USOO5343322A United States Patent (19) 11 Patent Number: 5,343,322 Pirio et al. 45 Date of Patent: Aug. 30, 1994

54 SYSTEM OF VERY-LONG-DISTANCE mum Dispersion in Optical Fibres, Optics Communica DIGITAL TRANSMISSION BY OPTICAL tions, vol. 48, No. 3, Dec. 1983, pp. 181-184. FIBER WITH COMPENSATION FOR Gnauck et al.: Dispersion Penalty Reduction using an DISTORTIONS AT RECEPTION Optical Modulator with Adjustable Chirp, IEEE Photon ics Technology Letters, vol. 3, No. 10, Oct. 1991, pp. 75 Inventors: Francis Pirio; Jean Thomine, both of 96-918. Paris, France K. Iwashita et al. Chromatic Dispersison Compensation in 73 Assignee: France Telecom, Paris, France Coherent Optical Communications Journal of Lightwave Technology, vol. 8, No. 3, Mar. 1990, New York US pp. 21 Appl. No.: 995,816 367-375. 22 Filed: Dec. 23, 1992 Blow and Doran Non linear effects in optical fibres and fibre devices IEEE Proceedings, Jun. 1987, pp. 138-144. 30) Foreign Application Priority Data Agrawal; Non linear fiber optics, Academic Press, Chap Dec. 31, 1991 FR France ...... 91 6488 ter 2, Chapter 6. Gnauck and al; Dispersion penalty reduction using a opti 51 Int. CI...... H04B 10/12; H04B 10/00 cal modulator with adjustable chirp OFC 1991. Post 52 U.S. C...... 359/173; 359/161; deadline in 17. 385/28 Koyama et al.; Compensation of nonlinar pulse distortion 58 Field of Search ...... 359/130, 161, 173, 181-182, in optical fiber by employing prechirp technique ECOC91. 359/188, 195, 111; 385/3, 27-28 WCC7-2, pp. 469-472. 56) References Cited Primary Examiner-Richard E. Chilcot, Jr. U.S. PATENT DOCUMENTS Assistant Examiner-K. Negash 4,067,642 1/1978 King et al...... 385/27 Attorney, Agent, or Firm-Merchant & Gould Smith, 4,261,639 4/1981 Kogelnik et al. . ... 385/27 Edell, Welter & Schmidt 4,969,710 1/1990 Ticket al...... 385/123 X 57 ABSTRACT 4,979,234 12/1990 Agrawal et al...... 359/173 5,119,447 6/1992 Trisno ...... A 385/3 A system for very-long-distance transmission of a digi 5,166,821 11/1992 Huber ...... 359/173 tal signal between a transmitter station and a receiver 5,224,183 6/1993 Dugan ...... 385/24 station, wherein the transmitter and receiver stations 5,261,016 l/1993 Poole ...... 385/28 are connected by a monomode optical fiber with nega FOREIGN PATENT DOCUMENTS tive chromatic dispersion at the operating wavelength of the system, having a length of at least one thousand 2269309 2/1990 Japan . kilometers. The receiver station comprises device to 2240683 8/1991 United Kingdom ...... 359/16 compensate for the distortions due to the non-linear OTHER PUBLICATIONS effects and to the chromatic dispersion introduced by the transmission line, the compensation device carrying Koch and Alferness: Dispersion Compensation by Active out a positive chromatic dispersion of the received sig Predistorted Signal Synthesis, Journal of Lightwave nal, the amplitude of the positive chromatic dispersion Technology, vol. LT-3, No. 4, Aug. 1985, pp. 800-805. being a function notably of the amplitude of the nega Saito et al.: Prechirp Technique for Dispersion Compensa tive chromatic dispersion induced by the optical fiber as tion for a High-Speed Long-Span Transmission, IEEE well as of the mean on-line optical power of the signal Photonics Technology Letters, vol. 3, No. 1, Jan. 1991, transmitted on the optical fiber. pp. 74-76. Blow et al.: Nonlinear Limits on Bandwidth at the Mini 12 Claims, 2 Drawing Sheets

2 22 23 25 RECEIVER / STATION

TRANSMITTER RECEIVER STATION

24 U.S. Patent Aug. 30, 1994 Sheet 1 of 2 5,343,322

SZ

ZZ U.S. Patent Aug. 30, 1994 Sheet 2 of 2 5,343,322

N011.d3D3? 30010010Hd

No.

