Patent Grant
10727940
The technical field generally relates to optical signal characterization in telecommunication applications, and more particularly, to a method and apparatus for spectrally characterizing optical signals propagating along an optical communication link, for example to discriminate data-carrying signal contributions from noise contributions and/or to perform optical signal-to-noise ratio (OSNR) measurements.
The performance of long-haul fiber-optic communication systems depends largely on the optical signal-to-noise ratio (OSNR). The OSNR is a conventional measurable characteristic of the quality of a signal carried by an optical telecommunication link. The dominant noise component in an optical communication link is typically unpolarized amplified spontaneous emission (ASE) noise. ASE noise is a spectrally broadband noise source contributed by the optical amplifiers in the link. In practice, the OSNR must therefore be measured somewhere along the link, for example at the receiver end. Current state-of-the-art systems achieve high spectral efficiency using both sophisticated modulation schemes and polarization multiplexing (PM). However, traditional OSNR measurement techniques fail in cases where both densely-packed channels and PM signals are combined. That is, the ASE noise spectrum generally cannot be measured outside of the signal-spectrum bandwidth because the channels are too closely spaced. Meanwhile, in-band measurement methods of the polarization-nulling type, which rely on the fact that the signal is 100% polarized whereas ASE is unpolarized, also suffer from drawbacks because the overall PM signal is also unpolarized.
A method of measuring noise level on PM signals using an acquired optical spectrum trace of the signal was proposed in co-owned U.S. Pat. No. 9,112,604 B2, the disclosure of which is incorporated herein by reference in its entirety. This method is based on the knowledge of the spectral shape of the data-carrying signal contribution provided by a reference signal. From this knowledge, the data-carrying signal contribution and the ASE-noise contribution, which otherwise appear as merged together on the optical spectrum trace, may be mathematically discriminated from each other. Knowledge of the spectral shape of the signal contribution may be derived from a prior acquisition of a reference signal taken. For example, the reference signal can be acquired at a different point, generally upstream, along the same optical communication link, where the OSNR is known or where the signal can be considered free of ASE noise. As such, the reference signal originates from the same optical transmitter as the signal under test. The method described in U.S. Pat. No. 9,112,604 B2 assumes that, within the optical signal bandwidth, the spectral shape of the signal does not significantly change along the communication link. The signal contribution of such a reference signal is therefore considered to be spectrally representative of the signal contribution of the signal under test. U.S. Pat. Appl. Pub. Nos. 2014/0328586 A1 and 2016/0127074 A1, the disclosures of which are incorporated herein by reference in their entirety, include provisions to account for spectral shape variations originating, for example, from nonlinear effects.
A method for determining in-band OSNR and other quality parameters in optical information signals, for example PM signals, is disclosed in U.S. Pat. Appl. Pub. No. 2016/0164599 A1. The method involves measuring an optical power spectrum of a noisy signal, for example by means of a conventional optical spectrum analyzer, and subsequently measuring correlations between a predetermined pair of spaced-apart time-varying frequency components in the optical amplitude or power/intensity spectrum of the signal by means of two optically narrow-band amplitude or power detectors. The in-band noise in the signal may be determined from the correlation measurement. A measurement of the signal power may be used to determine the OSNR based on the determined in-band noise. A drawback of this method is the complexity and high cost of the required apparatus, notably involving two full coherent receivers.
Challenges therefore remain in the development of techniques for discriminating signal from noise in optical signals used in telecommunication applications.
The present description generally relates to techniques for spectrally characterizing an optical signal propagating along an optical communication link, for example an optical fiber. The optical signal can include a data-carrying signal contribution modulated at a symbol frequency within a data-carrying signal bandwidth, and a noise contribution including ASE noise. The present techniques leverage a spectral correlation property that exists within pairs of spectral components spectrally spaced apart by a quantity corresponding to the symbol frequency in the optical signal to allow an in-band determination of the spectrum shape of the data-carrying signal contribution discriminated from the spectrum shape of the noise contribution. Given the shape of both the noisy and noise-free spectra of the signal at the measurement location along the link, the OSNR or other signal quality parameters can be determined. A feature of some embodiments of the present techniques is the use of a large number of such pairs of correlated spectral components over the spectrum of the optical signal. As such, although deviations from perfect spectral correlation may affect the overall signal spectrum, they will generally preserve its shape. Such deviations can be due, for example, to chromatic dispersion (CD) and/or polarization-mode dispersion (PMD). Another feature of some embodiments of the present techniques is the use of relative spectrum shapes, which can eliminate or at least reduce the need for high-precision measurements compared to absolute measurement techniques.
In accordance with an aspect, there is provided a method for spectrally characterizing an optical signal propagating along an optical communication link, the optical signal including a data-carrying signal contribution modulated at a symbol frequency within a data-carrying signal bandwidth and a noise contribution. The method includes:
In some implementations, obtaining the solution representing the data carrying signal power spectrum contribution includes:
In some implementations, obtaining the solution representing the data-carrying signal power spectrum contribution includes:
In some implementations, the nonlinear regression model includes a set of adjustable parameters, the adjustable parameters including at least one of:
In some implementations, the method further includes determining an optical signal-to-noise ratio (OSNR) based on the measured optical power spectrum and the solution representing the data-carrying signal power spectrum contribution.
In some implementations, the method further includes averaging the measured optical power spectrum and the measured spectral correlation function over a plurality of polarization states of the optical signal.
In some implementations, determining, from measurement, the measured spectral correlation function includes measuring a beatnote amplitude function for a set of pairs of beating components respectively associated with the set of pairs of spectral components spectrally separated by the symbol frequency in the optical signal, the beating components in each pair being spectrally separated from each other by a beatnote frequency lower than the symbol frequency, the beatnote amplitude function representing the measured spectral correlation function
In some implementations, a ratio of the symbol frequency to the beatnote frequency ranges from 103 to 106, preferably from 104 to 105.
In some implementations, measuring the beatnote amplitude function includes:
In some implementations, detecting and spectrally resolving the double-sideband signal further includes:
In some implementations, measuring the beatnote amplitude function includes:
In some implementations, the method further includes reducing harmonics in the double-sideband LO signal.
In some implementations, the measured optical power spectrum and the measured spectral correlation function are obtained using a heterodyne optical spectrum analyzer.
It is to be noted that other method and process steps may be performed prior, during or after the above-described steps. The order of one or more of the steps may also differ, and some of the steps may be omitted, repeated and/or combined, depending on the application. It is also to be noted that some method steps can be performed using various signal processing and computational techniques, which can be implemented in hardware, software, firmware or any combination thereof.
In accordance with another aspect, there is provided a non-transitory computer readable storage medium having stored thereon computer readable instructions that, when executed by a processor, cause the processor to perform a method for spectrally characterizing an optical signal propagating along an optical communication link, the optical signal including a data-carrying signal contribution modulated at a symbol frequency within a data-carrying signal bandwidth and a noise contribution. The method includes:
In some implementations, obtaining the solution representing the data carrying signal power spectrum contribution includes:
In some implementations, obtaining the solution representing the data-carrying signal power spectrum contribution includes:
In some implementations, the nonlinear regression model includes a set of adjustable parameters, the adjustable parameters including at least one of:
In some implementations, the method further includes determining an optical signal-to-noise ratio (OSNR) based on the measured optical power spectrum and the solution representing the data-carrying signal power spectrum contribution.
According to another aspect, there is provided a system or apparatus capable of implementing one or more steps of the above methods.
