FR2715254A1 - Soliton coding method for optical fibre transmission - Google Patents

Abstract

The method includes dividing a time period (T) into three equal intervals (I1,I2,I3). A stream of solitons is introduced into these intervals according to a coding law. If the binary data to be transmitted is a logic 0 then solitons are not placed in any interval. If it is logic 10 and is preceded by logic 0 or 11, the stream is placed in the first interval. If the data to be transmitted is binary 11 and is preceded by logic 0 or 11,the solitons are placed in the second time interval. The third interval is never filled. The time interval following each soliton-filled time interval is always empty.

Description

METHOD OF OPTICAL TRANSMSSSION BY SOLITONS, OORRESPONDING TRANSMITTER AND RECEIVER
DESCRIPTION
Technical area
The present invention relates to an optical transmission method by solitons, as well as a transmitter and a receiver implementing this method.

 The invention finds an application in optical telecommunications.

State of the art
The technique of transmission by optical fibers is well known. It has many advantages.

However, it is limited towards high speed and long distances by the phenomenon of index dispersion. However, this disadvantage could be remedied by the use of non-linear effects adjusted to compensate for this dispersion. Certain optical modes, called "solitons", can then propagate without distortion over long distances. These modes have been described by LF MOLLENAUER et al. in
Physical Review Letters 45, 1095 (1980).

 More particularly, the optical transmission technique by solitons is described in particular in the article by L.F. MOLLENXUER et al. entitled "Experimental Study of Soliton Transmission over more than 10000km in Dispersion-Shift Fiber", published in the journal Optics Letters, vol. 15, 1203-1205 (1990).

 The non-linear nature of optical transmission by solitons generates particular constraints that are not found in conventional optical transmission. In particular, it is imperative that the light pulses constituting the solitons are sufficiently spaced not to overlap and interfere. This is one of the reasons for using a “return-to-zero” coding (RZ for short), which guarantees minimum spacing between two successive light pulses.

The usual technique of transmission by solitons is illustrated in Figure 1 attached. We see there, line
A, a sequence of binary data (or bits) to be transmitted, in this case the sequence 01011. On line B, we see the corresponding sequence of signals coded in RZ. The pulses shown have a duration equal to one fifth of the period T separating the bits to be transmitted. In other words, the filling rate is equal to 20%. A soliton (line C) is matched to each of these pulses. These solitons are therefore at least spaced 4T / 5 (case of two successive 1s, as at the end of line B).

 We can consider the sequence of pulses RZ of line B as forming pulses of the non-return to zero type (NRZ for short) provided that reference is made to a period T / 5. In fact, throughout this fictitious period, the signal retains its polarity. Line D shows the sequence of bits, also called "over-sequence", which gives the same sequence of solitons as the pulses RZ of line B.

 The interest of the concept of excess frequency (line D) is to show that the spacing between solitons is represented by the number of "0" that presents the excess frequency between two successive "1". This number has a minimum value which will be called hereinafter "minimum distance" and will be noted "d". In Figure 1, we have d = 4.

The coding technique for transmission by solitons illustrated in FIG. 1 is advantageous in certain respects, in particular from the point of view of the simplicity of implementation, but it has at least three drawbacks
a) As already pointed out, due to the non-
linear transmission, spacing
minimum time is required between two
successive solitons. For transmission to
long distance (a few thousand
kilometers), an order of magnitude of the distance
minimum is 4 (which is what is
shown in Figure 1). This constraint can be
satisfied, for a given online speed, by
the use of narrow width solitons (therefore
strong power). However, the reduction in
width of the solitons results in a
share, an increase in the level of jitter, and,
on the other hand, an increase in sensitivity
imperfections in the transmission line. A
order of magnitude of the width, offering
currently an interesting compromise vis-à-vis
of the set of constraints of a system, is
20 ps. Online throughput by wavelength
is then limited to approximately 10 Gbit / s. This
inconvenience, which results in bad
filling of the transmission channel, is, in
fact, the major drawback of this coding
classic.

b) As in other transmission systems,
RZ or NRZ codes require presence
capable scramblers-descramblers
eliminate data sequences without
transitions (in this case the sequences of "0"
consecutive), harmful to
synchronization.

c) Transmission by solitons addressing, to
less in the short term, at high speeds (several
Gbit / s), the RZ code requires, on reception,
very wide band amplifiers (several GHz)
theoretically capable of passing the continuous, or at
less having a very low cut-off frequency
at low frequencies, depending on the length of
sequences of successive "0" or "1" allowed.