5,343,322 1. 2 of 1 W) or for very large propagation distances at rea SYSTEM OF VERY-LONG-DISTANCE DIGITAL sonable levels of power (some thousands of kilometers TRANSMISSION BY OPTICAL FIBER WITH in a periodic amplification system). COMPENSATION FOR DISTORTIONS AT When there is no chromatic dispersion, the Kerr RECEPTION effect induces a self-phase modulation of the optical pulse: the instantaneous frequency diminishes at the BACKGROUND OF THE INVENTION start of the pulse and then increases at its end, propor 1. Field of the Invention tionally to the derivative of the optical power. This The field of the invention is that of very-long-dis induces a widening of the spectrum and a spectral con tance digital transmission (several thousands of kilome 10 position that fosters a substantial widening for negative ters) by optical fiber in systems using on-line optical chromatic dispersions. amplification The distortion provided by the transmission fiber One of the main factors limiting the bit rate in very should be considered as the combination of the chro long-distance systems such as these is the distortion matic dispersion (the first phenomenon) and of the non induced by the transmission fiber. 15 linear effects (the second phenomenon). For, while these distortions may be more or less over The combination of these two effects may be de looked in standard fiber-optic systems (setting up links scribed by a non-linear equation with partial derivatives over distances in the range of some humdreds of kilome of distance and time, known as Schrödinger's non-linear ters), they have, on the contrary, very disturbing effects equation, the resolving of which is discussed notably in on long-distance transmission systems. 20 the work by G. Agrawal, "Non-Linear Fiber Optics', The invention relates to a system of transmission on an optic fiber line enabling compensation, at reception, Academic Press. for this distortion induced by the transmission line, in The numerical resolution of this equation shows that the case of transmission lines having a length of at least there are two forms of behavior which are qualitatively a thousand kilometers, such as those used for trans 25 very different depending on the sign of the chromatic ocean links. dispersion (D): The distortion contributed by the transmission fiber * First case: Did O. In this case, phenomena of insta arises out of the combined existence of two phenomena bility of modulation are observed. The pulses that occur in the monomode fibers: chromatic disper "burst' into very short pulses at the end of 1000 to sion and non-linear effects. 30 2000 km and the optical spectrum widens consider The first phenomenon is that of chromatic dispersion. ably: this may give rise to problems related to the This phenomenon results from the frequency depen optical passband. dency of the refractive index of silica. It entails different * Second case: D-30. There is no instability of modu propagation times depending on the operating wave lation and the pulses keep a certain degree of integ length. In general, chromatic dispersion tends to widen 35 rity while the spectrum widens quasi-monotoni the pulses of the digital trains and, hence, to create cally during the propagation, while keeping rea inter-symbol interferences. sonable widths. However, the pulses widen greatly In the commonly-used fibers, the chromatic disper temporally, thus creating inter-symbol interfer sion is zero around 1.3 um and takes a positive value of ences. These interferences become very trouble about 17 ps/nm/km around 1.55 um. It is also possible 40 some for example, as soon as the chromatic disper to use dispersion-shifted fibers which are designed to sion goes beyond 0.05 ps/nm/km in terms of abso have zero chromatic dispersion in the region of 1.55 lute value for bit rates of 5 Gbits/s on distances of in. 6000 to 8000 km. Very-long-distance transmission systems (covering The most efficient case is naturally the second one, several thousands of kilometers) work at 1.55 um. The 45 that of a negative chromatic dispersion. However, to excessive value of the chromatic dispersion of the com make the very-long-distance systems work in negative monly used fibers at this wavelength rules out their use. dispersion mode, the values of chromatic dispersion of Hence, dispersion-shifted fibers will be used in these the fibers used must obligatorily be very low. C2SeS. 2. Description of the Prior Art It must be noted that the effect of distortion by chro 50 Any method of compensation for the two phenomena matic dispersion depends greatly on the spectral com that are the cause of the distortion in the fiber (chro ponents of the pulses: if a pulse shows variations in matic dispersion and non-linear effects) is therefore of optical phase that are positive at its start and negative at great value since it can be used to overcome the draw its end, it will be greatly widened by a positive chro back of the low values imposed on the chromatic dis matic dispersion. The converse is true for negative dis 55 persion. Indeed, by using a method of compensation for persions. the distortion provided by the transmission fiber, it is The second phenomenon relates to the non-linear possible to envisage two strategies. effects. The most important non-linear effect in a fiber is In a first strategy, for given characteristics of nega the Kerr effect. This effect, which is described for ex tive chromatic dispersion of the transmission fiber, the ample in the document by K. W. Blow and N. J. Doran, 60 compensation can be used to increase the product: line “Non-linear Effects in Optical Fibers and Fiber De bit rate * range of the link. vices' (IEEE Proceedings, June 1987, pp. 138-144) In a second strategy, for a fixed line bit rate and a reflects a linear dependency of the refractive index of fixed range, the compensation makes it possible to use silica with respect to the optical power. transmission line fibers having less stringent constraints The non-linear effect is very low in the usual fields of 65 as regards the characteristics of chromatic dispersion. operation of the optical systems (distance smaller than These fibers are easier to manufacture on an industrial 400 km and power below about 10 mW), but becomes scale and to sort out for the setting up of an underwater non-negligible for very high power values (of the order link for example. 