In some implementations, there is provided a system for spectrally characterizing an optical signal propagating along an optical communication link, the optical signal including a data-carrying signal contribution modulated at a symbol frequency within a data-carrying signal bandwidth and a noise contribution. The system includes:
In some implementations, the spectrally resolved detector unit further includes:
In some implementations, the spectral shifter is an electro-optical modulator.
In some implementations, the system further includes a polarization analyzer disposed upstream of the spectrally resolved detector unit, the polarization analyzer including:
In some implementations, the system further includes an optical channel selector configured to select the optical signal from a selected one of a plurality of spaced-apart optical channels.
In some implementations, there is provided a system for spectrally characterizing an optical signal propagating along an optical communication link, the optical signal including a data-carrying signal contribution modulated at a symbol frequency within a data-carrying signal bandwidth and a noise contribution. The system includes:
In some implementations, the LO spectral shifter is an electro-optical modulator.
In some implementations, the system further includes a polarization analyzer disposed upstream of the spectrally resolved detector unit, the polarization analyzer including:
Other features and advantages of the present description will become more apparent upon reading of the following non-restrictive description of specific embodiments thereof, given by way of example only, with reference to the appended drawings. Although specific features described in the above summary and in the detailed description below may be described with respect to specific embodiments or aspects, it should be noted that these specific features can be combined with one another, unless stated otherwise.
In the present description, similar features in the drawings have generally been given similar reference numerals. To avoid cluttering certain figures, some elements may not be indicated, if they were already identified in a preceding figure. It should also be understood that the elements of the drawings are not necessarily depicted to scale, since emphasis is placed on clearly illustrating the elements and structures of the present embodiments. Furthermore, positional descriptors indicating the location and/or orientation of one element with respect to another element are used herein for ease and clarity of description. Unless otherwise indicated, these positional descriptors should be taken in the context of the figures and should not be considered limiting. More particularly, it will be understood that such spatially relative terms are intended to encompass different orientations in the use or operation of the present embodiments, in addition to the orientations exemplified in the figures.
Unless stated otherwise, the terms “connected” and “coupled”, and derivatives and variants thereof, refer herein to any connection or coupling, either direct or indirect, between two or more elements. For example, the connection or coupling between the elements may be mechanical, optical, electrical, magnetic, logical, or a combination thereof.
In the present description, the term “measured” when referring to a quantity or parameter is intended to mean that the quantity or parameter can be measured either directly or indirectly. In the case of indirect measurement, the quantity or parameter can be derived, retrieved, inferred or otherwise determined from directly measured data.
The terms “a”, “an” and “one” are defined herein to mean “at least one”, that is, these terms do not exclude a plural number of elements, unless stated otherwise. It should also be noted that terms such as “substantially”, “generally” and “about”, that modify a value, condition or characteristic of a feature of an exemplary embodiment, should be understood to mean that the value, condition or characteristic is defined within tolerances that are acceptable for the proper operation of this exemplary embodiment for its intended application.
The terms “light” and “optical” are used to refer herein to radiation in any appropriate region of the electromagnetic spectrum. More particularly, these terms are not limited to visible light, but can also include invisible regions of the electromagnetic spectrum including, without limitation, the terahertz (THz), infrared (IR) and ultraviolet (UV) spectral bands. For example, in non-limiting embodiments, the imaging systems that can implement the present techniques can be sensitive to light having a wavelength band lying somewhere in the range from about 1250 nm to about 1650 nm. Those skilled in the art will understand, however, that this wavelength range is provided for illustrative purposes only and that the present techniques may operate beyond this range.
The present description discloses various implementations of techniques for in-band—that is, within the signal bandwidth—spectral characterization of an optical signal propagating along an optical communication link, for example an optical fiber.
The present techniques may be useful in various applications where it is desired or required to measure spectral properties of an optical signal in telecommunication applications. For example, some of the present techniques can be applied to, or implemented in, different types of optical communication networks including, without limitation, typical metro and long-haul systems using signal modulation schemes in the ITU grid, such as QPSK or M-QAM (where M can be, e.g., 16, 32, 64, 128 or 256) at 28 Gb/s and higher rates. The signals can be pulse-shaped or not, and be polarized or polarization-multiplexed. The present techniques may be used to ensure or help ensure that an optical network is reliable and operates within acceptable industry specifications. The present techniques can be implemented in various environments and settings, including field-deployed networks, manufacturing facilities for network equipment, research and development laboratories, data centers, and the like. Furthermore, the present techniques can be employed during the installation, activation and/or operation phases of an optical communication network for characterization, error diagnosis and troubleshooting, and/or performing monitoring.
In some implementations, there is provided an in-band measurement method allowing determination of the spectrum shape of a PM signal discriminated from the ASE-noise background (ASE-free) at the measurement location. Given both the signal-plus-ASE and the ASE-free spectrum-shapes of the signal at the measurement location, the OSNR can be computed according to any standard or custom definitions. The present techniques, however, are not limited to OSNR measurements and have a larger scope of application, since they involve the determination of spectra over a large spectral range, both with and without ASE noise, and not only a single OSNR value. Computing the OSNR from this data can be viewed as one specific application among a variety of other possible applications.
Some implementations of the present techniques leverage an intrinsic or natural spectral correlation property of optical signals to provide a reference signal-spectrum measured “in situ” at a measurement location or point along an optical communication link. This reference signal-spectrum can then be used in OSNR measurement methods, such as the one described in co-owned U.S. Pat. No. 9,112,604 B2. However, in its full generality, the present techniques can provide a mathematical analysis of measured data based on the physics of the situation, where this measured data can be obtained with different systems and apparatuses. For example, theory, principles and specific algorithms of some embodiments disclosed herein are based on definitions of measured physical quantities regardless of the actual apparatus implementation.
Some implementations of the present techniques can overcome or reduce drawbacks or limitations of the method disclosed in U.S. Pat. Appl. Pub. No. 2016/0164599 A1. Such drawbacks and limitations can include the high-precision measurement requirements that stem from the assumption of a perfect spectral correlation between the predetermined pair of frequency components and from the absolute nature of the measurements, and the potential lack of robustness resulting from the high-accuracy measurements and the fine adjustments and calibration that the implementation of this method would require.
Method Implementations
Various aspects of a method for spectrally characterizing an optical signal propagating along an optical communication link will now be described. The optical signal can generally be described as including two contributions: a data-carrying signal contribution modulated at a symbol frequency within a data-carrying signal bandwidth, and a noise contribution that typically includes ASE noise.
Referring to
The method 100 can also include a step 104 of determining, from measurement, a spectral correlation function for a set of pairs of spectral components of the optical signal under test. In each pair, the spectral components are spectrally separated from each other by the symbol frequency and are centered on a respective one of a set of center frequency values in a center frequency range within the spectral range. The measured spectral correlation function represents a variation in an intensity of correlation within the pairs of spectral components as a function of center frequency. As described in greater detail below, in some implementations, the step 104 of determining the spectral correlation function can include steps of measuring a beatnote amplitude function for a set of pairs of beating components respectively associated with the set of pairs of spectral components, and identifying the measured beatnote amplitude function as the spectral correlation function to be determined. In such a case, the beating components in each pair are spectrally separated from each other by a beatnote frequency, which is the same for all the pairs and substantially lower than the symbol frequency. The beatnote amplitude function represents a variation in the beatnote amplitude between the beating components of each pair as a function of the center frequency of the associated pair of spectral components.
