METHOD FOR PRODUCING A QUALITY OF TRANSMISSION ESTIMATOR FOR OPTICAL TRANSMISSIONS

Abstract

A technique is provided for producing a quality of transmission estimator for optical transmissions. The technique includes defining a local dispersion value, defining a dispersion increment, and performing a propagation calculation of an optical signal along an elementary section. The elementary section is a propagation medium characterized by the local dispersion value. The elementary section length may correspond to the dispersion increment. The optical signal, which is incoming in the elementary section, is previously affected by a cumulative dispersion value equal to an integer number of the dispersion increment. For each elementary section, a variance of noise is determined, the noise representing a distortion due to Kerr nonlinear field contributions in the elementary section. For each couple of elementary sections, a covariance of noise is determined between the couple of elementary sections. The variances and covariances may be stored in a look-up table of a data repository.

Claims

1. A method for producing a quality of transmission estimator for optical transmissions, the method comprising: defining a local dispersion value, defining a dispersion increment having a sign identical to the local dispersion value, for each of a plurality of integer numbers, wherein the integer numbers range from 0 to an upper bound greater than or equal to 0, performing a propagation calculation by a propagation model and/or experiment, each propagation calculation and/or experiment dealing with the propagation of an optical signal along an elementary section, and wherein the elementary section is a propagation medium characterized by the local dispersion value, an elementary section length corresponding to the dispersion increment, and wherein the optical signal which is incoming in the elementary section is previously affected by a cumulative dispersion value equal to the sum of a predefined pre-compensation dispersion and the integer number times the dispersion increment, for each elementary section, determining a variance of noise, the noise representing a distortion due to Kerr nonlinear field contributions in the elementary section, for each couple of elementary sections, determining a covariance of noise between the couple of elementary sections, storing in a data repository a look-up table comprising each determined variance of noise in association with the corresponding local dispersion value and cumulative dispersion value and each covariance of noise, in association with a first couple of local dispersion value and cumulative dispersion value and a second couple of local dispersion value and cumulative dispersion value.

2. The method in accordance with claim 1, wherein the propagation model is Split Step Fourier Method.

3. The method in accordance with claim 1, wherein the noise represents a distortion further due to any nonlinear field contribution and/or association of nonlinear field contributions from the following list: second harmonic generation, frequency mixing, optical parametric amplification and oscillation, spontaneous parametric down conversion, sources of entangled photons based on SPDC, four-wave mixing, Raman scattering, spontaneous and stimulated Raman scattering, Raman amplification, Brillouin Scattering and two photons absorption.

4. The method in accordance with claim 1, wherein the method further comprises a non-dimensionalizing step comprising: for each elementary section, determining an input power that was employed in the propagation model or experiment, for each variance determined for an elementary section in the determining step, dividing the variance by the input power employed for the elementary section to the square, for each covariance determined for a couple of elementary sections in the determining step, dividing the covariance by the input power determined for the first elementary section of the couple and the input power determined for the second elementary section of the couple.

5. The method in accordance with claim 1, wherein the local dispersion value corresponds to an optical fiber.

6. The method in accordance with claim 5, wherein the optical fiber has a type selected in the following list: Single Mode Fiber, Dispersion Compensation Fiber, LEAF, multi-fiber, multicore fiber, multi-mode fiber, polarization-maintaining fiber, photonic-crystal fiber, multimode graded index optical fiber, Non-Zero Dispersion Shifted Fiber, True-Wave-Reduced Slope, True-Wave-Classic, Teralight and SMF-LS.

7. The method in accordance with claim 1, wherein the look-up table comprises a covariance matrix of the noise due to the Kerr nonlinear field contributions generated in the elementary sections.

8. The method in accordance with claim 1, wherein the dispersion increment corresponds to a dispersion cumulated by an optical signal propagating along a section of an optical link which length is comprised between 100 m and 20 km.