5,343,322 3 4 There are known methods of compensation for the the laser, this "chirp' of the laser being a parasitic effect distortion introduced by a fiber-optic transmission line. that consists of a variation of the instantaneous optical A first known method, described for example by A. frequency, during a pulse, especially at the beginning H. Gnauck et al in the article "Dispersion Penalty Re and end of this pulse). duction. Using an Optical Modulator with Adjustable These other methods use different structures (a Chirp' OFC 1991, post-deadline paper No. 17, consists Fabry-Pérot filter for example) which have the draw of the use, at transmission, of an optical amplitude mod back of correcting the distortions only very partially. ulator having adjustable chirp (or variation in instanta In particular, the non-linear effects are not compen neous optical frequency). sated for by these known equalizers or filters. Now, the This first method has been proposed solely in order to 10 change from a transmission line with a length of some provide partial compensation for the penalties of chro hundreds of kilometers to a line with a length of over a matic dispersion on standard optical fibers in short-dis thousand kilometers leads to a situation where it is not tance transmission systems (of some hundreds of kilo longer possible to overlook the non-linear effects. meters). Indeed, these non-linear effects become as important, Indeed, it has not been sought, in this method, to 15 if not more important, than chromatic dispersion. Con compensate for the non-linear effects, which are com sequently, a specific solution has to be found in order to pletely negligible in this case, but solely to compensate resolve this new problem. for chromatic dispersion. The invention is aimed notably at overcoming these Besides, this first known method specifically relates different drawbacks of the prior art. to standard optical fibers (with zero chromatic disper More specifically, it is an aim of the invention to sion around 1.3 m). Now, in very-long-distance trans provide a system capable of compensating for the dis mission systems (of several thousands of kilometers), the tortion induced by a transmission line with a length of at excessive value of the chromatic dispersion of the com least a thousand kilometers on a monomode optical fiber monly used fibers rules out their use, and then it is gen with negative chromatic dispersion. erally dispersion-shifted fibers (namely fibers with zero 25 Another aim of the invention is to provide a system chromatic dispersion around 1.55 um) that are used. such as this providing compensation for the combina Finally, the compensation takes place at the transmit tion of the two phenomena induced by the transmission ter station and hence by anticipation, and not at recep fiber, namely the negative chromatic dispersion and the tion, on the actually disturbed signal. non-linear effects, and not solely for either one of these It can therefore be clearly seen that this first known 30 phenomena. method is not suited to very-long-distance transmission The invention is also aimed at providing a system systems and does not relate to a system of compensation such as this, capable of being implemented irrespec for distortions at reception. tively of the type of laser used at transmission. A second known method of compensation, described Another aim of the invention is to provide a system for example by T. Koyama et al. in the article, "Com 35 such as this that is simple to implement and reliable and pensation for Non-Linear Pulse Distortion in Optical that costs little. Fiber by Employing Prechirp Technique', ECOC 91..., WeC7-2 p. 469, consists of compensation, at transmis SUMMARY OF THE INVENTION sion, for the self-phase modulation generated by the These aims and other that shall appear here below are non-linear effects. achieved according to the invention by means of a sys To this end, the pulses are made to undergo a contin tem for the very-long-distance transmission of a digital uous scanning of the optical frequency, for the duration signal between a transmitter station and a receiver sta of a bit, by over-modulation of the sending laser. tion, said transmitter and receiver stations being con What is used in this case is the fact that the optical nected by a monomode optical fiber with negative chro frequency of a semiconductor laser is a function of its 45 matic dispersion at the operating wavelength of the current. However, only certain structures of lasers can system, having a length of at least one thousand kilome thus display high efficiency of frequency modulation, ters, said optical fiber inducing a negative chromatic i.e. a major variation of this optical frequency, for a dispersion and non-linear effects, the combination of small variation of the control current and hence of the said non-linear effects and of said negative chromatic optical power. SO dispersion giving distortions to the digital signal re This second known method has the drawback of ceived by said receiver station, said receiver station requiring a particular over-modulation, which is syn comprising means to compensate for said distortions, chronous with the useful digital train, on the sending said compensation means carrying out a positive chro laser used. Consequently, only certain types of lasers, matic dispersion of said received signal, the amplitude those capable of providing a major variation of the 55 of said positive chromatic dispersion being a function optical frequency for a low variation of the control notably of the amplitude of the negative chromatic current, may be used in this second method. dispersion induced by said optical fiber as well as of the Furthermore, this method can be applied only for a mean on-line optical power of said signal transmitted on particular binary coding, namely the on-line RZ coding. said optical fiber. Furthermore, just as in the first method described, the The invention therefore relates specifically to sys compensation takes place at transmission and not at tems of very-long-distance transmission. According to reception. the invention, there is used a monomode optical fiber Finally, this method enables compensation solely for with negative chromatic dispersion which appears to be non-linear effects. the only one suited to long distances. There are other known methods of compensation that 65 According to the invention, the amplitude of the work for the correction, at reception, of the distortion positive chromatic dispersion that makes it possible to due to the combination of