The method 100 can further include a step 106 of obtaining, using a processor, a solution representing the data-carrying signal power spectrum contribution of the measured optical power spectrum. The solution, which uses the measured optical power spectrum as an input, is such that a calculated spectral correlation function for pairs of spectral components separated by the symbol frequency in the solution matches the measured spectral correlation function. In the present description, the terms “match”, “matching” and “matched” are meant to encompass not only “exactly” or “identically” matching the measured and calculated spectral correlation functions, but also “substantially”, “approximately” or “subjectively” matching the measured and calculated spectral correlation functions, as well as providing a higher or best match among a plurality of matching possibilities. The terms “match”, “matching” and “matched” are therefore intended to refer herein to a condition in which two elements are either the same or within some predetermined tolerance of each other. Furthermore, the term “representing” is used in this context to indicate that the match between the solution found by the method 100 and the data-carrying signal power spectrum contribution of the actual measured value of the optical power spectrum can be either absolute or relative. For example, in some implementations, the solution can be a normalized spectrum proportional to the data-carrying signal power spectrum contribution of the measured value of the optical power spectrum.
In some implementations, the step 106 of obtaining the solution for the data-carrying signal power spectrum contribution can include a step of performing a nonlinear regression analysis based on a nonlinear regression model. The nonlinear regression model can use a set of adjustable parameters to relate the measured optical power spectrum and the measured spectral correlation over the center frequency range. In some implementations, the adjustable parameters can include, without limitation, a model function representing a normalized model of the noise power spectrum contribution (which can itself be defined by one or more adjustable parameters); a noise-to-signal ratio parameter representative of a relative amplitude of the noise power spectrum contribution with respect to the data-carrying signal power spectrum contribution; and a dispersion parameter conveying information indicative of chromatic dispersion and/or polarization mode dispersion of the optical signal along the optical communication link. The measured data can be fitted using a method of successive approximations, in which an initial set of parameter values is iteratively refined until a suitable match is obtained between the model and the measured data.
In some implementations, the method 100 can include a step of determining an optical signal-to-noise ratio (OSNR) based on the measured optical power spectrum and the solution representing the data-carrying signal power spectrum contribution. For example, in some cases, the method 100 can yield normalized solutions for the data-carrying signal power spectrum contribution and the noise power spectrum contributions and a noise-to-signal ratio, and these three parameters can be used to determine the OSNR.
These and other possible method steps and variants according to the present techniques will be described in greater detail below.
Intrinsic Spectral-Correlation Property
Modulated signals have a “hidden” periodicity, even if the individual symbols are mutually independent random variables. This is reflected in the frequency domain (Fourier transform, spectrum) as a theoretical 100% correlation between spectral components separated by exactly fsb, where fsb is the symbol frequency (baud rate). More precisely, for any random sequence of symbols, the two spectral components of all such pairs have the same phase difference. Considering discrete signals (sampled), this can be expressed at the transmitter end as,
Definition: C(f⋅)≡|(f+)·
*(f−)
sb| (1)
Value: C(f⋅)≡[P(f+)·P(f−)]1/2
sbδ(f
(f±)|2, (2)
where (f⋅) is a Jones vector representing the discrete Fourier transform (DFT) of the time-varying optical field
(t) (two-component complex phasor), that is, the modulated optical carrier with state of polarization (SOP) included,
. . .
sb represents an average over all possible sequences of symbols (sb), and C(f⋅) is the correlation between the two spectral components f±=(f⋅±½fsb), where f⋅ is the center frequency. In other words, Eq. (1) indicates that the correlation between these components is equal to the square root of the product of their powers, and zero if (f+−f−)≠fsb. This signal property is made use of in embodiments described herein. Indeed, since ASE noise and other noise contributions generally do not have this property (C=0), knowledge of C(f⋅) gives, according to Eq. (2), the power of the signal at frequencies f+ and f−, or more exactly the square root of the product of the two powers, which in practice is integrated over a small frequency range [finite resolution, window w(f), see, e.g.,
Definitions of Spectral Quantities
Referring to
Definition of Measured Data
In some implementations, the quantities that constitute the measured data—regardless of the apparatus used to measure them—are as follows. These quantities are shown as thick solid lines in
Psη(f) is the optical power spectrum (i.e., “signal+ASE” spectrum) of the optical signal to be characterized: Psη(f)=Ps(f)+Pη(f) (3)
The values of Psη(f) at f+ and f− are shown as black circles in
{tilde over (P)}(f⋅) is a correlation term, referred to herein as the “spectral correlation function”, corresponding to a power representative of correlation (1). It can be written in all generality as
{tilde over (P)}2(f⋅)=|∫w2(ƒ)
(f++ƒ)·
*(f−+ƒ)dƒ|2
sb, where f±=(f⋅±½fsb) and w2(f)=|w(f)|2. (4a)
When w(ƒ) is made narrow enough to be approximated as w(ƒ)=δ(ƒ) (Dirac delta function), Eq. (4a) reduces to
{tilde over (P)}2(f⋅)=P+(f⋅)P−(f⋅), (4b)
Other Definitions
Referring to
Step 0: Raw-Data Acquisition
It is assumed that regardless of the apparatus used to implement the method 300 in
More particularly, step 0 in
Step 0 can also include a step of determining, from measurement, a spectral correlation function {tilde over (P)}2(f⋅) within a set of pairs of spectral components spectrally separated by the symbol frequency fsb in the optical signal under test. The set of pairs of spectral components are respectively centered on a corresponding set of center frequencies over a center frequency range. As mentioned above, the function {tilde over (P)}2(f⋅) describes how the correlation between the spectral components within each pair varies as a function of the center frequency of the pair.
In some implementations, step 0 of the method 300 may further include a step of determining, from measurement, a value for the DC term P:(f⋅). The determination of the DC term P:(f⋅) can be made using a different measurement process than the one used to measure optical power spectrum Psη(f). This different measurement process can be the same as the one used to determine the spectral correlation function {tilde over (P)}2(f⋅).
The Square Modulus:
It should be noted that a difference exists between Eq. (1) and the definition of Eq. (4a) for {tilde over (P)}(f⋅), in that the modulus operator | . . . | is inside the averaging operator . . .
sb in Eq. (4a). A consequence of this difference is an offset δ∥(f⋅) adding to {tilde over (P)}2(f⋅) compared to (4b), that is,
{tilde over (P)}2(f⋅)=P+(f⋅)P−(f⋅)+δ∥(f⋅), where δ∥(f⋅)˜Psη(f+)Psη(f−)÷[1+Beq·Δt] with Beq=∫−∞∞w(ƒ)dƒ/w(0), (7)
where Δt is the finite duration of the signal (t) from which the DFT
(t) is obtained, and Beq is the equivalent width of w(f) (i.e., a rectangular window of width Beq gives the same).
It is to be noted that the seemingly inconvenient offset δ∥ becomes negligible if Beq·Δt>>1, which is expected to be readily fulfilled in practice. For example, in the case of a Gaussian window of ENB, Bw=100 MHz, corresponding to Beq=141 MHz, even an acquisition time Δt as short as 10 millisecond (ms) already gives Beq·Δt=1.4×106. Consequently, the otherwise advantageous average of the square modulus can be used because, while (f) is the DFT of a finite-duration sequence, which implies a finite resolution (bin) df=1/Δt, the correlation is integrated over a width Beq much larger than df. This imposes a lower bound to Bw given some Δt, or vice versa, but this lower bound is generally not an expected limitation. An advantage of taking the average of the square modulus is that it is generally insensitive to CD and PMD, which can remove the requirement present in some existing systems to measure CD and PMD.