The latter requirement can lead to
significant technological difficulties.

 To avoid these difficulties, one can think of being inspired by studies carried out in a completely different technical field, that of codes for digital magnetic recording. These codes are called "Runlength Limited" or limited current length. They are noted RLL for short. Such codes are described, for example, in the article by B.H.

MARCUS and PH SIEGEL, entitled "Finite-State
Modulation Codes for Data Storage ", published in IEEE
Journal on Selected Areas in Communications ", vol. 10, no. 1, January 1992, p.5-37.

 An RLL code is defined by the minimum number of "0" separating two successive "1", number noted d, and by the maximum number of "0" being able to separate two successive "1", number noted k. An RLL code can therefore be noted (d, k).

 However, the RLL codes known in the digital recording technique do not make it possible to solve the problems specific to transmission by solitons. In particular, they are generally too complex for such a technique, which requires great simplicity because of the bit rates involved.

These bit rates are 100 to 1000 times higher than in the case of digital recording and there is no question of using known RLL codes, which would require coders and decoders that are much too complex, therefore too slow and subject to errors.

 The object of the present invention is precisely to remedy these drawbacks by proposing a transmission method based on a very simple code. It is indeed an RLL type code, but specially adapted to the problems of transmission by solitons.

The code of the invention makes it possible to increase the information rate, at a given soliton width, (or to reduce the width of the solitons at a given rate).

The invention leads to a very simple and very reliable coder and decoder.

Statement of the invention
The method of the invention takes up the general scheme of any method of transmission by solitons.

In this sense, it includes the following operations
A) on emission
- we start from binary data to be transmitted formed
a series of bits equal to "0" or to "l",
- clock periods of duration are constituted
determined,
- we form a train of solitons by placing at most
a soliton in such periods according to a law
determined coding,
- we emit this train of solitons;
B) at the reception
- we receive the train of solitons,
- we reconstitute clock periods similarly
duration of transmission,
- we detect the presence or absence of solitons
in the different periods,
- by a law of inverse decoding of the law of
coding used at the emission one restores the continuation
binary data.

 According to the invention, this transmission method comprises a first characteristic according to which the data are not coded one by one according to their value "0" or "1", but by considering them in pairs (or doublets). Furthermore, the coding depends on the bit or bits which precede the torque to be coded. It is therefore not an absolute code in the sense that, for example, the doublet "11" taken at random from a series of binary elements, is coded differently according to the nature of the sequence which precedes it. Finally, the coding law used is very simple and makes it possible to increase the bit rate.

More specifically, the method of the invention is characterized in that
A) on emission
a) each clock period is subdivided into one
first, second and third intervals,
all of duration equal to a third of the period
clock,
b) we place, in these intervals, solitons
according to the following coding law
- if the binary data is a "0", we do not place
no soliton in any interval
either a period,
- if the binary data is a couple "10" and
if this couple was preceded by a "0" or a
couple "11", we place a soliton in the
first interval of a clock period,
we broadcast this period, and we forward it
of a period without soliton,
- if the binary data is a pair "11" and if
this couple was preceded by a "0" or a
couple "11", we place a soliton in the
second interval of a clock period,
we broadcast this period and we forward it
of a period without soliton,
B) at the reception
- the presence or absence of soliton is detected
in the first and second intervals of
each clock period, and we restore the following
binary data according to this presence or this
absence.

Preferably
-if we detect a period without soliton, we
returns a "0",
-if we detect a soliton in the first
interval of a period and no soliton in the
period which follows, a couple "01" is restored if
we previously detected an "O" or a pair 1 1 n,
- if a soliton is detected in the second
interval of a period and no soliton in the
period which follows and if one has previously detected
a "0" or a couple "11", we restore a couple
"11".