high positive chromatic dis carry out the compensation is not only a function of the persions and non-linear effects (in particular the chirp of amplitude of the negative chromatic dispersion induced 5,343,322 5 6 by the optical fiber but takes account also of the mean chromatic dispersion to a received signal having a wide on-line optical power. spectrum owing to the non-linear effects. The fact of taking account of the mean on-line optical In other words, the invention provides a particularly power can be used to take account of the non-linear simple solution to the problem, specific to long-distance effects. Indeed, the non-linear effects, which are spe 5 transmission, represented by the combined compensa cific to the very-long-distance links, are a function of tion for negative chromatic dispersion and for non-lin this mean on-line optical power. ear effects, in showing that, contrary to ideas conven In other words, while the known methods consist in tionally accepted in the field of optical transmission, the compensating for the negative chromatic dispersion use of well chosen positive chromatic dispersion means induced by the fiber in carrying out a positive chro O can enable the restitution of an exploitable corrected matic dispersion with an amplitude that is precisely signal. equal to that of the negative chromatic dispersion, the Advantageously, said amplitude of the positive chro invention proposes to compensate for the distortions matic dispersion is also a function of the electrical pass due to the combination of the negative chromatic dis band of said receiver station as well as of the digital persion and the non-linear effects induced by the fiber, 15 signal sent out by said transmitter station. in achieving a positive chromatic dispersion, the ampli In this way, the compensation for the distortions tude of which is not equal (except in particular cases) to induced by the combination of the negative chromatic that of the negative chromatic dispersion and is a func dispersion and of the non-linear effects is further im tion of the optical power. proved. Furthermore, in short-distance transmission, since the 20 Advantageously, said digital signal is encoded ac mean on-line optical power does not come into play in cording to the NZ or NRZ binary format. the computation of the compensation for the distor In a first advantageous embodiment of the invention, tions, it may advantageously be increased in order to said compensation means include at least one section of increase also the signal-to-noise ratio and hence to re optical fiber with positive chromatic dispersion at said duce the bit error rate (BER). 25 wavelength of operation. By contrast, in long-distance transmission, the mean Advantageously, said compensation means include at on-line optical power plays a role in the computation of least one optical amplifier and at least two fiber sections the compensation for distortions since the non-linear with positive chromatic dispersion, each of said amplifi effects are a function of this optical power. In the case ers being interposed between two consecutive fiber of the invention, the value of the optical power results 30 sections. therefore from a compromise: it should be neither too In this way, the attenuation of the fiber is compen high, to prevent the non-linear effects from being re sated for and the receiver receives sufficient power for duced, nor too low in order that the signal-to-noise ratio its operation. may be high enough. Advantageously, said compensation means include at After propagation in a fiber that displays negative 35 least two fiber sections with positive chromatic disper dispersion and induces varying levels of non-linear ef. sion chosen from among a set of fiber sections having at fects, a pulse tends to widen. This widening notably least two different lengths, so as to adjust said amplitude prompts inter-symbol interference. of the positive chromatic dispersion. In the system according to the invention, the pulses In this way, the assessment of the amplitude of the received are compressed at reception by being made to positive chromatic dispersion to be applied, which gen undergo a positive chromatic dispersion. This compres erally calls for the use of intensive and complicated sion causes the disappearance of the inter-symbol inter computer programs of digital simulation, may be car ference and therefore improves the quality of the trans ried out simply by a trial-and-error system on an in mission. stalled link. It must be noted that the setting up of a positive chro 45 This also makes it possible to determine the optimum matic dispersion at reception, firstly, in manner that can compensation when the characteristic parameters of the be easily understood, enables compensation of the nega transmission system (notably the amplitude of the nega tive dispersion due to the transmission fiber but also, tive chromatic dispersion induced by the transmission secondly and far more surprisingly, enables compensa fiber, the electrical pass band of the receiver station, the tion for the non-linear effects. 50 signal sent out-whether or not it is phase/frequency Indeed, it is known in the field of optical transmission modulated in addition to the amplitude modulation for that non-linear effects appear during long-distance example-, the mean on-line optical power, etc.) are not transmissions. It is also known that, owing to these perfectly known before the installation of the link. non-linear effects, the optical spectrum at output of the In other words, the length of the correction fiber is transmission line is greatly widened, the factor of wid 55 first of all assessed roughly: then the addition or the ening being dependent on the optical power. removal of different lengths makes it possible to obtain Now, according to conventionally accepted ideas, the length of fiber corresponding to the optimum com any means that induces a high chromatic dispersion pensation. produces major distortions on a wide spectrum. Those Advantageously, a first part of said compensation skilled in the art are therefore clearly encouraged to 60 means is within the receiver station and a second part of refrain from using means that give high (positive) chro said compensation means is outside the receiver station. matic dispersion to try and reduce thye distortions due Thus, by placing a first part of the compensation to the combination of the (negative) chromatic disper means (i.e. a part of the length of the correction fiber) sion and of the non-linear effects induced by the trans outside the receiver station, space is saved in the re mission fiber. 