Step 1: Normalization of Psη(f)
The method 300 can include a step 1 of computing a normalized value sη(f) for the measured optical power spectrum Psη(f). Step 1 can involve dividing each data point of the measured optical power spectrum Psη(f) by the value of Psη(f) measured at the center (i.e., at f=0) of the data-carrying signal bandwidth of the optical signal under test, such that
sη(f)=Psη(f)/Psη(0) with
sη(0)=1.
It is to be noted that, in the present description, all normalized spectra represented by the symbol “” have the same definition, that is,
(0)≡1 (definition of a normalized spectrum). (8)
Step 2: Scaling of {tilde over (P)}(f⋅) and P:(f⋅)
As mentioned above, in some implementations, the method 300 in
It is to be noted that depending on the application, step 2 may be omitted, but can be useful to provide a quantitative physical interpretation of the parameter α found by nonlinear regression (see, e.g., step 4 below). It is also to be noted that, in practice, the measurement process to obtain {tilde over (P)}(f⋅) and P:(f⋅) may differ from the measurement process to obtain Psη(f), so that the raw determined values for P:(f⋅) may differ from those which would be obtained from Eq. (6) using the measured value of Psη(f). In such a case, the scaling step 2 can involve multiplying both the DC term P:(f⋅) and the spectral correlation term {tilde over (P)}(f⋅) by a common scaling factor αN. The scaling factor αN is chosen such that when P:(f⋅) is scaled by αN, the relationship P:(f⋅)=½[sη(f+)+
sη(f−)] is fulfilled, where the right-hand side is computed from the measured normalized spectrum
sη(f). It should be noted that, in general, measuring or determining the parameter P:(f⋅) may be omitted when the scaling step 2 is not performed.
Step 3: Guess Values
Referring still to s(f)—from
sη(f), {tilde over (P)}(f⋅) and, optionally, P:(f⋅). Performing a nonlinear regression analysis generally involves providing initial values or estimates for a set of adjustable parameters used in a nonlinear regression model that relates the measured optical power spectrum and the measured spectral correlation over a certain spectral range. Various techniques exist for determining initial estimates of a nonlinear regression analysis and the choice of initial estimates may influence the convergence of the algorithm.
In the present embodiment, the set of adjustable parameters can be represented as a vector denoted V. The vector V can include the parameters (ρα), introduced below, and a set of Nη additional parameters grouped in the vector Vη and representing the adjustable parameters of a model function ĝη(Vη, f). This can be written as
V=(ραVη). (9)
Definition:
In some embodiments, the model function ĝη(Vη, f) is defined as a function of f that is normalized according to Eq. (8), that is, ĝη(Vη, 0)≡1, where Vη is a vector whose Nη elements represents Nη adjustable parameters. Different values of Vη correspond to different shapes of the curve ĝη(f) as a function of f. Depending on the application, the Nη elements of Vη can be scalars, vectors, matrices or any other appropriate mathematical entity.
Condition:
In some implementations, a necessary and sufficient condition regarding ĝη(Vη, f) is the existence of a value of Vη such that ĝη(Vη, f) is substantially equal to the actual normalized ASE-noise spectrum η(f). In general, there is no a priori prescription regarding the form of ĝη, the number Nη of parameters, or the nature of the parameters (scalars, vectors, matrices, etc.). A simple form of ĝη(Vη, f) with a few scalar parameters may be sufficient in some cases, and a more complex function with more parameters may be needed in other cases. It is foreseeable that simple shapes of
η(f) over the width of a channel slot that are still common presently, such as a simple constant, a straight line or a parabola, may change as system architectures evolve. The determination of whether a particular form for ĝη(Vη, f) is sufficiently general will depend on the application, notably on the required or desired accuracy of the regression analysis. Non-limiting examples of model ĝη(Vη, f) that may be used for
η(f) are provided in Table I below.
As mentioned above, various techniques can be used to establish guess values in nonlinear regression. For example, in some embodiments, both ρ and α can be initially set to zero in V=(ραVη) and be iteratively refined as the regression progresses.
Step 4: Nonlinear Least-Square Regression
In the present embodiment, the measured data include the optical power spectrum Psη(f), the spectral correlation function {tilde over (P)}(f⋅) and, optionally, the DC term P:(f⋅). This means, more particularly, that the data-carrying signal power spectrum contribution Ps(f) and the noise power spectrum contribution Pη(f) of the measured optical power spectrum Psη(f) are a priori unknown at the start of the method 300.
The method 300 of
In some implementations, a nonlinear least-square regression analysis can be used to find the unknown ASE-noise spectrum Pη(f) such that Eq. (4b) is satisfied, or in other words, find Pη(f) such that the following difference is zero—or minimized to an appropriate degree—all over the range of f⋅,(f⋅)=Ps(f+)·Ps(f−)−{tilde over (P)}2(f⋅)=0, (10)
where the unknown Ps(f) can be written as,
Ps(f)=Psη(f)−Pη(f). (11)
Substituting Eq. (11) for Ps(f) in Eq. (10), yields(f⋅)=[Psη(f+)−Pη(f+)]·[Psη(f−)−Pη(f−)]−{tilde over (P)}2(f⋅), (12)
in which the only unknown is Pη(f). Once Pη(f) is found by minimizing (f⋅), the signal spectrum Ps(f), which is to be obtained in accordance with the present method 300, can be found from Eq. (11).
In some implementations, Eqs. (10) to (12) may be expressed differently to facilitate their use. For example, defining the following normalized spectra in accordance with Eq. (8) yieldss(f)=Ps(f)/Ps0 and
η(f)=Pη(f)/Pη0, where Ps0=Ps(0) and Pη0=Pη(0), (13)
from which Eq. (11) can be normalized as follows:s(f)=(1+ρ)·
sη(f)−ρ·
η(f)↔
sη(f)=[
s(f)+ρ·
η(f)]/(1+ρ), where ρ=Pη0/Ps0. (14)
Then, using Eq. (14) to express Eq. (10) in terms of normalized spectra yields
Δ(f⋅)=[(1+ρ)sη(f+)−ρ·
η(f+)]·[(1+ρ)
sη(f−)−ρ·
η(f−)]−α2·{tilde over (P)}2(f⋅)=0, (15)
where Δ(f⋅)=(f⋅)/P20. In Eq. (15),
sη(f±) and {tilde over (P)}(f⋅) are the experimental data, and (ρα) and
η(f) are the unknowns.
To carry out the nonlinear regression analysis, η(f) in Eq. (15) can be replaced by the model function ĝη(Vη, f), where the vector Vη represents the Nη adjustable parameters of the model function ĝη. The adjustable parameters of the regression are V=(ραVη). The experimental data
sη(f±) and P(f⋅) can be fitted using a method of successive approximations. In such a case, an initial set of parameter values for vector V is iteratively refined until a suitable match is obtained between the model and the measured data, that is, until Δ(f⋅) in Eq. (15), as computed with the parameters V=(ραVη) found by the regression, approaches zero or a predetermined value to within a specified degree of tolerance.