9. A quality of transmission estimator device for optical transmissions, the device comprising: a data repository in which is stored a look-up table, comprising a plurality of variance entries, each variance entry being stored in association with a corresponding local dispersion value and a corresponding cumulative dispersion value, the cumulative dispersion value being chosen in a set of cumulative dispersion values consisting of the sum of a predefined pre-compensation dispersion and a predefined dispersion increment multiplied by an integer number ranging from 0 to an upper bound greater than or equal to 0, the look-up table further comprising a plurality of covariance entries, each covariance entry being stored in association with a first couple of local dispersion value and cumulative dispersion value and a second couple of local dispersion value and cumulative dispersion value, an input interface adapted to receive an optical transmission system description, the system description defining a plurality of system segments and, for each system segment, an input power of the system segment, a local dispersion value of the system segment and an input cumulative dispersion of the system segment, a calculation unit configured to perform: for each system segment, selecting a variance entry in the look-up table, so that the local dispersion and input cumulative dispersion of the system segment substantially match the local dispersion value and cumulative dispersion value associated with the variance entry, for each couple of the system segments selecting a covariance entry in the look-up table, so that the local dispersion and input cumulative dispersion of the system segment substantially match the first couple associated with the covariance entry and so that the local dispersion and input cumulative dispersion of the system segment substantially match the second couple associated with the covariance entry calculating a quality of transmission estimate ${\mathrm{SNR}}_{\mathrm{NL}}^{-1}=\underset{k=1}{\overset{N}{.Math.}}.Math.P_k^2.Math.{\sigma }_{\mathrm{match}\left(k\right).Math.\mathrm{match}\left(k\right)}+2.Math.\underset{k=1}{\overset{N}{.Math.}}.Math.\underset{k^{\prime }=1}{\overset{k-1}{.Math.}}.Math.P_k.Math.P_{k^{\prime }}.Math.\mathrm{Re}\left[{\sigma }_{\mathrm{match}\left(k\right).Math.\mathrm{match}\left(k^{\prime }\right)}\right]$ where N is the number of system segments in the optical transmission system description, an output interface for transmitting the calculated quality of transmission estimate.

10. The device in accordance with claim 9, wherein the look-up table comprises a covariance matrix.

Description

BRIEF DESCRIPTION OF THE DRAWINGS

[0080] These and other aspects of the invention will be apparent from and elucidated with reference to the embodiments described hereinafter, by way of example, with reference to the drawings.

[0081] FIG. 1 is a schematic graph of cumulative dispersion values of an optical signal propagating over an optical link as a function of distance from the source node, the optical link comprising an optical fiber having no in-line dispersion compensation.

[0082] FIG. 2 is a look-up table comprising variances of a nonlinear noise of the optical signal propagating over the optical link of FIG. 1, the variances being in association with corresponding cumulative dispersion values estimated in FIG. 1, the look-up table further comprising covariances of the noise, in association with a cumulative dispersion value of different sections of the optical link.

[0083] FIG. 3 is a covariance matrix of the nonlinear noise of the sections of the optical link of FIG. 1, the covariance matrix being identical to the look-up table of FIG. 2, the covariance matrix being constructed thanks to the look-up table of FIG. 2 by selecting the variances and covariances of the cumulative dispersion values corresponding to the successive sections of the optical link of FIG. 1.

[0084] FIG. 4 is a schematic graph of the cumulative dispersion of an optical signal propagating over an optical link as a function of distance from the source node, for an optical link with a dispersion compensation device being half-way between the source and the destination nodes situated at both ends of the optical link.

[0085] FIG. 5 is the look-up table of FIG. 2, showing the selection of the variances and covariances of the cumulative dispersion values corresponding to the sections of the optical link of FIG. 4.

[0086] FIG. 6 is a covariance matrix of the nonlinear noise of the sections of the optical link of FIG. 4, the covariance matrix being constructed thanks to the look-up table of FIG. 2 by selecting the variances and covariances of the cumulative dispersion values corresponding to the successive sections of the optical link of FIG. 4.

[0087] FIG. 7 is a flow chart showing a method that carries out the calculation of the fast estimation of the quality of transmission of an optical signal along an optical link.

[0088] FIG. 8 is a functional drawing of a computing device that may be employed for computing the estimation of the quality of transmission of an optical signal along an optical link.

[0089] FIG. 9 is a functional drawing of an example of optical link from a transmitter to a receiver.

[0090] FIG. 10 is a graph showing the numerical simulation results of the variance over the section length calculated thanks to the method of FIG. 7, compared to the variance calculated thanks to a SSFM simulation.

[0091] FIG. 11 is a schematic of a 6-nodes network having optical fibers of two types.

[0092] FIG. 12 shows the discretization of the dispersion map of two optical transmission paths of the 6-nodes network of FIG. 11.

DETAILED DESCRIPTION OF THE EMBODIMENTS

[0093] In an illustrative example, a WDM optical network carries 100 Gb/s per wavelength on uncompensated optical links, i.e. without inline dispersion compensation between the optical fiber spans comprised in the optical links. The network must carry a plurality of 100 Gb/s demands, the 100 Gb/s payload being transported over Polarization Division Multiplexed Quaternary Phase Shift Keying PDM-QPSK wavelengths.

[0094] It is well known that for such systems, the mains sources of impairments are the Amplified Spontaneous Emission (ASE) noise and the distortions due to the Kerr Effect. Both effects can be modeled extremely accurately by additive white Gaussian noise. The total signal-to-noise ratio at the receiver is defined by:

[00003]${\mathrm{SNR}}^{-1}=\frac{P_{\mathrm{ASE}}+P_{\mathrm{NL}}}{P}$

[0095] and the nonlinear-distorsion-to-signal-ratio is defined by

[00004]${\mathrm{SNR}}_{\mathrm{NL}}^{-1}=\frac{P_{\mathrm{NL}}}{P},$

[0096] where P is the power of the signal, P.sub.ASE the power of the amplified spontaneous emission (ASE) noise measured in a reference bandwidth (e.g. 0.1 nm as for the traditional definition of optical SNR (OSNR)) and P.sub.NL the power of nonlinear distortions.