65 ceiver station. The invention therefore runs counter to these con Furthermore, by placing a second part of the com ventionally accepted ideas since it specifically recom pensation means inside the receiver station, it becomes mends the use of such means giving a high (positive) easy to work on the length of the compensation fiber by 5,343,322 7 8 adding or removing small-sized fiber sections. Simi Now, following another strategy, by preserving a larly, problems of optical level are easily handled by the fixed on-line bit rate and a fixed range, the system ac positioning, in the reception station, of optical amplifi cording to the invention enables the use of line fibers e.S. that have less stringent constraints as regards the char In a second advantageous embodiment of the inven acteristics of chromatic dispersion and are therefore tion, said compensation means comprise at least one set easier to manufacture on an industrial scale. of at least two diffraction gratings mounted so as to FIGS. 1a to 1c each show two curves: induce a positive chromatic dispersion. - a first curve of variation of the optical power PO as This second embodiment is slightly more complex a function of time, shown in solid lines; than the previous one but, in return, makes it possible to 10 - a second curve of variation of the instantaneous obtain better quality results. optical frequency FI (corresponding to the deriva Advantageously, the receiver station then comprises tive in relation to the time of the phase) as a func means to adjust the spacing and/or the inclination of at tion of the time t. This second curve is shown in least one of said diffraction gratings. dashes. Finally, it is quite possible to combine the two em 15 FIG. 1a corresponds to a pulse sent. The pulse has bodiments described here above. width Lo. The instantaneous frequency FI is constant. This pulse is transmitted through a very-long-distance BRIEF DESCRIPTION OF THE DRAWINGS fiber (length greater than 1000 km) of negative chro Other features and advantages shall appear from the matic dispersion. following description of a detailed preferred embodi This same pulse received at the end of the line is ment of the invention, given by way of a non-restrictive shown in FIG. 1b. After propagation in a fiber of nega example, and from the appended drawings, of which: tive chromatic dispersion on the one hand, and showing FIGS. 1a to 1c show curves of variation in optical major non-linear effects on the other hand, the pulse has power and instantaneous optical frequency as a function widened (length L1 greater than Lo), thus prompting of time, the three figures corresponding respectively to 25 inter-symbol interference. This pulse has an instanta the same pulse during the three successive steps follow neous frequency FI (derived with respect to the time of ing a transmission with a system according to the inven the phase) that is quasi-linear, with a slope correspond tion: ing to the chromatic dispersion undergone. Its spectrum * sending of the pulse, has widened because of non-linear effects. The lowest * reception of the pulse at the end of the line, 30 frequencies are at the start of the pulse and the highest * compensation for the pulse after reception. frequencies at the end of the pulse. FIGS. 2 to 4 each show a simplified drawing of a According to the invention, this pulse is compressed distinct embodiment of a transmission system according at reception by being made to undergo a positive chro to the invention, the compensation means respectively matic dispersion. FIG. 1c shows a pulse that has under comprising: 35 gone a compression such as this. The low frequencies have been slowed down (10) and the high frequencies * a fiber section with positive chromatic dispersion, have been accelerated (11). This compression makes the * fiber sections with positive chromatic dispersion inter-symbol interference disappear and therefore im and of different lengths, between which optical proves the quality of the transmission. amplifiers are interposed, It will be noted that this compensation does not re * two diffraction gratings. store the initial pulse precisely. Because of the non-lin DETAILED DESCRIPTION OF THE ear effects, the optical frequency diminishes at the start INVENTION (12) of the pulse and then rises at its end (13). However, this is of no consequence from the viewpoint of the In a system of long-distance digital transmission (ex 45 digital signal. tending to several thousands of kilometers) by optical The pulse obtained after compensation is also short, fiber, using on-line optical amplitication, the optical even shorter than the pulse emitted. This makes it possi pulses sent out undergo a distortion. This distortion is ble to recover an eye-opening rate, at reception, that is due to the fiber itself and comes from the combination equal to or better than the rate at transmission. of the non-linear effects (the Kerr effect in particular) FIGS. 2 and 3 show two variants of a first embodi and of the chromatic dispersion in the transmission ment of the transmission system according to the inven fiber. tion. Studies on the combination of these two effects (non The simplest transmission system to implement is linear effects and chromatic dispersion) have shown shown in FIG. 2. that it is useful to make very-long-distance systems 55 The signal transmitted is formed by a train of pulses, work in negative chromatic dispersion mode. However, each of these pulses (such as the one shown in FIG. 1a) the very small values of chromatic dispersion required is generated by a transmitter station 21. This pulse gets make it difficult to obtain the transmission-line fiber. propagated in a segment 22 of fiber having a length Liot The transmission system, which shall be described in (in kilometers) and having a mean chromatic dispersion detail here below, comprises means of compensation for per kilometer Dmoy (in ps/nm/km) such that Dnoy is the non-linear effects as well as for negative chromatic negative. dispersion, these two types of phenomena being in After propagation, the pulse will have undergone a duced by the fiber of the transmission line. negative dispersion D (in ps/nm) with D=Dmoy'Liot. Thus, by preserving a transmission fiber that has The pulse then has the characteristics of a pulse such as given negative