It is also within the scope of the present techniques for the regression to be performed iteratively. In this case, a first regression is performed using a first form of ĝη, then if the resulting difference Δ0(f⋅) deviates significantly from zero or from another convergence criterion, a second regression is performed with a different, generally more complex form of ĝη selected on the basis of the result Δ0(f⋅) of the first regression, then a third regression on the basis of the result Δ1(f⋅) of the second regression, and so on, if necessary, until the result Δq(f⋅) of the (q−1)th regression (qth iteration) matches or satisfies a specified criteria or degree of substantial “zeroness”.
Meaning of ρ:
ρ is a noise-to-signal ratio parameter at f=0, proportional to the ASE ratio rη=1/OSNR (see step 5 below for more details).
Meaning of α:
α is a scaling factor for {tilde over (P)}(f⋅) that can convey information indicative of propagation effects, such as of CD and/or PMD, affecting the optical signal under test along the optical communication link. In the absence of propagation effects such as CD and PMD, it has been found from simulation results that α becomes substantially equal to (1+ρ) or αN·(1+ρ), depending on whether step 2 is performed or not. Consequently, if CD and PMD can be neglected and step 2 is performed, then, in principle, α can be omitted as an independent adjustable parameter in Eq. (15), and be replaced by (1+ρ). However, in practice, CD and PMD are rarely negligible, so that α generally cannot be omitted. The effect of CD and PMD is to reduce the value of the correlation term {tilde over (P)}(f⋅), with respect to the value given by Eqs. (4a) and (4b), by a global relative correlation factor Cr, referred to herein as the “relative correlation”, that is substantially independent of f⋅ in practice. The factor Cr can be accounted for through the adjustable parameter α. The fact that Cr is substantially independent of f⋅, on the one hand, and the introduction of the parameter α in the regression, on the other hand, can be advantageous of some implementations of the method 300. In this way, both CD and PMD can be compensated for with a single adjustable parameter, α, and without having to be measured.
The normalized data-carrying signal spectrum, s(f), can be determined from the result for
η(f) obtained by the regression, as follows:
s(f)=(1+ρ)·
sη(f)−
η(f), where
η(f)={circumflex over (g)}η(Vη,f). (16)
In some implementations, the error Δs(f) on
s(f) as a function of the difference Δĝη(f) between the model function and the actual normalized ASE-noise spectrum can be approximated by
Δs(f)˜−rηΔgη(f). (17)
It is noted that if the normalized data-carrying signal power spectrum contribution s(f) obtained using the present techniques is used as the reference signal-spectrum defined in U.S. Pat. No. 9,112,604 B2, Eq. (17) implies that, at worst, the relative error on the measured ASE noise ratio would be equal to some average of the relative difference Δĝη.
Step 5: Computation of the OSNR
Referring still to the embodiment of s(f) for the data-carrying signal power spectrum contribution, the normalized solution
η(f)=ĝη(Vη, f) for the noise power spectrum, and the adjustable parameter ρ. Indeed, with the knowledge of
s(f),
η(f)=ĝη(Vη, f) and ρ, the OSNR may be computed according to various standard definitions. For example, from the vocabulary used in the present description, a traditional definition of the OSNR can be written as,
rη=∫−f
In Eq. (18), Ss(f) and Sη(f) are the spectral densities of the data-carrying signal and ASE noise respectively, Bch is the width of the channel slot, and fr=½Bν where Bν is the standard reference bandwidth. In some applications, the standard reference bandwidth is specified as a wavelength interval Bλ=0.1 nm. In such a case, Bν is given as a function of Bλ by Bν·(c/λ2i), where λi is the center wavelength of the ith channel, and c is the speed of light. According to this definition, the ASE-noise ratio and OSNR can be computed from the measured normalized spectra as follows,
rη=ρ·∫−fη(f)df÷∫−1/2B
s(f)df OSNRdB=−10 log(rη), (19)
where OSNRdB is the OSNR expressed in dB, that is, OSNRdB=10 log(OSNR). It is noted that Eq. (19) assumes that the window w(f) is narrow enough in the sense that,
∫−1/2Bs(f)df≈∫−1/2B
η(f)df≈∫−f
where Šs(f) and Šη(f) are the normalized spectral densities of the signal and ASE noise, respectively. Since a narrow w(f) is typically used (e.g., 50-500 MHz), the two approximate equalities in Eq. (20) are generally accurate in practice.
System Implementations
In accordance with another aspect, there is provided a system or apparatus for spectrally characterizing an optical signal propagating along an optical communication link and capable of implementing the methods described herein. As will be described in greater detail below, some system and apparatus implementations described herein can use what is referred to herein as a “low-frequency beatnote” (LFB) approach. In the LFB approach, the spectral correlation function within pairs of spectral components spectrally separated by the symbol frequency in the optical signal under test is determined from a LFB amplitude function whose measurement can involve bringing the spectral components of each pair spectrally closer together and measuring a low-frequency beatnote therebetween.
Referring to
The system 400 includes a spectral shifter 406 configured to generate, in a first acquisition mode, a double-sideband signal 408 from the optical signal. The double-sideband signal 408 includes a first image signal 410a and a second image signal 410b. The first and second image signals 410a, 410b represent sideband images of the optical signal 402 that separated from each other by a spectral shift 2δ↔ equal to the symbol frequency fsb plus or minus a beatnote frequency fb, the case 2δ↔=(fsb+fb) being represented in
The system 400 also includes a spectrally resolved detector unit 412 operable within a spectral range in the data-carrying signal bandwidth. The spectrally resolved detector unit 412 is configured to, in the first acquisition mode, detect and spectrally resolve the double-sideband signal 408 and output a first detected signal 414, and, in the second acquisition mode, detect and spectrally resolve the optical signal 402 and output a second detected signal 416. Because of the provision of the optical switch 446, the optical signal 402 and the double-sideband signal 408 in the illustrated embodiment are detected one at a time and independently from each other. However, in other embodiments, it could be envisioned that the optical signal 402 and the combined optical signal 408 be detected at least partly concurrently, for example if the spectrally resolved detector unit includes multiple detectors and/or delay lines.
In the present description, the term “spectrally resolved detector unit” broadly refers to any optical detector or receiver, or combination of optical detectors, capable of measuring a spectrally dependent response (e.g., as a function of frequency or wavelength over a certain spectral range) of an input signal. The or each optical detector in the spectrally resolved detector unit 412 generally operates as an opto-electrical receiver configured for receiving an input signal and for outputting an electrical signal representing the received input signal. The electrical signal can be sampled and digitized as spectral data representative of the detected input signal. The spectrally resolved detector unit 412 can be embodied by, or be part of, different types of spectrally sensitive detectors, including optical spectrum analyzers (OSAs), swept-wavelength systems or any other type of spectral measurement devices. In some implementations, the spectrally resolved detector unit 412 can be a high-resolution OSA, non-limiting examples of which can include a heterodyne OSA, a coherent OSA, a Brillouin OSA and a coherent receiver.
The system 400 further includes a processor 418 coupled to the spectrally resolved detector unit 412. The processor 418 is configured to receive, after analog-to-digital conversion, the first and second detected signals 414, 416, and to deriving therefrom spectral information about the double-sideband signal 408 and the optical signal 402, respectively.