 Thus, for a clock period of value T, we obtain solitons of width T / 3 corresponding to the intervals into which the period has been divided, and the distance between two solitons is at least four intervals (case where two couples " 11 "follow each other), a duration of 4T / 3. In the prior art, for the same duration T, there was a soliton duration of T / 5 and an interval of 4T / 5.

 At constant pulse width, the information rate in the present invention is therefore equal to 5/3 of the rate obtained in the prior art. We therefore obtained a gain of 1.67 on the flow.

 Conversely, at constant flow, the width of the solitons is equal to T / 3 instead of T / S, or a gain of 5/3 = 1.67.

 The coding law which has just been defined can be implemented directly optically by acting on an appropriate soliton source. However, it can also be implemented indirectly electronically. In this second case, there are already produced logical electrical signals formed by periods of duration T subdivided into first, second and third intervals of duration one third of the period, and these intervals are put at a logical level n 1 n where you want to place a soliton in the soliton train. Using these logic signals, optical gates which allow solitons to pass are then controlled.

The present invention also relates to a transmitter for implementing the method which has just been defined. Like any transmitter usable in transmission by solitons, that of the invention comprises
- an input capable of receiving binary data
formed of a series of bits equal to "1" or to "0",
- clock means capable of constituting
fixed term periods,
- optical means capable of delivering solitons,
- coding means capable of placing these solitons
in the periods according to the information to
transmit and according to a coding law
predetermined.

The transmitter of the invention is characterized in that the clock means are capable of subdividing each clock period into a first, a second and a third interval, all of the same duration equal to one third of the clock periods and by the fact that the coding means are able to arrange the solitons as follows
- if the binary data is a "0", the means of
coding do not place any soliton in
whatever interval of a period
clock,
- if the information is a couple "10" and if this
couple was preceded by a no or another
pair "11", the coding means place a
soliton in the first interval and don't place
no soliton in the following period,
- if the information is a pair "11" and if this
couple was preceded by a n o or another
pair "11", the coding means place a
soliton in the second interval and don't place
no soliton in the following period.

Finally, the subject of the present invention is a receiver for implementing the method which has been defined. This receiver includes, like any soliton transmission receiver
- an entrance receiving a train of solitons,
- clock means for reconstructing
reception clock periods of the same duration
that on the show,
- means for detecting the presence or absence
of solitons in the different periods
clock,
- a decoder to restore the rest of the data
binary transmitted.

The receiver of the invention is characterized in that
- the clock means are able to subdivide
each clock period first, one
second and third intervals, all of
same duration equal to a third of the period
clock,
- the means to detect the presence or absence
of solitons are able to detect this presence
or this absence in each of the first and
second intervals of the clock periods.

Brief description of the drawings
- Figure 1, already described, illustrates the technique
known optical transmission by solitons
- Figure 2 shows the division of each period
in three intervals;
- Figure 3 shows the result of the coding of a
pair "10" preceded by a "0" or a pair
"11";
- Figure 4 shows the result of the coding of a
pair "11" preceded by a "0" or a pair
"11";
- Figure 5 illustrates a borderline situation where the
number of "0" between two "1" is 4;
- Figure 6 is the state diagram of an encoder;
- Figure 7 illustrates an example of an encoder for
implementation of the invention;
- Figure 8 illustrates an embodiment of a
transmitter according to the invention;
- Figure 9 illustrates an embodiment of a
receiver according to the invention;
- Figure 10 shows an example of a decoder for
implementation of the invention.

Presentation of specific embodiments
In FIG. 2, we see clock periods of duration T, each subdivided into three intervals of duration T / 3, respectively a first (I1), a second (I2) and a third (I3) interval.

 According to the invention, the solitons are placed in the first or the second interval for the pairs "10" and "11" (preceded by "0" or "11"); no soliton is placed in any interval whatsoever in case of "O". We never place a soliton in the third interval 13.