chromatic dispersion characteristics, the 65 the one shown in FIG. b. system according to the invention can be used to in According to the invention, the transmission system crease the product: (on-line bit rate) * (range of the comprises, at reception, a fiber with positive chromatic link). dispersion that makes it possible to compensate for the 5,343,322 10 distortions due to the combination of the negative chro Furthermore, it may be that the parameters of the matic dispersion, on the one hand, and of the non-linear transmission line are not defined before the line is laid. effects, on the other. For example, it may be that there is no precise a priori If, at reception, there is a correction fiber available knowledge of Dinoy because of uncertainties related to with a mean positive chromatic dispersion per kilometer manufacture, or because Dncy varies from one link to Dcor (in ps/nm/km), it is possible to make a segment 23 another. In the same way, it is sometimes impossible to of this fiber with a length Leo. simulate the mean on-line optical power with precision. The amplitude of the positive chromatic dispersion It should therefore be possible to adjust Leo to each induced by this segment 23, namely Leo'Doshould be link. To do this, it is useful to divide the correction fiber Such that the distortions are compensated for. O into sections that are very short with respect to the This amplitude of the positive chromatic dispersion is expected Leo. The addition or removal of these elemen a function of several parameters, and notably of: tary sections during the installation then makes it possi - the amplitude of the negative chromatic dispersion ble to obtain the optimum length with a precision that is induced by the transmission fiber segment 22; all the more efficient as the elementary sections are - the mean on-line optical power transmitted by the 15 small with respect to the expected Lcor. fiber segment 22: If it is sought to further improve the precision of the - the electrical pass band of the receiver 24 of the compensation, it is possible to resort to a system accord receiver station 25; ing to the “boxes of weights' principle. Use is made, - the signal sent by the transmitter station 21. firstly, of the greater lengths which give approximate In this first embodiment, it is assumed that all these 20 parameters are well defined. In this case, the amplitude compensation and, secondly, shorter lengths which of the chromatic dispersion Leo of the correction fiber enable the compensation to be adjusted to its optimum segment 22 to be used may be computed, for example, value. by means of digital simulations taking account of all the Finally, the question that arises is that of the physical parameters brought into play. 25 location of the correction fiber. This physical location There is thus a device available, providing for a chro may be in a transmission line cable on a terminal section, matic dispersion with a sign opposite that of the line and or else inside the receiver station. enabling the elimination of the distortions. The receiver The first approach, which consists in placing the receives a pulse possessing the characteristics of a pulse correction fiber in a transmission line cable, namely such as is shown in FIG. 1c (the elimination of inter 30 outside the receiver station, has the advantage of saving symbol interference and the improvement of quality of space in the receiver station but has two drawbacks: the received and corrected signal). - it will be very difficult to make adjustments in the The receiver station 25, located downline from the event of error on Loor (especially in the case of an transmission fiber 22, includes the correction fiber 23 as underwater link where the cable is in the sea), well as the receiver 24. 35 - any repairs to the cable in a terminal section is liable Of course, in order that the compensation may be to modify Leo and hence to degrade the compensa applicable, the section Loor should not be too long in tion. relation to Liot. It is therefore extremely desirable to The second approach, in which the correction fiber is have Dcor>> Dmoy. If, for example, Door/D- placed inside the receiver station, appears to be more noy = 100, the length Leo will correspond to only 1% flexible. This second approach is worthwhile notably if of the length Liot, which would appear to be reasonable the precise compensation length is not known or if this from the viewpoint of its construction. - length changes following modifications on the line. The transmission system shown in FIG. 2 seems how Indeed, it is then easier to take action on the length of ever to be relatively difficult to implement as such. compensation fiber by adding or by removing small Indeed, it does not take a certain number of problems 45 sized fiber sections. Similarly, the problems of optical into account. level can be easily managed by placing optical amplifi Thus the correction fiber, especially if it has a sub ers in the receiver station. Stantial length, shows a certain degree of attenuation. If Finally, a third approach combines the above two this attenuation is excessive, it will not be possible to approaches: a part of the correction length is in a cable add on the fiber as such as the end of the line, for the 50 in a terminal section and another part, which may be receiver in this case would not receive sufficient power adjustable, is in the receiver station. to ensure its operation. An exemplary structure of this first embodiment An alternative to this system is shown in FIG. 3. In gives excellent results. The system, in this example, is a this variant, the length Lco is divided into n smaller trans-ocean link with a length of 8,000 kilometers, at a sections 11, 12, ... ln, between which there are interposed 55 bit rate of 5 Gbits/s and an operating wavelength of optical amplifiers 311 to 31, to maintain the level of 1.55 um. The line is constituted by sections of shifted power. dispersion fibers with a negative dispersion Dmoy at 1.55 In this way, a pulse generated by a transmitter station un. 32 is propagated in a fiber with a length Lto and under Between these sections, at every 40 km, there are goes a negative dispersion D. Inside the receiver station interposed optical amplifiers that compensate for the 33, the pulse is then propagated by n fiber sections with losses. a length 11, 12, . . . ln, providing in all for a chromatic The fact of working at a wavelength equal to 1.55 un dispersion