In the present description, the term “processor” refers to an entity that controls and executes, at least in part, the operation of the system 400. The processor 418 can be provided within one or more general purpose computers and/or within any other dedicated computing devices. It should be noted that the term “processor” should not be construed as being limited to a single processor, and accordingly any known processor architecture may be used. The processor 418 can be implemented in hardware, software, firmware, or any combination thereof, and be connected to various components of the system 400 via appropriate communication links and ports. The processor 418 may be embodied by a computer, a microprocessor, a microcontroller, a central processing unit (CPU), a programmable logic device such as, for example, a field-programmable gate array (FPGA), or any other type of processing resource or any combination of such processing resources configured to operate collectively as a processor. The processor 418 may include or be coupled to one or more memory elements 420 capable of storing computer programs and other data to be retrieved by the processor 418. Each memory element 420 can also be referred to as a “computer-readable storage medium”. Depending on the application, the processor 418 may be integrated, partially integrated or physically separate from the optical hardware of the system 400, including, but not limited to, the spectral shifter 406 and the spectrally resolved detector unit 412.
In
The processor 418 is also configured to determine a spectral correlation function {tilde over (P)}(f⋅) for a set of pairs of spectral components of the optical signal 402, the set of pairs being respectively centered on a corresponding set of center frequencies f⋅ over a center frequency range within the spectral range. The spectral components in each pair are spectrally separated from each other by the symbol frequency fsb. In the embodiment of
Once the optical power spectrum Psη(f) and the spectral correlation function {tilde over (P)}(f⋅) have been determined as measured data, the processor 418 can be configured to obtain a solution representing the data-carrying signal power spectrum contribution Ps(f), such that a calculated spectral correlation computed based on the solution and in view of the measured optical power spectrum Psη(f) matches the measured spectral correlation function {tilde over (P)}(f⋅). This step can be carried out such as described, for example by using a nonlinear regression analysis.
It is to be noted that in some implementations, the processing steps described above may, but need not, be performed by the same physical processor. For example, it may be envisioned to determine the spectral correlation function {tilde over (P)}(f⋅) and the optical power spectrum Psη(f) based on the first and second detected signals 414, 416, respectively, using a processor included in the spectrally resolved detector unit 412, and to obtain the solution representing the data-carrying signal power spectrum contribution Ps(f) with another processor, separate from the optical hardware of the system 400.
More detail regarding the structure, configuration and operation of these and other possible components and features of the system according to the present techniques will be described in greater detail below.
Referring to
The channel λ-selector 522 is configured to receive (at point “IN” in
The function of the channel λ-selector 522 is to filter out a given channel i from all the other channels, as illustrated by the spectra IN and [1]. In some implementations, the measurements are performed sequentially, channel by channel: the signal spectrum of a first channel i is measured; then a second channel is selected, and its signal spectrum measured; then a third channel is selected, and so on. Depending on the application, the channel λ-selector 522 may be embodied by a tunable filter, an array of fixed-filters with optical switches, a wavelength-division multiplexing (WDM) demultiplexer, or any other commonly available device that can perform or be configured to perform the function described above.
It is to be noted that with respect to the terminology used herein, each one of the N optical channels can be said to contain an optical signal 502 to be spectrally characterized. As such, the optical channel selector 522 is configured to select the currently characterized optical signal 502 from a selected one of a plurality of spaced-apart optical channels Ni. The optical signal 502 in a selected channel can therefore be described as including two contributions, namely a data-carrying signal contribution modulated at a symbol frequency fsb within a data-carrying signal bandwidth corresponding to spectral width of the selected channel, and a noise contribution that typically includes ASE noise. The optical signal 502 in a selected channel also has an optical power spectrum Psη(f) including a data-carrying signal power spectrum contribution Ps(f) associated with the data-carrying signal contribution and a noise power spectrum contribution Pη(f) associated with the noise contribution.
The LFB approach introduced above can be implemented based on the function provided by the 2-ν shifter 506, which corresponds to the spectral shifter 406 of the embodiment of
The system 500 of
The principles of the LFB approach may be more easily understood if presented first from an alternative point of view in which there would be no 2-ν shifter and no channel λ-selector. Instead, the incoming lightwave 524 at IN in
In contrast, in the LFB approach depicted in the embodiment of
In current fiber-optic systems, the symbol frequency fsb can be up to 25 GHz and more. On the other hand, Bw can be expected to lie in the range from a few tens to a few hundreds of MHz, so indeed Bw<<fsb is generally satisfied. Accordingly, a value of a few hundreds of kHz to about 2 MHz should generally be both large enough and a judicious setting for the beat frequency fb, so that the inequalities fb<<Bw<<fsb are satisfied. Therefore, considering what would be a typical case with fsb=25 GHz and fb=1 MHz, fb is indeed a very low frequency compared to fsb.
For further digital processing the output of the photodetector, the beatnote P(t, f\f⋅), is sampled and converted to digital data. For this purpose, the system 500 of
Referring still to
Shifter-Off:
When the 2-ν shifter 506 is put off or bypassed (second acquisition mode), the mean value of P(t, f) over an acquisition time Δt constitutes the measurement of the optical power spectrum Psη(f) of the optical signal 502. In this case, no specific processing is required. That is, the processor 518 can receive spectral data 516 corresponding to the detected optical signal 502 and simply identify or determine that this spectral data 516 represent Psη(f). In some implementations, Psη(f) may be averaged by the processor 518 over a number K of acquisitions.
Shifter-on:
When the 2-ν shifter 506 is put on (first acquisition mode), the square modulus of the DFT of the sampled beatnote 514, P(t), can be computed as
S(f)=|DFT[P(t)]|2. (21)
In general, S(f) exhibits three large narrow peaks or tones over a comparatively small, wideband background as schematically depicted by graph [4] in
{tilde over (P)}2(f⋅)=½[S(fb)+S(−fb)], (22)
P:2(f⋅)=¼S(0). (23)
In some implementations, the above procedure can be done K times while keeping the same value of f or f⋅ either when the 2-ν shifter 506 is on (f⋅) or off (f). For example, when the 2-ν shifter 506 is on, the measured value {tilde over (P)}2k obtained according to Eq. (22) is stored in memory at each iteration k, k=0 . . . (K−1), and the averaged result is computed as {tilde over (P)}2(f⋅)=(1/K)Σk{tilde over (P)}2k. Alternatively, averaging can be performed through an accumulated sum, that is, a first value {tilde over (P)}20 is measured and stored in memory as first sum Σ(0), then a second value {tilde over (P)}21 is measured and added to first sum to obtain second sum Σ(1)=Σ(0)+{tilde over (P)}21, then a third value {tilde over (P)}22 is measured and added to second sum to obtain third sum Σ(2)=Σ(1)+{tilde over (P)}22, and so on, computing sum k at each iteration as Σ(k)=Σ(k−1)+{tilde over (P)}2k, the averaged result being obtained as {tilde over (P)}2(f⋅)=(1/K) Σ(K−1). Yet another alternative is averaging the whole curves Sk(f) instead of the single values {tilde over (P)}2k; the procedure is the same as described above, just by replacing {tilde over (P)}2k by Sk(f). Then {tilde over (P)}2 is computed according to Eq. (22) from the averaged S(f). An advantage of this last alternative is that an eventually non-negligible background of additive noise (electronics) may be determined and subtracted with greater accuracy. In some embodiments, P:2 may also be averaged along with {tilde over (P)}2, using a procedure that may correspond to any of the above alternatives, and likewise for Psη(f) when the 2-ν shifter 506 is off.