 In FIG. 3, the place occupied by various logic signals and by the soliton emitted is seen when the data to be transmitted comprise a pair "10" preceded by a "0" or a pair "it". Line A shows a logic signal at level 1 during the interval I1.

This signal can be generated by a logic electronic circuit, as will be seen in connection with FIG. 6. Line B shows the corresponding binary over-sequence. This excess sequence includes a "1 n in the interval I1, then" 0 "in the intervals I2 and I3. The entire second period only includes" O ". We therefore have the excess sequence 100,000. Line C shows the soliton sl centered on the interval I1.

 FIG. 4 similarly shows the signals and the soliton for a pair "11" preceded by a no or a pair "11". The logic signal (line A) is at level 1 in the second interval I2 and the soliton s2 is emitted in this interval.

 In these two cases (Figures 3 and 4), the second period of duration T is empty of soliton. This means that three intervals without soliton necessarily follow a period containing a soliton. And since the third interval of a period containing a soliton is always empty, this implies that there are always at least four empty intervals after a soliton. In other words, there are always at least four "0" between two successive "1" in the over-sequence. But there may be more.

 The case where this number is precisely equal to 4 is shown in FIG. 5. The illustrated sequence is 010000100000. It corresponds to the data sequence 01011.

 To obtain the appropriate logic signals from the binary data stream to be transmitted, a logic circuit called an encoder must be made. This coder receives the binary data which will be noted "a" with an index k marking their rank. We will thus have successive data ak, akl, etc. The coder delivers three logic signals x, y and z with a rank k, that is xk, yk, zk. The signal xk relates to the first interval I1 of a period, the signal yk to the second interval I2 of this same period and the third signal zk to the third interval I3. This third signal is always at level 0.

 The coder therefore delivers a triplet xk, yk, zk which defines the state of each interval of a period as to the presence or absence of solitons. As the coding of the data must take account of the data previously received, it is also necessary to provide an auxiliary variable ck which defines the internal state of the coder. The triplet delivered xk, yk and zk does not depend in fact only on the data ak received but also on the data which have already been received, therefore on the state of the coder. As has already been said, the coding is not absolute.

Among the different ways of forming the signals xk, yk, zk, one can describe the following mode of implementation which is defined by the following coding equations Xk = ctaka * +, Yk = Zk = 0 Ck + l =
We can easily verify that if ak = 1 and ak + 1 = O and if ck = O then xk = 1, Yk = 0: level 1 is in the first interval I1.

 Similarly, if ak = l and ak + 1 = 1 and Ck = 0 then xk = O and yk = 1: level 1 is the second interval I2.

These coding equations can be obtained by the following reasoning
* at initialization, we assume that the signal
detected is zero. As long as a succession of two
"1" was not detected, (what we can
symbolize by a state ck = O of the encoder), the output
Xk is zero, except for the combination
"10" (which is represented by the first and the
single term of the equation giving x),
* if two successive "1s" are detected, the
triplet "010". In this case, we must take
arrangements to issue on the next stroke
triplet in000 ", regardless of the element
binary according to the detected pair "11", since
the scope of information of the triplet "010"
extends to the next symbol. It is therefore
necessary to memorize this detection by a
toggle encoder state. Insofar
where ck = O symbolized the absence of two "1s"
successive, ck + 1 = l translates the new state of the
coder. On the next move, we therefore issue xi + 1 = 0 (the
encoder in state c = î can only transmit xi + 1 = 0,
which implies that the equation of xk has no
second term) and yak + 1 = 0 thanks to the variable
representing the state of the encoder, and we switch back
encoder in its initial state, since the memory
of the system is limited to one symbol. The coder
is only in state 1 when we have just emitted what explains the identity of ck + 1 and Yk,
which simplifies the practical realization of the
coder.

 To obtain the minimum and maximum distances of the different possible over-sequences, the code can be presented in the form of a state diagram.

The branches of this diagram associate an input binary element with an output triplet.