[0097] P.sub.ASE depends on the distance traveled by the signal and specifically on the number of traversed optical amplifiers and their characteristics.

[0098] P.sub.NL depends also on the distance traveled by the signal, therefore of the length of the optical link, on the nature of the optical link, therefore of the local dispersion value, and on the optical power of the signal under consideration.

[0099] To estimate the quality of transmission of an optical link, the fiber types comprised in the optical link are to be taken into account. In an optical network, different fiber types may be used having a plurality of different local dispersion values. For example, a Single Mode Fiber (SMF) has a local dispersion value of 17 ps/nm/km. By contrast, a Dispersion Compensation Fiber (DCF) has a negative local dispersion value.

[0100] Method for Producing a Quality of Transmission Estimator for Optical Transmissions:

[0101] With reference to FIG. 7, a numerical method for producing a quality of transmission estimator for optical transmissions will now be explained.

[0102] The method yields the nonlinear distortion variance of all possible connections in a meshed network, for an arbitrary modulation format, dispersion management and possibly heterogeneity of fiber types and fiber lengths. The resulting nonlinear distortion variance is stored in a memory and is a main component of the quality of transmission estimator.

[0103] The method comprises the performing of the following steps:

[0104] As a preamble, determining a chromatic dispersion map of the transmission system, i.e. predicting the cumulated chromatic dispersion at each and every point in the optical transmissions. The dispersion map is employed for sampling the optical transmissions into sections denoted S.sub.k. A cumulative dispersion value, denoted D.sub.k for a section S.sub.k of the network is a pre-dispersion previously cumulated by the optical signal while propagating along the fiber and measured at an input point of the section S.sub.k, where k is an integer number.

[0105] In a step 21, the method performs: determining the different fiber types present in an optical network F.sup.(l), with l=1, 2, . . . , m for m different fiber types and choosing a fixed increment of cumulative dispersion ΔD.sup.(l), for each one of the different fiber types F.sup.(l), with l=1, 2, . . . , m.

[0106] In step 22, considering all possible point-to-point links in a network and separating them into sections S.sub.k, with the fiber type of the section denoted F.sub.k and the corresponding dispersion increment ΔD.sub.k which is cumulated during the propagation of an optical signal along a section k, where F.sub.k and ΔD.sub.k are chosen among the possible fixed values F.sup.(l) and the corresponding ΔD.sup.(l) of the step 21.

[0107] In step 23, the method performs the gathering all couples (D.sub.k, F.sub.k) corresponding to each section S.sub.k of the network and the removing of the duplicates from the gathering. This step 23 leads to a total of M distinct couples (D.sub.k, F.sub.k).

[0108] In step 24, that is a calibration step, the method performs, for each couple (D.sub.k, F.sub.k), running a SSFM nonlinear propagation simulation over a section S.sub.k. The series of M simulations is referred to as the calibration phase step 24. As the M simulations are completely independent, they can all be run in parallel, thus speeding up the calibration step 24 of the network.

[0109] In an illustrative case of an optical signal propagating in the sections with a QPSK modulation, the following step 25 is performed for each of the output signals resulting from the M simulations of the calibration phase step 24. Step 25 comprises: the removing of the initial phase modulation, the compensation of the nonlinear phase shift and the removing of the input signal coming from a filtering at the transmitter side, if any. Indeed, as shown on FIG. 9, a filter 94 is optionally placed between the transmitter 95 and the fiber 91. The removing of the input signal from the output signals resulting from the M simulations allows extracting M nonlinear distortion fields of the optical signal propagating in the sections. The removing is performed on both the input and output signals sampled at the middle of each symbol time.

[0110] In step 26, the method further comprises:

[0111] Constructing a M×M table, denoted E in the following. Table Σ comprises the calculation of the variance σ.sub.ii and covariance σ.sub.ij of the nonlinear distortion fields of the optical signal propagating in the sections i and j, i and j denoting the values taken by k, for each distinct couple D.sub.k, F.sub.k.

[0112] In step 27, the method performs the storing of E in a look-up table in an external memory for future usage.

[0113] The dispersion increment ΔD.sup.(l) for each fiber type is calculated as follows: ΔD.sup.(l)=∫DF.sup.(l)(z) dz, wherein DF.sup.(l) denotes the local dispersion for the fiber type F.sup.(l) and wherein the integration is performed over the section of length dz.

