level of pulses received at a station does or does not correspond to the sum of the peak levels of the pulses regenerated by the immediately upstream station and pulses admitted in derivation thereof, one can use the technique described below.

At the beginning of transmission, a typical signal sample of the information transmitted by the optical link is considered before sending the messages themselves. This signal is received at a station 16 along this link. It comprises pulses regenerated in the station immediately upstream and pulses transmitted in derivation thereof and whose level is somewhat attenuated compared to that of the regenerated pulses. The received signal is sampled, preferably digitally, by a fast analog-to-digital converter. The sampling frequency is sufficiently high and the number of samples sufficient so that even if the resulting signal remains on a plateau value only for a brief moment, for example, in the case where the rising transition of a weakened pulse is followed almost immediately by the rising or falling transition of a regenerated pulse, several samples of this signal portion are obtained. In addition, the sampling is preferably performed asynchronously with respect to the rate of information conveyed by the received pulses, so as to avoid favoring certain states. In particular, this makes it possible to avoid that the samples on the edges of the received signal which are formed by relatively steep portions of the signal representative curve connecting successive steps, fall to all identical values.

The sample values corresponding to the entire sampled signal are stored and analyzed to deduce a histogram 201 shown in Figure 8A. The curve 201 is obtained by classifying all the samples representative of the digitized signal interval by increasing values starting from a first sample N0 of practically zero value these samples are marked from Ng to Nm on the abscissa axis by order of increasing rank. Sample N1 has a value greater than that of N0. Similarly, the sample N2 has a value greater than that of N1 and so on. Thus, a uniformly increasing curve 201 of the sample Ng to the sample N is obtained which represents the histogram of the distribution of the samples. levels of the samples Ng to sm taken on the selected reference signal portion. This curve has four steps, respectively 200 for a series of samples near the zero value, 202 for samples in the vicinity of a value A corresponding to the peak level of the weakened pulses, 204 for samples substantially equal to one value. B corresponding to the peak level of the regenerated pulses and, finally, 206 for samples with a value substantially equal to A + B corresponding to the sum of the peak levels A and B. Thus, for each sample of rank P carried along the axis ahscisses corresponds a value Sp ordinate on the curve 201.

Analysis of this histogram 201 makes it possible to determine the number of steps such as 202, 204 and 206 and their level in order to know the peak level of the received signal and suitably determine the setting of the analysis threshold of the signals received by the station. . The detection of these bearings is carried out as follows.

Each sample Np is considered in turn and is successively compared to all the other samples No to Nm. The result of this comparison is noted by defining a range centered on NP so as to accept a certain tolerance for the noises that may have been detected. assign the histogram 201, during the measurements. In practice, it is thus determined whether the sample considered is between tNp and NP (being less than 1 and for example equal to 0-, 9), or if it is between NP and d Np with d greater than 1 and for example equal to 1.1).

This process can be implemented for example using the circuit of FIG. 7. The numerical values of the samples are placed in memory 210. A counter 212 makes it possible to read the value of each sample from the corresponding memory position for the first time. register in a buffer memory 214 whose output is coupled to the respective inputs 215 and 216 of two multipliers 217 and 218 which multiply the value C in the buffer memory respectively by a coefficient 4 less than 1 for the multiplier 217 and a coefficient d greater than 1 for the multiplier 218.

For each value of C in the buffer memory, this counter 212 successively addresses all the positions of the memory 210 to extract the values D from the respective samples and send them to inputs 2221, 2231, and 2241 of three comparators 222, 223 and 224. On the second inputs 2222, 2232 and 2242 of these three comparators, the signals from the multiplier 217 (p C), directly from the buffer memory 214 (C) and the multiplier 218 (GC) are respectively received. The output of the comparator 222 is connected to the input of an AND gate 226 whose other input is coupled through an inverter 227 to the output of the comparator 223. The output of the latter comparator is also applied to the input d an AND gate 228 whose other input receives the output of the comparator 224. The outputs of the AND gates 226 and 228 are respectively coupled to a counter 230 and to a counter 231.

The comparator 222 produces a signal on its output, opening the AND gate 226, when the value D is greater than the value> C. The comparator 223 produces an output signal opening the AND gate 228 and closing 226 when D is greater than C and conversely and the comparator 224 produces a signal on its output opening the gate 228 when D is less than> C.As a result, when the contents of the memory 210 are scanned by the counter 220, the counter 230 counts all the samples D of which the value is defined by the inequality:
D <D # C (I) while the counter 231 counts all the samples D that satisfy the inequality
C / D o C (II)
The case D = C can obviously be related to 230 or 231.

The count levels reached by the counters 230 and 231 for each value of the sample C are stored in memory. These values are analyzed when buffer 214 has successively received the contents of memory 210 for all values C.