 Figure 6 shows such a state diagram. The notations have the following meaning: the rectangles represent states of the coder and the number carried in each of these rectangles indicates the rank of each state. Thus, the number 5 in the bottom right rectangle represents the state N05 of the encoder. The arrows between two rectangles represent the passage from one state to another. The notation ak + 1 / xkykzk associated with the arrows on a branch connecting a state i to a state j means that the coder being in state i transmits the triplet Xkykzk when it receives the data ak + l. For example, if the coder is in state 2, it sends the triplet "100" if the data received is "0" and the coder goes to state 3. But, still from state 2, the coder transmits the triplet "010" if the data is "1" and the encoder goes to state 4.

 The loop which leaves from state "1" and which brings back there shows that, receiving a continuation of "O", the coder delivers only "0", in an unlimited way. The maximum distance (k) between two "1" is therefore a priori not bounded and can take any value.

FIG. 7 shows an example of an encoder capable of delivering logic signals conforming to the coding equations which have just been described. This C encoder includes
- a first flip-flop 10 receiving on an input D
the binary DB data to be transmitted. This
flip-flop has two complementary outputs Q and Q,
- a second flip-flop 12 whose input D is connected
at the output Q of the first flip-flop 10;
- a first AND gate 14, one entry of which is
connected to the output Q of the flip-flop 12;
- a second AND gate 16, one input of which is
connected to the output of gate ET 14
- a third door AND 18, one entry of which is
connected to the output of gate ET 14 and a
other input is connected to output Q of the
rocker 10;
- and finally a third flip-flop 20 whose input D
is connected to the exit of door 16. The exit
Q of this third rocker is connected to the
second entry of gate AND 14.

 The outputs Sx, Sy of the encoder are taken respectively at the output of the third AND gate 18 and at the output of the second AND gate 16. These outputs respectively deliver the signals xk and yk. The output Sz is directly connected to a permanent logic level 0 since the signal zk is always at zero.

It is easy to see that the encoder in FIG. 7 delivers the desired logic signals:
- if the input binary data is an "O", the
three outputs Sx, Sy, Sz are at the logic level
0,
- if the input binary data is a pair "10"
and if this couple was preceded by a "0" or a
other couple "11", the first Sx output is at
logic level 1 and the second and third
Sy, Sz outputs are at logic level 0,
- if the input binary data is a pair "Il"
and if this couple was preceded by a "0" or a
other couple "11", the second Sy output is one
logic level 1 and the first and third
outputs Sx, Sz are at level 0.

 A transmitter according to the invention comprises optical means controlled by the logic signals delivered by the encoder of FIG. 7. These optical means must be able to deliver a soliton in the first or the second interval of a period depending on whether one finds a logic level 1 on the first (Sx) or the second (Sy) output of the encoder.

FIG. 8 shows an example of such optical means. As shown, they include
an optical source 30 emitting a train of
solitons s of width substantially equal to T / 3,
- a coupler 32 to an input connected to the source 30
and two outputs respectively delivering a
first and second train of soliton sl, s2,
- a delay line 34 of duration T / 3 placed on the
path of the second train s2,
- a first optical door P1 with an optical input
receiving the first train of solitons sl, a
control input connected to the first Sx output
encoder C, and an optical output; this door
pass the soliton sl it receives if the
control signal xk which it receives from the encoder C
is at logical level 1,
- a second optical door P2 to an optical input
receiving the second train of solitons s2 delayed
via delay line 34, a command entry
connected to the second output Sy of the encoder C and a
optical output; this second door P2 leaves
pass the delayed soliton s2 it receives if the
control signal yk which it receives from the encoder C
is at logical level 1,
- an optical coupler 36 with two inputs and one
output, the two inputs being connected
respectively at the optical outputs of the
first and second optical doors P1, P2,
the output 40 of this coupler 36 constituting the
transmitter output. It is on this exit
that we find the train of solitons (S) which is
finally issued.

 The CD encoder and the source 30 are controlled by a transmission clock He.