[0114] The choice of the dispersion increment has an incidence on the section length of the section considered. Indeed, for a fiber section k having a fixed local dispersion value DF.sub.k, the higher the length, the higher the cumulative dispersion value D.sub.k is. Therefore, the choice of the dispersion increment for each fiber type ΔD.sup.(l) depends on the available system memory, the desired accuracy and the variety of span lengths existing in the network to be estimated. Short sections yield a higher accuracy but they demand both a large amount of memory and a higher calculation time.

[0115] For the sake of illustration, in the above method, the SSFM nonlinear propagation simulation is performed on a bit sequence with quasi-random distribution of ones and zeros. With reference to FIG. 9, an example of optical link is described. The optical link is a fiber 91. A SSFM simulation is performed for the fiber 91. As shown on FIG. 9, the dispersion is brought back to zero by a full post compensation fiber 92 before the optical signal arrives to the receiver 93.

[0116] For the sake of illustration, with reference to FIGS. 1 to 3, such a method is performed on an optical network that consists of a single optical link ab to produce a look-up table. The look-up table is useful to estimate the quality of transmission of an optical link cd as it will later be explained with reference to FIGS. 4 to 6. FIG. 1 is a schematic graph of cumulative dispersion values D.sub.1, D.sub.2, D.sub.3, D.sub.4, D.sub.5 and D.sub.6 of an optical signal propagating over the optical link ab, as a function of the sections 1 to 6 of the optical link ab. For example, D.sub.1 is a negative dispersion value equal to the value of the pre-compensation at the input of the optical link ab. The optical link ab comprises two spans 123 and 456, characterized by the same fiber type, i.e. the local dispersion value along the two spans is the same. The local dispersion value F.sub.k for each section k, with k=1 to 6, is the same. The spans 123 and 456 are respectively split into 3 sections 1, 2, 3 and 4, 5, 6. The sectioning of the spans 123 and 456 is performed as follows: for each section 1, 2, 3, 4, 5 and 6 of the optical link ab, the cumulative dispersion value represented on the axis 11 increases by a same increment of dispersion 14. The sections of the optical link ab incorporate no in-line dispersion compensation. Therefore the cumulative dispersion of each section S.sub.k denoted D.sub.k is equal to Σ.sub.i D.sub.i. For each section S.sub.k having a distinct couple (D.sub.k, F.sub.k), the method performs the running of a SSFM simulation, as a function of the couple comprising the cumulative dispersion and the local dispersion D.sub.k, F.sub.k characterizing the section S.sub.k of the optical link ab. P.sub.0 denotes a fixed power value at the input of each section S.sub.k employed in the SSFM simulations.

[0117] Denoting u.sub.NL,k the nonlinear distortion field of the optical signal at the output of the section S.sub.k calculated by the SSFM simulation, each coefficient σ.sub.ij of the matrix Σ.sub.1 represented on FIG. 2 is then calculated as follows:

σ.sub.ij=cov(u.sub.NL,i,u.sub.NL,j)/P.sub.0.sup.2

[0118] with cov(X,Y)=E[(X−μ.sub.X).sup.2 (Y*−μ.sub.Y*).sup.2] being the covariance of the random variables X and Y with averages μ.sub.X and μ.sub.Y and E[.] the expected value. The matrix Σ.sub.1 is then stored in a data repository for future usage.

[0119] Indeed, with the matrix Σ.sub.1 the transmission performance of the optical link ab may be calculated by a quality of transmission estimator device comprising the data repository, by selecting the variances and covariances σ.sub.ij of Σ.sub.1 corresponding to the couple D.sub.k, F.sub.k characterizing the sections S.sub.k of the optical link ab, with k=1, . . . 6.

[0120] The optical system described with reference to FIG. 1 does not necessary really exist. FIG. 1 is described uniquely in order to illustrate the method of calculation of the coefficients of a quality of transmission which allows estimating a large number of optical systems. Similarly, the coefficients may be calculated for a large number of local dispersion values in order to produce an estimator capable of calculating the coefficients for a large number of optical fiber types. In special cases, the coefficients of one fiber type l′ may be deduced by the already calculated coefficients of the fiber type l, by replacing the fiber type F.sup.(l′) by F.sup.(l) the cumulative dispersion D.sup.(l′) by the cumulative dispersion

[00005]$\frac{{\mathrm{DF}}^{\left(l^{\prime }\right)}}{{\mathrm{DF}}^{\left(l\right)}}.Math.D^{\left(l\right)}$

and finally the coefficient α.sub.ij by the coefficient

[00006]$\frac{{\mathrm{DF}}^{\left(l^{\prime }\right)}}{{\mathrm{DF}}^{\left(l\right)}}.Math.{\sigma }_{\mathrm{ij}}.$

[0121] In an embodiment, the dispersion increment ΔD.sup.(l) for each fiber type is defined to be the same and is denoted ΔD. Therefore, the length of the sections of different fiber types varies and is denoted dz.sup.(l).