As per the above discussion regarding step 0 and Eq. (7), the duration Δt of the sampled beatnote, from which S(f) is computed, may be selected such that Beq·Δt>>1. When averaging is performed as described above, the relevant Δt to be put in Eq. (7) becomes the acquisition time of individual iterations, rather than the total acquisition time K·Δt. Therefore, unless for other practical reasons (see, e.g., the exemplary embodiment of
Reference is now made to
It is to be noted that if the measurements of {tilde over (P)}(f⋅) were performed strictly such as described above with reference to the embodiments of
In the present embodiment, the polarization analyzer 728 is located before the 2-ν spectral shifter 706 in
The time-varying polarization state selected by the polarization analyzer 728 can be represented by a three-dimensional normalized Stokes vector â (â·â=1), which can be viewed as the position of a point on the surface of a unit sphere, known as the Poincaré sphere. In some implementations, it may be desirable, or required, that the polarization scrambling performed by the polarization scrambler 730 be such that the vector â can reach any point on the surface of the Poincaré sphere, so as to cover approximately its entire surface over a measurement period, preferably uniformly (e.g., uniform random scrambling). However, it has been found that in many cases, the uniformity condition is readily fulfilled, as it is generally sufficient â can reach at least one point in each of the eight octants of the Poincaré sphere. More particularly, it is also generally best avoided to have all possible values of â concentrated around only one or two points on the sphere.
In some embodiments, to randomize the axis â of the polarization scrambler 730 to get meaningful mean-values by averaging over K random polarization state settings corresponding to K random settings of â, it is often preferred to perform the averaging from K acquisitions with acquisition time Δt, so that â is changed between each iteration, instead of performing a single acquisition with acquisition time K·Δt. In this way, a desirable random scrambling of a can be performed without increasing the overall measurement time. Of course, some embodiments may use different acquisition schemes without departing from the scope of the present techniques. In some implementations, it has been found that only a few different settings of â can be sufficient to avoid vanishing values of Cr, and that for physical interpretation purposes, meaningful mean values can already be obtained with K=10. Furthermore, it has also been found that a value of K=100 can be a reasonable and adequate default value in some applications.
Referring now to
In
vm(t)=(Δφ/π)·vπ cos(ωmt), with Δφ=asin(ΔA) and ωm=2πfm, (24)
where ΔA is the modulation depth of the amplitude. In this configuration, a 100% modulation depth (ΔA=1) can be achieved, corresponding to a phase swing Δφ=π/2, but then non-negligible odd harmonics at ±3fm are generated in the Fourier transform of the output lightwave 808, instead of only the two desired images 810a, 810b at ±fm. The amplitude c3 of the third harmonic relative to the amplitude c1 of the fundamental at fm is then r3=c3/c1=0.122. In some cases, if not accounted for, such value of the relative amplitude r3 of the spurious image could lead to non-negligible measurement errors. In some embodiments, two solutions can be combined to overcome or at least alleviate the effect of the third harmonic: a) reducing the modulation depth ΔA, thus reducing r3 accordingly, the downside being an equivalent overall power loss, as if the power of the input lightwave were smaller; and b) taking the non-negligible third harmonic—and other harmonics, if required—into account in the nonlinear regression analysis described above.
In some implementations, it may be envisioned to deal with such spurious images, provided that the relative amplitudes rq of the harmonics |q|=2 . . . q+ in question are reasonably well known. Instead of Eq. (15), a somewhat more complex expression of the difference Δ(f⋅) to be minimized could be used, where the contribution Δ{tilde over (P)}h(f⋅) of the spurious harmonics to the correlation term would be considered, although the overall analysis would remain essentially the same. For completeness, the contribution of the spurious harmonics to the correlation term may be written as
Δ{tilde over (P)}h(f⋅)=Σq=[2,q
In
Returning to
In the illustrated embodiment, the optical heterodyne OSA 712 can generally include a local oscillator (LO) source 750, an optical coupler 752, a heterodyne receiver 754, and a sweep controller 756. The LO source 750, for example a laser source, generates an LO signal 758 having a tunable LO frequency v and sends it toward the optical coupler 752, for example a PM fiber (PMF) coupler. In the first acquisition mode (spectral shifter 706 on), the optical coupler 752 combines the LO signal 758 and the double-sideband signal 708 into a first combined signal 760, while in the second acquisition mode (spectral shifter 706 off), the optical coupler 752 combines the LO signal 758 and the optical signal 702 into a second combined signal 762.
The heterodyne receiver 754 is configured to, in the first acquisition mode, receive the first combined signal 760 and generate therefrom the first detected signal 714, and, in the second acquisition mode, receive the second combined signal 762 and generated therefrom the second detected signal 716. The heterodyne receiver 754 can include a set of photodetectors 764, for example in a balanced detection scheme, to convert detected light signals into electrical signals, and electronic circuit 766 to further process the electrical signals before outputting them as detected signals 714, 716. For example, the electronic circuit 766 can include a bandpass electronic filter 768 followed by a square-law or power detector 770 including a lowpass filter 772.
The sweep controller 756 is coupled to the LO source 750 for sweeping, in both the first and second acquisition modes, the tunable LO frequency ν of the LO signal 758 within the spectral range of interest.
As described above, the detected signals 714, 716 can be sent to the processor 718 for determining or identifying the measured spectral correlation function {tilde over (P)}(f⋅) and optical power spectrum Psη(f), and for obtaining therefrom a solution representing the data-carrying signal power spectrum contribution Ps(f) of Psη(f).
It should be noted that in the present embodiment, the optical section of the heterodyne OSA 712, whose function is to mix the input signal—the optical signal 702 or double-sideband signal 708 depending on the acquisition mode with the LO signal 758, need not be provided with polarization diversity. This is because the polarization of the input signal 702, 708, which is determined by the polarizer 732 of the polarization analyzer 728, is fixed and known. Therefore, if the optical path between the polarization analyzer 728 and the optical coupler 752 of the heterodyne OSA 712 consists of PM fibers, the known polarization state of the input signal 702, 708 at the output of the polarization analyzer 728 would generally be preserved at the input of the optical coupler 752. This feature can simplify the optical section of the heterodyne OSA 712.
It should also be noted that a scanning OSA whose function is to display spectra includes an electronic or digital lowpass filter 772 at the output of the photodetectors 764, whose bandwidth is typically only a few tens of kHz or even much less if slow or low-resolution scans are acceptable. In the present embodiment, what is observed as output of the square-law or power detector 770 in the first acquisition mode is the beatnote amplitude function, from which the spectral correlation function {tilde over (P)}(f⋅) is to be determined. Therefore, the cutoff frequency fc of this lowpass filter 772 is preferably higher than the selected beat frequency fb which, as previously mentioned, can be expected to be around 1 MHz, for example in the range from a few hundreds of kHz to 2 MHz.
Referring now to
The heterodyne OSA 1012 in
The optical coupler 1052 is configured to, in the first acquisition mode (LO spectral shifter 1078 on), combine the double-sideband LO signal 1076 and the optical signal 1002 into a first combined signal 1060, and, in the second acquisition mode (LO spectral shifter 1078 off), combine the LO signal 1058 and the optical signal 1002 into a second combined signal 1062. As in
As described above, the detected signals 1014, 1016 can be sent to the processor 1018 for determining or identifying the measured spectral correlation function {tilde over (P)}(f⋅) and optical power spectrum Psη(f), and for obtaining therefrom a solution representing the data-carrying signal power spectrum contribution Ps(f) of Psη(f).