On the receiver side, there are symmetrical means to those of the transmitter. FIG. 9 shows, for example, an embodiment of such a receiver. As shown, this receiver includes
- a clock Hr defining periods of duration T
subdivided into three intervals I1, I2, I3,
- an optical divider 50 at an input 52 receiving the
soliton train S transmitted, and three outputs
delivering three trains of solitons sl, s2, s3,
- a first control circuit CCî to generate
a first corresponding control signal SC1
at the first interval I1 of the clock periods,
- a second control circuit CC2 to generate
a second corresponding control signal SC2
at the second interval I2 of the clock periods,
- a third control circuit CC3 for
generate a third control signal SC3
corresponding to the third interval I3 of
clock periods,
- a first detection means D1 receiving the
first train of sl solitons from
divider 50, this first detection means D1
being controlled by the first control signal
SC1 and delivering a first signal xk whose
logical level translates presence or absence
of soliton in the first interval I1 of a
period,
- a second detection means D2 receiving the
second train of s2 solitons from
divider 50, this second detection means D2
being controlled by the second control signal
SC2 and delivering a second signal yk whose
logical level translates presence or absence
of soliton in the second interval I2 of a
period,
- a third detection means D3 receiving the
third train of soliton s3 from
divider 50, this third detection means D3
being commanded by the third signal from
command SC3 and delivering a third signal Zk.

 The receiver of FIG. 9 finally comprises a decoder DC, which receives the first and the second signal xk, yk and which restores on its output 62 the binary data DB according to the decoding law already indicated.

The clock signal delivered by the clock Hr of the receiver is affected by an ambiguity of order 3, insofar as nothing imposes that the first interval I1 of a reception clock period corresponds to the first interval I1 </ RTI
current signal xk and whose output Q delivers the
previous signal xk-1;
- a second flip-flop 72 whose input D is connected
to the second detection means D2 and receives the
current signal yk and whose output Q delivers the
previous signal Yk-1
- a first OR 74 logic gate, both of which
inputs are connected to the outputs of flip-flops 70
and 72 and which therefore receive xk-l and Yk-1
- a third flip-flop 76 whose input D is
connected to the output of the second flip-flop 72 and
therefore receives Yk-1;
- a second OR 78 logic gate including one input
is connected to the exit of door 74 and the other
at the exit of flip-flop 76 and receives Yk-2.

 The output of gate 78 constitutes the general output 62 of the receiver which restores the binary data DB from which one had started on transmission.

Claims (13)