Leo'Dco. The optical amplifiers 311 to 31, makes it possible easily to obtain the above-mentioned interposed between the fiber sections with a length ll, 12, fibers with high positive dispersion. It suffices to make ... ln, make it possible to compensate for the attenuation 65 use of usual fibers with a chromatic dispersion of about due to the correction fiber formed by n sections. The 17 ps/nm/km at 1.55 um. pulses transmitted to the receiver 34 therefore does not In taking, for example, the case Dmoy= -0.1 show any inter-symbol interference. ps/nm/km, the pulses then undergo a total negative 5,343,322 11 12 dispersion during the propagation of: Dno.8000 up to a receiver station 43. The transmission fiber 42 has km = -800 ps. a total length Liot and a total negative chromatic disper In this specific case, the inventors have found that the sion (-D). correction fiber length to be used for the compensation The receiver station 43 comprises a receiver photodi is equal to 50 km. It must be noted that if only the nega ode 44 as well as an assembly based on diffraction grat tive chromatic dispersion induced by the transmission ings 45 and 46. This assembly is aimed at inducing a fiber is taken into account, then the length would be positive chromatic dispersion that makes it possible to equal to: compensate for the negative chromatic dispersion -(-0.1 x8000)/17=47 km. (-D) of the fiber 42 as well as the non-linear effects 10 induced by this same fiber 42. It is clear, in this example, that the compensation for The general principle of this assembly is described in the combination of the non-linear effects and of the G. P. Agrawal, "Non-Linear Fiber Optics', Academic negative chromatic dispersion is different from a simple Press, 1989, chapter 6. However, in this document, the compensation for the negative chromatic dispersion assembly is specifically designed for the compression of (compensation of the type implemented for short-range 15 ultra-short pulses. There is no question whatsoever of transmissions). compensating for the distortions in a transmission sys As the case may be, the length of the correction fiber te. may be greater or smaller than the computed length, In other words, the invention adapts known pulse taking account solely of the negative chromatic disper compression means to a totally distinct field, namely sion of the transmission fiber. 20 very-long-distance transmission, and for a different In short, the following are the characteristics of the application, namely that of compensation for distances, transmission system: - length of the link: 8000 km; and more specifically that of compensation for distor -bit rate: 5 Gbit/s; tions resulting from the combination of the negative - wavelength of use: 1.55 um; 25 dispersion and of the non-linear effects induced by the - NRZ coding; fiber. - optical amplifiers at a distance of 40 km from each Pulses propagated in the fiber 42 have their spectra other and each having a noise excess of 6 dB; widened after propagation. They go through a first lens - transmission fiber: 47. The optical beam 48 coming out of this first lens 47 * attenuation: 0.2 dB/km, 30 is then sent through two diffraction gratings 45 and 46. * chromatic dispersion: -0.1 ps/nm/km, The pulses are compressed during a dual passage - level of signal at output of amplifier: -3dBm, through these two diffraction gratings. The optical - correction fiber: beam is reflected on a first mirror 49 and goes through * length: 50 km, the two diffraction gratings 45 and 46 a second time. * chromatic dispersion: 17 ps/nm/km. 35 This dual passage makes it possible to restore a circular The following results are obtained: cross-section to the optical beam (a section that has - without compensation: there is a eye-opening pen become ellipsoidal again because of the first passage) alty of 2 dB at reception. This penalty is related to and to multiply the compression factor by two. the phenomena of propagation and leads to an The first mirror 49 is slightly inclined in order to error rate of about 10-9, 40 separate the incident beam from the reflected beam. A - with compensation: the cancellation of the eye second mirror 410 (shown in dashes and located in a opening penalty makes it possible to obtain an error raised plane with respect to the plane of the figure) rate better than 1012. If an increase is made in the receives the beam reflected by the first mirror 49 and value of the corrective positive chromatic disper deflects it, without introducing additional losses sion (D greater than 800 ps/nm), it is even possible 45 therein, towards a second lens 411. The reception diode to obtain an improvement of the eye diagram (+1 44 finally receives compressed pulses without inter dB) with respect to the diagram at transmission. symbol interference. A second exemplary embodiment has also given very Thus, by computing the spacing and inclination be good results. This second exemplary embodiment is tween the two diffraction gratings, it is possible to ob differentiated from the first one solely by the following 50 characteristics: tain an induced chromatic dispersion that enables the - length of the link: 7960 km, distortions to be compensated for with precision. - transmission fiber: If the chromatic dispersion of the link is not known * chromatic dispersion double that of the first ex exactly, it is possible to modify the spacing and the ample: -0.2 ps/nm/km, 55 angle between the two diffraction gratings by mechani - correction fiber: cal translation-rotation of these two gratings. This ena * length double that of the first example: 100 km. bles the positive chromatic dispersion of compensation The following results are obtained: to be adjusted as efficiently as possible. - without compensation: there is a eye-opening pen In this compensation assembly based on diffraction alty of 5.5 dB that leads to an error rate of about 60 gratings, the passing of the beam into the open air prob 10-5, ably results in substantial losses. It is then possible to - with compensation: the cancellation of the eye provide for a system in which this compensation assem opening penalty makes it possible to obtain an error bly is preceded (or followed) by an optical amplifier so rate better than 10-12. that the pulses arrive with the right level at the receiver FIG. 4 shows a second embodiment of the system of 65 photodiode. transmission according to the invention. The system This second embodiment is slightly more difficult to comprises a transmitter station 41 generating optical implement than the first one, but makes it possible to pulses. These optical pulses are transmitted by a fiber 42 obtain more precise compensation. 