It is noted that compared to the embodiment of
In some implementations, several advantages can result from applying the double-sideband spectral shifting to the LO signal rather than to the optical signal. First, the use of a channel λ-selector may in principle be avoided and, consequently, the scans of f and f⋅ could encompass all channels in one acquisition, since there would be no need to acquire the raw data sequentially channel by channel Second, the output of the LO spectral shifter could be used to actively lock the operating point through a feedback loop if a Mach-Zehnder modulator such as that of
In embodiments such as
Referring now to
In some implementations, under the assumption that condition 1 is satisfied from careful manufacturing of the DDMZ LO spectral shifter 1378, the remaining conditions can be met by finely setting the parameters mentioned in conditions 2 to 4, based on the power of harmonic q=2 in the spectrum of the 2-ν-shifted lightwave.
Referring still to
In some implementations, undesirable harmonics in the spectrum of the 2-ν-shifted laser lightwave may be suppressed or at least significantly reduced by filtering them out using a tunable 2-ν optical filter, which can be a tunable optical filter having two narrow (i.e., <<fm) passbands centered at ν±=(ν±fm), where ν is the optical frequency of the tunable laser. Referring to
According to another aspect, there is provided a non-transitory computer readable storage medium having stored thereon computer readable instructions that, when executed by a processor, cause the processor to perform a method for spectrally characterizing an optical signal propagating along an optical communication link, such as disclosed herein.
In the present description, the terms “computer readable storage medium” and “computer readable memory” are intended to refer to a non-transitory and tangible computer product that can store and communicate executable instructions for the implementation of various steps of the method disclosed herein. The computer readable memory can be any computer data storage device or assembly of such devices, including random-access memory (RAM), dynamic RAM, read-only memory (ROM), magnetic storage devices such as hard disk drives, solid state drives, floppy disks and magnetic tape, optical storage devices such as compact discs (CDs or CDROMs), digital video discs (DVD) and Blu-Ray™ discs; flash drive memories, and/or other non-transitory memory technologies. A plurality of such storage devices may be provided, as can be understood by those skilled in the art. The computer readable memory may be associated with, coupled to, or included in a computer or processor configured to execute instructions contained in a computer program stored in the computer readable memory and relating to various functions associated with the computer.
In some implementations, the computer program stored in the computer readable storage medium can instruct a processor to perform the following steps: receiving a measured optical power spectrum of the optical signal over a spectral range within the data-carrying signal bandwidth, the measured optical power spectrum including a data-carrying signal power spectrum contribution associated with the data-carrying signal contribution of the optical signal and a noise power spectrum contribution associated with the noise contribution of the optical signal; receiving a measured spectral correlation function for a set of pairs of spectral components of the optical signal, the spectral components in each pair being spectrally separated from each other by the symbol frequency, the set of pairs being respectively centered on a corresponding set of center frequencies over a center frequency range within the spectral range, the measured spectral correlation function relating a correlation intensity between the spectral components of each pair to the center frequency of the pair over the center frequency range; and obtaining a solution representing the data-carrying signal power spectrum contribution based on the measured optical power spectrum of the optical signal, such that a calculated spectral correlation function for pairs of spectral components spectrally separated by the symbol frequency in the solution representing the data-carrying signal power spectrum contribution matches the measured spectral correlation function.
In some implementations, the step of obtaining the data carrying signal power spectrum contribution can include determining a solution representing the noise power spectrum contribution, and deriving the solution representing the data-carrying signal power spectrum contribution from the solution representing the noise power spectrum contribution and the measured optical power spectrum.
In some implementations, the step of obtaining the solution representing the data-carrying signal power spectrum contribution can include providing a nonlinear regression model relating the measured optical power spectrum and the measured spectral correlation function, and using the nonlinear regression model to determine the solution representing the data-carrying signal power spectrum contribution. In some implementations, the nonlinear regression model can include a set of adjustable parameters, such as: a model function representing a normalized model of the noise power spectrum contribution; a noise-to-signal ratio parameter representative of a relative amplitude of the noise power spectrum contribution with respect to the data-carrying signal power spectrum contribution; and a dispersion parameter conveying information indicative of chromatic dispersion and/or polarization mode dispersion of the optical signal.
In some implementations, the method can further include a step of determining an optical signal-to-noise ratio (OSNR) based on the measured optical power spectrum and the solution representing the data-carrying signal power spectrum contribution.
It will be understood that the methods and systems described herein can find applications in maintenance, monitoring and/or troubleshooting.
Although the above description refers to portable test instruments (such as portable OSAs), it should be mentioned that some signal characterization methods described herein may be used for monitoring applications which employ fixed (as opposed to portable) test instruments.
It should be noted that all the equations provided herein as a function of frequency, could be adapted to be expressed as a function of wavelength, wave number, or the like. Accordingly, all the equations given herein could be readily adapted to find their equivalent as a function of the wavelength or wave number.
It should be appreciated that the methods described above are not limited to the characterization of an optical signal having a unique signal carrier wavelength. The optical signal under test may include a plurality of data-carrying signal contributions multiplexed using Nyquist wavelength division multiplexing (N-WDM) (also referred to as “superchannels” in the scientific literature), such as dual-carrier PM-16-QAM (which is currently deployed for 400 G transmission) or all-optical orthogonal frequency-division multiplexing (OFDM), for example, provided that the variation of the signal portion of such an optical signal under test is significantly greater than the ASE-noise variation across at least a portion of the optical signal bandwidth.
Of course, numerous modifications could be made to the embodiments described above without departing from the scope of the appended claims.
The present application is a continuation of International Application No. PCT/IB2018/052423 filed Apr. 6, 2018, which claims priority from U.S. Provisional Patent Application No. 62/490,113 filed on Apr. 26, 2017, the specifications of which are incorporated herein by reference in their entirety.
| Number | Name | Date | Kind |
|---|---|---|---|
| 6256103 | Sorin et al. | Jul 2001 | B1 |
| 6515276 | Baney et al. | Feb 2003 | B2 |
| 6898001 | Ishihara | May 2005 | B2 |
| 9008508 | Liu | Apr 2015 | B2 |
| 9112604 | Gariepy | Aug 2015 | B2 |
| 9413455 | Oda | Aug 2016 | B2 |
| 9438336 | He | Sep 2016 | B2 |
| 9596027 | He | Mar 2017 | B2 |
| 9825698 | Heismann | Nov 2017 | B2 |
| 9954610 | He | Apr 2018 | B2 |
| 10128960 | Okabe | Nov 2018 | B2 |
| 10128975 | He | Nov 2018 | B2 |
| 10348412 | Xue | Jul 2019 | B1 |
| 10567077 | Oyama | Feb 2020 | B2 |
| 20140328586 | He et al. | Nov 2014 | A1 |
| 20160072579 | Schroeder | Mar 2016 | A1 |
| 20160127074 | He et al. | May 2016 | A1 |
| 20160164599 | Heismann | Jun 2016 | A1 |
| 20180034550 | Rakich | Feb 2018 | A1 |
| 20190187051 | Green | Jun 2019 | A1 |
| Entry |
|---|
| Moench et al., Measurement of Optical Signal-to-Noise-Ratio in Coherent Systems using Polarization Multiplexed Transmission, Optical Fiber Communication Conference 2017 Technical Digest, Th2A.42.pdf, 2017, 3 pages. |
| Number | Date | Country | |
|---|---|---|---|
| 20190305845 A1 | Oct 2019 | US |
| Number | Date | Country | |
|---|---|---|---|
| 62490113 | Apr 2017 | US |
| Number | Date | Country | |
|---|---|---|---|
| Parent | PCT/IB2018/052423 | Apr 2018 | US |
| Child | 16445992 | US |