FIG. 8B plots the curves representing the counting levels 240 and 241 obtained at the outputs of the counters 230 and 231 for each of the values of the samples.
N0 to Nm. It has been chosen to orient the counting levels 240 downwards of FIG. 8B with respect to a counting level axis 211 along which the samples are indicated by their rank from N0 to Nm and towards the the upper part of the figure for the curve 241. It can be seen that for the bearings 200, 202, 204 and 206, the curves 240 and 241 have lobes which each have a maximum substantially in the middle of the respective bearing but which have a marked dissymmetry with respect to to the vertical axis passing through the middle of each plateau, such as 2llbis. This dissymmetry reflects the fact that, for example, for a sample of value Np to the right of the middle of the plateau 202, there are a larger number of samples of which the value is between ssNp (ss <1) and Np only samples whose value is between Np and Np. It is assumed here that the deviations enter etc. respectively and the value 1 on the other hand are equal. This dissymmetry is used to accurately detect the sample values which correspond to the maximums of the lobes of the curves 240 and 241. According to one embodiment, this detection is performed by calculating for each sample value along the length of the sample. 211, a number whose value is of the form P = - z ss and P = o if 5 -Zz O
In the example, we took Z = 10, which gives
P = Z - 10h where is the sum of the absolute value accounts of the curves 240 and 241 for each level of sample C considered and is the difference between these accounts (in absolute value).

Any other formula that results in pinching the lobe may be appropriate. The representative curve of the magnitude P is illustrated at 252 by the lobe corresponding to the plateau 206. It consists of a very marked maximum lobe at 253 which corresponds to the value of the sample N q for the middle of the plateau 206.

An analogous peak of curves 254, 256 and 258 representative of P is detected for the bearings 204, 202 and 200 respectively.

The value of the sample corresponding to this peak is equal to the level of the respective step.

Analysis of the sample values corresponding to the peaks of the curves 252, 254, 256 and 258 makes it possible to determine the number of steps. Peak 253 is recognized as the largest and eliminates subsequent comparisons, the values of
P relative to samples of amplitude close to that relative to the peak 253. The judicious choice of this range, which can be taken from> to times the value of the sample relative to 253, eliminates all the main lobe P . We search for the most important lobe remaining and proceed in the same way.

The detection threshold is set halfway between the first non-zero level (signal passing through the shunt) and the second level (regenerated signal).

Claims (11)

Claims. A method for transmitting signals over a link, in particular an optical link (10), along which reception stations equipped with means for retransmitting the received signals on said link are distributed, characterized in that a portion of the energy transmitted on said bypass connection (35) of each station (162) to the downstream station (163), so that the signals received by the latter are normally formed by the superposition of signals retransmitted by an upstream station (162); ) and signals accepted in derivation thereof and the signals received by the downstream station (163) are processed to remove them at least in part from said signals transmitted by shunting. 2. Method according to claim 1, characterized in that the processed signals are used to re-transmit the signal received on said link in a regenerated form. 3. Method according to claim 1 or 2, for the transmission of digital signals, characterized in that the peak level (81) of the signals received at the downstream station (163) is detected and the only signals selected from among them are selected. whose level is greater than a threshold (S), set according to said peak level. 4. Method according to claim 3, characterized in that the downstream station is detected if the peak level is constituted by the superposition of the peak levels of the signals transmitted by the upstream station and signals admitted bypass, to set the threshold detection. 5. Method according to claim 1 or 2 for the transmission of analog signals, characterized in that subtracted from the amplitude of the signal (150) received at a first instant a fraction of the amplitude of the signal (154) received at a second instant shifted from the first instant of a predetermined time interval to clear the received signal of the signals transmitted bypass from the previous station. 6. Received digital signals transmitted on a link, including an optical link (lu), characterized in that it comprises a detector (65) of the peak level of the received signals and a comparator (60) suitable for comparing the the level of the received signals with a reference signal fixed to a predetermined fraction of the peak value detected by the peak detector, to enable the signals received on this link to be eliminated the signals having passed through a bypass (35) an upstream station (16) on this link whose level is attenuated with respect to that of the signals transmitted by this upstream station. Receiving station according to claim 6, characterized in that the comparator comprises a first input (58) connected directly to the input of the station (50) and a second input (70) connected to the peak detector via an attenuator (71), at the input of this peak detector being connected to the input of the station, through at least one low-pass filter (55). 8. Hece-psc-eur station for analog signals transmitted over a link (10), in particular an optical link, to which this station is connected, characterized in that it comprises a suppressor (FIG. 7) capable of subtracting from the detected signal at a previous time a predetermined fraction of the signal received at the current time, the time offset being a predetermined time interval statically or dynamically. 9. Station according to claim 7 or 8, characterized in that it comprises a charge transfer memory (122) for shifting, and thus delay in its output of the duration of said time interval, samples of the received signals, a attenuator (125) for producing a correction signal whose level is a fraction of the level of the signals received at the current time and a differential amplifier (128) for subtracting the correction signal from the samples from the charge transfer memory. 10. Installation for transmitting signals over a link, in particular an optical link, characterized in that it comprises at least one upstream station on this link comprising means for receiving signals transmitted on this link and means for transmitting signals. on this link suitable for re-transmitting the received signals in a regenerated form, a ramp for the permanent transmission of a fraction of the energy bypass of this station, and at least one downstream station comprising receiving means suitable for ridding at least in part the signals it receives from the influence of the signals having passed through the ramp of the branch of the upstream station or the previous station in the case of failure of the transmitter of this upstream station. 11. Installation according to claim 10, characterized in that said means for receiving the downstream station comprise a peak detector comprising, downstream of a digital analogue converter, signals received by the downstream station, a memory and means adapted to storing the successive digital samples at the output of the converter in this memory means for re-reading said samples and process them to detect the maximum level of plateau of said samples, means being further provided for discriminating the signals received by the downstream station; function of said maximum level.