[0122] Since the covariance matrices constructed thanks to the method above described have a Hermitian symmetry, only M*(M+1)/2 terms are to be calculated. This determines the memory usage of the method. Considering for example, 400 sections and assuming that the variance and covariance terms are stored with a double float accuracy, i.e. 8 bytes, the memory takes up 8*400*(400+1)/2=641.6 Kbytes.

[0123] The complexity of the method is upper-bound by N x (N+1)/2 additions, where N x N is the size of the sub-matrix D corresponding to the point-to-point optical link of interest with N<M.

[0124] The method described above allows a fast computation of the transmission performance of all links in a network simultaneously. Indeed, as described in step 24, the number of simulations depends on the number of the distinct couples of fiber type F.sub.k and cumulative dispersion D.sub.k of the sections S.sub.k appearing in the network.

[0125] An advantage of the method is to be able to construct a look-up table comprising the variances and covariances of the nonlinear noises generated by all possible combinations of fiber types and input cumulative dispersions for a given modulation format, that may be used to calculate the performance of any possible optical link. The overall calculation time and cost gain compared to existing methods can be found by dividing the total number of sections in a network by the number of the discrete couples found in step 24.

[0126] In an embodiment, a SSFM simulation of the method is performed for a WDM optical signal, with the assumption that each optical channel of the bandwidth is occupied. The assumption leads to over-estimate the transmission degradation.

[0127] The method described above is a numerical method. However, it may also be implemented as an experimental method according to the same steps.

[0128] Above we have described the method for producing the quality of transmission estimator. Now, the exploitation of the quality of transmission estimator will be described.

[0129] In order to exploit the quality of transmission estimator to estimate the quality of transmission of any optical system, for example the optical system of FIG. 4, a first step is necessary. The first step consists in providing a suitable description of the optical system to be estimated.

[0130] A Method for Determining an Optical Transmission System Description:

[0131] Providing a suitable description of the optical system to be estimated is achieved by the following method, which will be illustrated with reference to FIG. 11. The optical transmission system is a point-to-point optical transmission in an optical network. For example, the optical transmission system is defined as a connection between two nodes of the network. Any possible connections between any nodes of an optical network may be described thanks to the following method.

[0132] The method for determining the optical transmission system description comprises a few steps.

[0133] In a first step, the method performs determining a dispersion map of the optical transmission system. A dispersion map plots the cumulative dispersion as a function of transmission distance along an optical communication path. The dispersion compensation devices at the input and/or output ends of optical fibers, if any, may produce abrupt changes in cumulative dispersion along the optical transmission system. To construct the dispersion map, the cumulative dispersion value D may be calculated as follows: D(z)=∫DF(z)dz, where DF denotes the local dispersion value and z denotes the distance through which the optical signal has propagated. A suitable computer program may be employed to establish the dispersion map of any optical transmission system or at least substantially match those values. This step generally assumes that local dispersion is known in all links of the system.

[0134] In a second step, the method performs placing a set of discrete cumulative dispersions onto the dispersion map. The set of discrete cumulative dispersions should be comprised in the cumulative dispersion values employed in the method for producing the quality of transmission estimator. Preferably, two consecutive discrete cumulative dispersions of the set are separated by a dispersion increment ΔD defined as the smallest dispersion increment of the set of dispersion increments ΔD.sup.(l) defined for each fiber type in the method for producing the quality of transmission estimator, or an integer multiple of that increment ΔD.

[0135] As an illustration, referring to the optical system of FIG. 4, the set of cumulative dispersion values is represented by D.sub.1, D.sub.2, D.sub.3, D.sub.4, D.sub.5 and D.sub.6.

[0136] In a third step, the method performs defining a plurality of sequential system segments S.sub.k of the optical transmission system, wherein each system segment has an input point that corresponds to a point in the optical transmission system where the input cumulative dispersion matches an input cumulative dispersion of the set.

[0137] In a fourth step, the method performs, for each system segment S.sub.k, determining an input power P.sub.k of the system segment, and if necessary a local dispersion value of the system segment,

[0138] In a last step the method comprises, for each system segment S.sub.k, storing the input power and the local dispersion value determined in relation with the input cumulative dispersion of the system segment S.sub.k in a data repository.

[0139] A sequence number of the system segments S.sub.k is also stored. The sequence number defines the ordering of the system segments in a point-to-point optical system. It may be represented by an integer number. In order to describe a meshed network, the system may be represented as a combination of point-to-point links.

[0140] For the sake of illustration, a simplified network will be described with reference to FIG. 11. FIG. 11 is a schematic of an optical network 100 comprising six nodes denoted by A to D. The optical fibers linking the nodes A and B, B and E, B and D, D and F, A and C, C and E and E and F have the same local dispersion value, which is equal to half the local dispersion value of the optical fiber linking the nodes E and D. No in-line dispersion compensation is performed along the optical fibers of the network 100 and no dispersion compensation device is provided in the optical nodes. The optical connections established are transparent.