 1. Method of optical transmission by solitons, in which A) on emission  - we start with binary data (DB) to be transmitted  formed of a series of bits equal to "0" or to "1",  - clock periods of duration are constituted  determined (T),  - we form a train of solitons by placing at most  a soliton (s) in such periods according to a  determined coding law,  - this train of solitons (S) is emitted; B) at the reception  - we receive the train of solitons (S),  - we reconstitute clock periods similarly  duration (T) that on transmission,  - we detect the presence or absence of solitons  in the different periods,  - by a law of inverse decoding of the law of  coding used at the emission one restores the continuation  binary data (DB), this process being characterized in that A) on emission  a) subdividing each clock period (T) into a  first (I1), second (I2) and third  (I3) intervals, all of duration (T / 3) equal to  third of the clock period (T),  b) placing, in these intervals (I1, I2, I3),  solitons according to the following coding law  - if the binary data (DB) to be transmitted is a  "O", no soliton is placed in  whatever interval of a period,  - if the binary data to be transmitted is a  couple "10" and if this couple was preceded  of a "0" or of a couple "li", we place a  soliton in the first interval (I1) of a  clock period, we issue this period and  it is followed by a period without soliton,  - if the binary data to be transmitted is a  couple "11" and if this couple was preceded  a "0" or a couple "11", we place a  soliton in the second interval (I2)  of a clock period, we issue this  period and it is followed by a period  without soliton, B) at the reception  - the presence or absence of soliton is detected  in the first (I1) and second (I2)  intervals of each clock period, and we  restore the binary data sequence according to this  presence or absence.  2. Method according to claim 1, characterized in that, on receipt  - if we detect a period without soliton, we  renders an "O",  - if we detect a soliton in the first  interval (I1) of a period and no soliton  in the period that follows, we restore a couple  "01" if one previously detected an "O" or a  couple "11",  - if a soliton is detected in the second  interval (I2) of a period and no soliton  in the period that follows and if one has detected  previously an "O" or a couple "11", we restore  a couple "11".  3. Method according to claim 1, characterized in that to implement the coding law, there are already produced logic control electrical signals formed by periods of duration T subdivided into a first (I1), a second (I2) , a third (I3) intervals of duration (T / 3) equal to a third of the clock period (T), and we put these intervals at a logical level 1 where we wish to place a soliton in the train of solitons.  4. Method according to claim 3, characterized in that  - we generate a periodic train (s) of solitons  (30),  - we divide this train into a first (sl) and into a  second (s2) trains,  - we address the first train (sl) on a first  optical door (P1),  - the second train (s2) is delayed by a duration (T / 3)  equal to a third of the period and we address this  second train delayed at a second optical gate  (P2),  - the first and second doors are ordered  optical (P1, P2) by logical levels (xk,  yk) appearing respectively in the first  (I1) and the second (I2) period intervals  logic control signals,  - we optically combine (36) the outputs of the  first (P1) and second (P2) doors  optics to form the train of solitons  issued (S).  5. Method according to any one of claims 1 to 4, characterized in that on receipt  - we divide the received train of solitons (S) into one  first (sl) and second (s2) trains,  - we address the first (sl) and the second (s2)  trains respectively on a first (D1) and a  second (D2) detection means,  - a clock signal of the same period (T) is formed  that at issue and we subdivide each period  of this clock signal into a first (11), a  second (I2) and third (I3) intervals,  these intervals being synchronized on the  intervals used during transmission,  - a first control signal (SC1) is formed (CC1)  in the first interval (I1) of each  period,  - a second control signal is formed (CC2)  (SC2) in the second interval (I2) of each  period,  - the first detection means (D1) is controlled by  the first control signal (SC1), which gives  a first logic signal (xk) linked to the presence  or the absence of a soliton in the first  interval (I1),  - the second detection means (D2) is controlled  by the second control signal (SC2), which  gives a second logic signal (Yk) linked to the  presence or absence of soliton in the  second interval (I2),  - we decode the first (xk) and second (yak)  signals and by the decoding law, which  restore binary data (DB).  6. Method according to claim 5, characterized in that the train of solitons received (S) is divided into a third train (s3), this third train is addressed to a third detection means (D3), we form (CC3) a third control signal (SC3), the third detection means (D4) are controlled by the third control signal (SC3) which gives a third logic signal (zk), the signal d is wedged (66) reception clock so that the output (zk) of this third detection means (D3) is always at logic level 0, which synchronizes the reception periods with the transmission periods.  