5,343,322 13 14 It is therefore possible also to provide for a system two diffraction gratings connected to said monomode that successively comprises, at reception, a optical cor optical fiber so as to induce a positive chromatic disper rection fiber according to the first embodiment (and sion. makes it possible to obtain a rather rough compensa 9. A system according to claim 8, comprising means tion) and an assembly of diffraction gratings according to adjust a spacing and/or an inclination of at least one to the second embodiment (this assembly being de of said diffraction gratings. signed to obtain a finer compensation from the rough 10. A system according to claim 1, wherein Said com compensation that has already been obtained). pensation means includes: What is claimed is: at least one section of optical fiber with positive chro 1. A system for a very-long-distance transmission of a O matic dispersion at said wavelength of operation; digital signal between a transmitter station and a re and ceiver station, wherein said transmitter and receiver at least one set of at least two diffraction gratings. stations are connected by a monomode optical fiber 11. A system for a very-long-distance transmission of with negative chromatic dispersion at an operating a digital signal between a transmitter Station and a re wavelength of the system, having a length of at least 15 ceiver station, wherein said transmitter and receiver one thousand kilometers, said optical fiber inducing a stations are connected by a monomode optical fiber negative chromatic dispersion and non-linear effects, a with negative chromatic dispersion at an operating combination of said non-linear effects and of said nega wavelength of the system, having a length of at least tive chromatic dispersion giving distortions to the digi one thousand kilometers, said optical fiber inducing a tal signal received by said receiver station, and wherein 20 negative chromatic dispersion and non-linear effects, a said receiver station comprises compensation means for combination of said non-linear effects and of said nega compensating for said distortions, said compensation tive chromatic dispersion giving distortions to the digi means being connected to said monomode optical fiber, tal signal received by said receiver station, and wherein Said compensation means carrying out a positive chro said receiver station comprises compensation means for matic dispersion of said received signal, an amplitude of 25 compensating for said distortion, said compensation said positive chromatic dispersion being a function no means being connected to said monomode optical fiber, tably of an amplitude of the negative chromatic disper said compensation means carrying out a positive chro sion induced by said optical fiber as well as of a mean matic dispersion of said received signal, an amplitude of on-line optical power of said digital signal transmitted said positive chromatic dispersion being a function no on said optical fiber. 30 tably of an amplitude of the negative chromatic disper 2. A system according to claim 1, wherein said ampli sion induced by said optical fiber as well as of a mean tude of the positive chromatic dispersion is also a func on-line optical power of said signal transmitted on said tion of an electrical pass band of said receiver station as optical fiber, and wherein said compensation means well as of the digital signal sent out by said transmitter include at least two fiber sections with positive chro station. 35 matic dispersion chosen from a set of fiber sections 3. A system according to claim 1, wherein said digital having at least two different lengths, so as to adjust said signal is encoded according to the RZ or NRZ binary amplitude of the positive chromatic dispersion. format. 12. A system for a very-long-distance transmission of 4. A system according to claim 1, wherein said con a digital signal between a transmitter station and a re pensation means includes at least one section of optical ceiver station, wherein said transmitter and receiver fiber with positive chromatic dispersion at said wave stations are connected by a monomode optical fiber length of operation, which is connected to said mono with negative chromatic dispersion at an operating mode optical fiber. wavelength of the system, having a length of at least 5. A system according to claim 4, wherein said com one thousand kilometers, said optical fiber inducing a pensation means includes at least one optical amplifier 45 negative chromatic dispersion and non-linear effects, a and at least two fiber sections with positive chromatic combination of said non-linear effects and of said nega dispersion, each of said amplifier being interconnected tive chromatic dispersion giving distortions to the digi between two consecutive fiber sections. tal signal received by said receiver station, and wherein 6. A system according to claim 4, wherein said com said receiver station comprises compensation means for pensation means includes at least two fiber sections with 50 compensating for said distortions, said compensation positive chromatic dispersion chosen from among a set means being connected to said monomode optical fiber, of fiber sections having at least two different lengths, so said compensation means carrying out a positive chro as to adjust said amplitude of the positive chromatic matic dispersion of said received signal, an amplitude of dispersion by choosing which fiber section or fiber said positive chromatic dispersion being a function no sections among said at least two fiber sections must be 55 tably of an amplitude of the negative chromatic disper connected to said monomode optical fiber. sion induced by said optical fiber as well as of a mean 7. A system according to claim 4, wherein a first part on-line optical power of said signal transmitted on said of said compensation means is within the receiver sta optical fiber, and wherein said compensation means tion and a second part of said compensation means is includes at least one set of at least two diffraction grat outside the receiver station. 60 ings mounted so as to induce said positive chromatic 8. A system according to claim 1, wherein said com dispersion. pensation means comprises at least one set of at least ck ck ck xc

65