[0141] The method for determining an optical transmission system description thus performs the third step over each link of the optical network 100.

[0142] Each optical fiber of the network is segmented into system segment S.sub.k in order that each system segment S.sub.k has an input cumulative value of the set of discrete cumulative dispersions.

[0143] The local dispersion value of the optical fiber linking the nodes E and D is twice the local dispersion value of the other optical fibers of the network 100. As a result, the elementary length of the system segments between nodes E and D is equal to a value L/2 which is half the elementary length value L of the system segments of the other optical fibers of the network 100. In other words, in this particular example, the discretization of the optical system employs a uniform mesh-size in terms of cumulated dispersion, resulting in a non-uniform mesh-size in terms of distance.

[0144] An optical connection between nodes A and D will follow the shortest path A-B-D with a total length of 10 times the increment ΔD. An optical connection between nodes C and F, will follow the shortest path C-E-F with a total length of 6 times the length L.

[0145] For example, the optical paths A-B-D and C-E-F have no dispersion management and are of one single fiber type, i.e. one local dispersion value. Therefore, the sectioning of the optical paths A-B-D and C-E-F results in similar system segments S.sub.1 to S.sub.6 of same local dispersion and cumulated dispersion. Therefore the coefficients stored in the matrix corresponding to the cumulative dispersion values and local dispersion value of the system segments are usable twice for the estimation of the quality of transmission along the system segments S.sub.1 to S.sub.6.

[0146] With reference to FIG. 12, the dispersion map of the optical paths A-B-D and E-D is represented. The set of discrete cumulative dispersions placed onto the dispersion map has been shown on the vertical axis, represented by D.sub.1, D.sub.2, D.sub.3, D.sub.4, D.sub.5, D.sub.6, D.sub.7, D.sub.8, D.sub.9 and D.sub.10. The distance has been represented on the horizontal axis.

[0147] The squares represent the inputs of the system segments for the optical path E-D. The dots represent the inputs of the system segments S.sub.1 to S.sub.10 for the optical path A-B-D. The input cumulative dispersion at the input of the last system segment of the optical path A-B-D has a value D.sub.10. The local dispersion value of the optical path A-D being twice the local dispersion value of the optical path A-B-D, the slope of the line 101 is twice the slope of the line 102.

[0148] Therefore the coefficients stored in the matrix corresponding to the cumulative dispersion values and local dispersion value of the system segments for the optical path E-D are not the same as the coefficients for the optical path A-B-D.

[0149] For realistic homogeneous networks, the length of each fiber is not necessary a multiple of the dispersion increment. Therefore, for realistic networks, the length of each optical fiber is rounded to the closest multiple of the dispersion increment ΔD. The method inaccuracy due to this fiber length approximation is negligible especially for low values of the dispersion increment.

[0150] Exploitation of the Quality of Transmission Estimator:

[0151] Thanks to the method for determining an optical transmission system description above described, any optical transmission may be described in a convenient format for the exploitation of the quality of transmission estimator obtained thanks to the method for producing a quality of transmission estimator.

[0152] With reference to FIG. 8, such a quality of transmission estimator device for optical transmissions will now be described. The device comprises a data repository 113 and a calculation unit 114 that is linked to the data repository 113 as shown by line 116. The device further includes an output interface 115 that is linked to the calculation unit 114 as shown by arrow 117. In embodiments, the output interface may be connected to a data repository, a network interface, a printer and the like. The look-up table Σ constructed thanks to the above described method is stored in the data repository 113. The device also comprises an input interface 118 able to receive an optical transmission system description and to transmit it to the calculation unit 114, as shown by arrow 119.

[0153] The optical transmission system description is constructed thanks to the above described method for determining optical transmissions description as follows:

[0154] The system description defines a plurality of sections S.sub.k of the optical transmission system and, for each section S.sub.k, an input power P.sub.k of the system segment, a local dispersion value of the system segment F.sub.k and an input cumulative dispersion D.sub.k of the system section S.sub.k.

[0155] The calculation unit 114 is configured to perform, for each system segment S.sub.k of the system description received from the input interface 118 the selection of a variance entry σ.sub.match(k)match(k) in the look-up table stored in the data repository 113, so that the local dispersion value DF.sub.k and input cumulative dispersion value D.sub.k of the system segment k substantially match the local dispersion value DF.sub.match(k) and input cumulative dispersion value D.sub.match(k) associated with the variance entry σ.sub.match(k)match(k).

[0156] For example, the input cumulative value D.sub.k substantially matches D.sub.match(k) if

[00007]$\frac{.Math.D_{\mathrm{match}\left(k\right)}-D_k.Math.}{.Math.D_k.Math.}<\mathrm{.Math.},$

wherein ε=5%.