7. Transmitter for implementing the method according to claim 1, this transmitter comprising  - an input suitable for receiving data  binary (DB) formed by a series of bits equal to  n or  - clock means (He) capable of constituting  fixed term periods (T),  - optical means capable of delivering solitons  (s),  - coding means capable of placing these solitons  in the periods according to the information to  transmit and according to a coding law  predetermined, this transmitter being characterized in that the clock means are capable of subdividing each clock period (T) into a first (11), a second (I2) and a third (I3) intervals all of the same duration equal to a third (T / 3) of the clock period (T) and by the fact that the coding means are able to arrange the solitons as follows  - if the binary data is an "on, the means of  coding do not place any soliton in  whatever interval of a period  clock (T),  - if the binary data is a pair "10" and if  this couple was preceded by a "0" or another  pair "11", the coding means place a  soliton in the first interval (I1) and does not  place no soliton in the period that follows,  - if the binary data is a pair "11" and if  this couple was preceded by a n O or another  pair "11", the coding means place a  soliton in the second interval (I2) and does not  place no soliton in the following period.  8. Transmitter according to claim 7, characterized in that the coding means (C) comprise  - an input (D) capable of receiving the data  binaries (DB) to transmit,  - a first (Sx), a second (Sy) and a  third (Sz) outputs,  - suitable logical means (10, 12, 14, 16, 18, 20)  to be delivered on the first (Sx), second (Sy)  and third (Sz) logic signal outputs  (xk, yk, zk), function of binary data (DB)  received on entry according to the following law  - if the data is a "0", the three outputs  (Sx, Sy, Sz) are at logical level 0,  - if the data is a couple "10" and if this  couple was preceded by a "0" or another  couple "11", the first output (Sx) is one  logic level 1 and the second and third  outputs (Sy, Sz) are at logic level 0,  - if the data is a pair "11" and if this  couple was preceded by a "0" or another  couple "11", the second output (Sy) is one  logic level 1 and the first and the  third outputs (Sx, Sz) are at level  logic 0.  9. Transmitter according to claim 8, characterized in that it comprises optical means controlled by the signals (xk, yk, yz) delivered by the coding means (C), these optical means being capable of delivering a soliton in the first or the second interval of a period depending on whether a logic signal is found at level 1 on the first (Sx) or the second (Sy) output from the coding means (C).  10. Transmitter according to claim 9, characterized in that the optical means comprise  - an optical source (30) emitting a train of  solitons (s) of width substantially equal to T / 3,  - a coupler (32) to an input connected to the source  (30) and with two outputs delivering respectively  a first (sl) and a second (s2) train  solitons,  - a delay line (34) of duration T / 3 placed on the  path of the second train (s2),  - a first optical door (P1) at one input  optics receiving the first train of solitons  (sl), a command input connected to first  output (Sx) of the coding means (C), and a  optical output,  - a second door (P2) to an optical input  receiving the second train of solitons (s2) delayed  via the delay line (34), an entry of  command connected to the second output (Sy) of  coding means (C) and an optical output,  - an optical coupler (36) with two inputs and one  output, the two inputs being connected  respectively to the two optical outputs of the  first and second optical doors (P1,  P2), the output (40) of this coupler (36)  constituting the output of the transmitter and delivering  the train of solitons (S).  11. Receiver for implementing the method according to claim 1, comprising  - clock means (H) for reconstructing  reception clock periods of the same duration (T)  that on the show,  - an input (52) receiving a train of solitons (S),  - means for detecting the presence or absence  of solitons in the different periods  clock,  - a decoder to restore the rest of the data  transmitted binaries, this receiver being characterized by the fact that  - the clock means (Hr) are able to subdivide  each clock period as a first (I1), a  second (I2) and third (I3) intervals,  all of the same duration (T / 3) equal to a third of the  clock period (T),  - the means (50, CC1, Dl, CC2, D2) for detecting the  presence or absence of solitons are suitable for  detect this presence or absence in  each of the first (I1) and second (I2)  clock period intervals.  12. Receiver according to claim 11, characterized in that the means for detecting the presence or absence of solitons in each of the intervals comprise  - an optical divider (50) at one input (52)  receiving the train of solitons (S), and at least  two outputs delivering a first and a second  soliton trains (sl, s2),  - a first control circuit (CC1) for  generate a first control signal (SC1)  corresponding to the first interval (I1) of  clock periods,  - a second control circuit (CC2) for  generate a second control signal (SC2)  corresponding to the second interval (I2) of  clock periods,  - a first detection means (D1) receiving the  first train of solitons (sl) issued by  divider (50), this first detection means (D1)  being controlled by the first control signal (ground),  - a second detection means (D2) receiving the  second train of solitons (s2) issued by the  divider (50), this second detection means  (D2) being controlled by the second signal from  command (SC2).  13. Receiver according to claim 11, characterized in that the optical divider (50) comprises a third optical output delivering a third train of solitons (s3), and in that the detection means comprise a third detection means ( D3) receiving the third train of solitons (s3) from the divider (50) and controlled by a third control signal (SC3) corresponding to the third interval (I3) of the periods, synchronization means (66) being further provided for wiring the clock signal (Hr) so that the signal (zk) delivered by the third detection means (D3) is always at logic level 0.