[0157] The calculation unit 114 is also configured to perform, for each couple of this system segments S.sub.k and S.sub.k, the selection of a covariance entry σ.sub.match(k)match(k′) in the look-up table, so that the local dispersion value DF.sub.k and input cumulative dispersion value D.sub.k of the system segment S.sub.k, substantially match the first couple S.sub.k and S.sub.k, associated with the covariance entry of the look-up table σ.sub.match(k)match(k′) and so that the local dispersion and input cumulative dispersion of the system segment S.sub.k, substantially match the second couple associated with the covariance entry.

[0158] The calculation unit 114 is also configured to perform the calculation of the SNR.sub.NL.sup.−1, that is a quality of transmission estimate, as follows:

[00008]${\mathrm{SNR}}_{\mathrm{NL}}^{-1}=\underset{k=1}{\overset{N}{.Math.}}.Math.P_k^2.Math.{\sigma }_{\mathrm{match}\left(k\right).Math.\mathrm{match}\left(k\right)}+2.Math.\underset{k=1}{\overset{N}{.Math.}}.Math.\underset{k^{\prime }=1}{\overset{k-1}{.Math.}}.Math.P_k.Math.P_{k^{\prime }}.Math.\mathrm{Re}\left[{\sigma }_{\mathrm{match}\left(k\right).Math.\mathrm{match}\left(k^{\prime }\right)}\right]$

[0159] The device is further configured for transmitting the calculated quality of transmission estimate SNR.sup.−1.sub.NL through the output interface 115.

[0160] Turning back to the example of optical link ab which look-up table Σ.sub.1 is represented on FIG. 2, the matching performed by the calculation unit 114 leads to a covariance matrix Σ.sub.1 represented on FIG. 3 comprising all the selected variances and covariances corresponding to the characterizing couples F.sub.k and D.sub.k of the sections S.sub.k of the optical link comprising the two spans 123 and 456.

[0161] As the F.sub.k is the same for the two spans 123 and 456, and as no in-line dispersion is comprised on the optical link ab, the covariance matrix Σ.sub.1 represented on FIG. 3 is constructed thanks to the look-up table Σ.sub.1 represented on FIG. 2 by selecting the upper-left 3×3 sub-matrix σ.sub.123 represented in a solid-line frame comprising the variance of the first span 123, the down-right 3×3 sub-matrix σ.sub.456 represented in a dashed-line frame comprising the variance of the second span 456 and the other two 3×3 sub-matrices α.sub.c and α.sub.c′ respectively represented in a dotted-line frame and in a dash-dotted line frame comprising the covariance terms between the span 123 and the span 456. In this particular case, there is no overlap between the used sub-matrices of Σ.sub.1, while all its elements are used. Indeed, in this particular example, match(k)=k for each k from 1 to 6.

[0162] In an embodiment, the look-up table is a covariance matrix which has been constructed by blocks. Indeed, for two covariance matrices X and Y, a joint covariance matrix Σ.sub.X,Y of X and Y may be written in the following block form:

[00009]${\Sigma }_{X,Y}=\left[\begin{array}{cc}{\Sigma }_{\mathrm{XX}}& {\Sigma }_{\mathrm{XY}}\\ {\Sigma }_{\mathrm{YX}}& {\Sigma }_{\mathrm{YY}}\end{array}\right]$

where Σ.sub.XX=var(X), Σ.sub.YY=var(Y) and Σ.sub.XY=Σ.sub.YX.sup.T=cov(X,Y).

[0163] For example, by using the method above described, a first covariance matrix has been constructed for a first fiber span 123 and a second covariance matrix has been constructed for a second fiber span 456. The look-up table comprises the covariance matrix Σ.sub.1 calculated from the first covariance matrix σ.sub.123 and second covariance matrice σ.sub.456.

[0164] Similarly, for example, by using the method above described, a first covariance matrix may be constructed for a first local dispersion value and a second covariance matrix may be constructed for a second dispersion value. In that case, the look-up table comprises the covariance matrix calculated from the first and second covariance matrices. Therefore, with two types of fiber, the look-up table comprises an extended matrix having more rows and columns

[0165] In a second illustrative example, represented on FIGS. 4 to 6, the same covariance matrix construction is performed from the pre-calculated look-up table Σ.sub.1 for an optical link cd different from the optical link ab. Elements which are similar to those of FIGS. 1 to 3 are represented by the same numbers. Optical link cd has a partial dispersion compensation at the end of the first span 123, as represented on the position 13. In this example, the matching is performed by the device represented on FIG. 8, as shown by the above table.

TABLE-US-00001 k Match (k) 1 1 2 2 3 3 4 2 5 3 6 4

[0166] Therefore, a fraction of the matrix Σ.sub.1 is to be used to construct the matrix Σ.sub.2 represented on FIG. 6. As shown on FIG. 5, the sub-matrices corresponding to the variance and covariance matching terms overlap.

[0167] The SNR.sup.−1.sub.NL is then calculated as follows:

[00010]${\mathrm{SNR}}_{\mathrm{NL}}^{-1}=\underset{k=1}{\overset{N}{.Math.}}.Math.P_k^2.Math.{\sigma }_{\mathrm{match}\left(k\right).Math.\mathrm{match}\left(k\right)}+2.Math.\underset{k=1}{\overset{N}{.Math.}}.Math.\underset{k^{\prime }=1}{\overset{k-1}{.Math.}}.Math.P_k.Math.P_{k^{\prime }}.Math.\mathrm{Re}\left[{\sigma }_{\mathrm{match}\left(k\right).Math.\mathrm{match}\left(k^{\prime }\right)}\right]$

[0168] Wherein P.sub.k denotes the input power of the section S.sub.k.

[0169] In an embodiment, the power P.sub.k is calculated as follows:

P.sub.k=P.sub.1e.sup.−α(z.sup.k.sup.−z.sup.1)

[0170] Wherein P.sub.1 denotes the input power of the optical link AB, α denotes the absorption and z.sub.k the abscissa of the input of the section S.sub.k.

[0171] The device is therefore able to provide an estimation of the quality of transmission SNR.sub.NL.sup.−1 for an optical system which is not yet estimated, for example cd, thanks to a look-up table constructed as described above, for the optical link ab.

[0172] For the sake of comparison, the numerical simulation by SSFM of a 9-channel Polarization Division Multiplexed Quaternary Phase Shift Keying PDM-QPSK signal over 100 km, requires about 15 minutes running on a server with a CPU at 2.67 GHz and 16 Gb of memory. This simulation duration makes SSFM simulations unsuitable for real-time applications. By contrast, the device above described allows performing the same calculation in much shorter time, e.g. less than a few seconds.

[0173] The above device is fast and accurate for estimating the performance of transmission of an optical link in an optical network. In a preferred embodiment, the section length is comprised between 100 m and 20 km.

[0174] Therefore, the above device is able to estimate the performance of optical links having a minimal length L.sub.s equal to the section length, i.e. between 100 m and 20 km. FIG. 10 is a graph showing the numerical simulation results of the variance over the section length calculated thanks to the method of FIG. 7, compared to the variance calculated with SSFM simulations. The vertical axis 129 represents the variance in dB for a range from −50 to −35 dB and the horizontal axis 128 represents the section length in km for a range from 0 to 20 km. The round markers 131 represents an example of the nonlinear distortion variance evolution for a Single Mode Fiber SMF span of a maximum length of 20 km and the triangle markers 130 represents the evolution of the nonlinear variance estimation considering the covariance matrix constructed thanks to the method above described. The cumulated dispersion of the span is fixed to 30000 ps/nm and the input power is Pin=0 dBm. An excellent accuracy is observed as the triangle markers match the round markers.

[0175] In the above embodiments, the Kerr effect has been presented as the main source of nonlinear impairments of the optical signals for the sake of simplicity. The optical Kerr effect can result in impairments on the optical signal for many reasons listed below: optical bistability, self-focusing effect, self-phase modulation, solitons formation and others.

[0176] Many other possible nonlinear effects from the list below may be taken into account: second order nonlinearities, second harmonic generation, frequency mixing, optical parametric amplification and oscillation, spontaneous parametric down conversion, sources of entangled photons based on SPDC, third order nonlinear effects as four-wave mixing, Raman scattering, spontaneous and stimulated Raman scattering, Raman amplification, Brillouin Scattering, two photons absorption and others.

[0177] The computation device described hereinabove may be implemented through the use of dedicated hardware as well as hardware capable of executing software in association with appropriate software. When provided by a processor, the corresponding functions may be provided by a single dedicated processor, by a single shared processor, or by a plurality of individual processors, some of which may be shared. Moreover, explicit use of the term “processor” or “controller” should not be construed to refer exclusively to hardware capable of executing software, and may implicitly include, without limitation, central processing unit (CPU), digital signal processor (DSP) hardware, network processor, application specific integrated circuit (ASIC), field programmable gate array (FPGA), read-only memory (ROM) for storing software, random access memory (RAM), and non-volatile storage. Other hardware, conventional and/or custom, may also be included.

[0178] The computation device above described may be implemented in a unitary manner or in a distributed manner.

[0179] The invention is not limited to the described embodiments. The appended claims are to be construed as embodying all modification and alternative constructions that may be occurred to one skilled in the art, which fairly fall within the basic teaching here, set forth.

[0180] The use of the verb “to comprise” or “to include” and its conjugations does not exclude the presence of elements or steps other than those stated in a claim. Furthermore, the use of the article “a” or “an” preceding an element or step does not exclude the presence of a plurality of such elements or steps.

[0181] In the claims, any reference signs placed between parentheses shall not be construed as limiting the scope of the claims.