IN-BAND OPTICAL SIGNAL TO NOISE RATIO DETERMINATION METHOD AND SYSTEM

Abstract

There is provided a method for determining the in-band noise in agile multichannel Dense Wavelength Division Multiplexing (DWDM) optical systems, where the interchannel noise is not representative of the in-band noise in the optical channel. The method relies on the analysis of two observations of the same input optical signal. In the two observations, the linear relationship between the optical signal contribution and the optical noise contribution (i.e. the observed OSNR) is different, which allows the discrimination of the signal and noise contributions in the input optical signal. In a first approach, the two observations are provided by polarization analysis of the input optical signal. In a second, the input optical signal is obtained using two different integration widths.

Description

TECHNICAL FIELD

The invention relates to the determination of the in-band noise in optical telecommunication applications. More specifically, the invention relates to the determination of the in-band noise in Dense Wavelength Division Multiplexing (DWDM) optical networks.

BACKGROUND

The IEC 61280-2-9 Fiber-optic communication subsystem test procedures Part 2-9 standards (ed. 1.0 b:2002) provides a standard method for determining OSNR in DWDM networks. This method is based on the assumption that the interchannel noise level is representative of the noise level at the signal peak position. The method interpolates the power level of the noise outside the signal bandwidth to evaluate the in-band noise in the signal bandwidth. Increased modulation rates, which enlarge the signal bandwidth, and increased channel density reduce the interchannel width, therefore resulting in severe spectral characteristics requirements for the optical spectrum analyzers used to perform the measurement. The procedures described in the standards are able to cope with these difficulties when the noise level of adjacent peaks is mostly continuous. For example, the standards propose a two-scan procedure to first measure a broad modulated peak with a larger resolution bandwidth to capture the entire signal peak and then determine the noise using a narrow resolution bandwidth to minimize the contributions of the main and adjacent peaks on the interchannel noise level.

Alternatively, commercial Optical Spectrum Analyzers (OSA) (such as EXFO's FTB-5240, in its versions available before 2007) implement a somewhat equivalent procedure by performing an integrated peak calculation and fine noise determination in a single scan.

However, to strictly comply with the standards recommendation, the noise level should be determined at the mid-channel spacing between peaks. In the case where noise is spectrally filtered with the signal peak, after passing through multiplexers or demultiplexers—such as Reconfigurable Optical Add Drop Multiplexers (ROADM)—the mid-spacing noise level is no longer representative of the in-band noise level, which is the relevant parameter for the OSNR determination. The interpolation of the interchannel noise level becomes unreliable. This can be mitigated by relying on a very sharp spectral response of the OSA filter and adaptive processing to determine the noise level at the shoulders where the noise meets the base of a signal peak within the channel bandwidth. However, increased modulation rates combined with narrow filtering of multiplexers and demultiplexers is making it increasingly difficult to achieve a reliable measurement of the noise level within the channel bandwidth.

Active polarization nulling (see J. H. Lee et al., “OSNR Monitoring Technique Using Polarization-Nulling Method”, IEEE Photonics Technology Letters, Vol. 13, No. 1, January 2001) provides an alternative to a direct analysis of the optical spectrum. This method uses the fact that the signal peak is generally polarized while the noise is generally unpolarized. Using a polarization controller cascaded with a polarizer, it is possible to actively control the polarization of the input signal in order to find a condition where the signal peak is substantially suppressed by the polarizer. An optical spectrum trace is acquired while the signal peak is suppressed and reveals the in-band noise within the optical channel bandwidth, discriminated from the signal peak by the signal peak suppression. The noise level within the optical channel bandwidth can be determined using the acquired optical spectrum trace.

Other methods for determining the OSNR of an optical signal have also been proposed. One of them is disclosed in U.S. Patent Application Pub. No. 2006/0098980 to Lee et al. This method uses a dithering function for measuring the noise and the signal simultaneously using a different resolution bandwidth and in a single measurement step, as opposed to the IEC method where the measurement is done in two optical spectrum scans.

SUMMARY

There is provided a method for determining a noise parameter, such as the in-band noise or the Optical Signal-to-Noise Ratio (OSNR), of a Dense Wavelength Division Multiplexing (DWDM) input optical signal having a signal and a noise contribution within an optical signal bandwidth. The method can be implemented with only passive optical devices and a commercially available optical spectrum analyzer (OSA), as opposed to actively controlled polarization controllers. At least two optical spectrum traces of the input optical signal obtained by the OSA are processed to discriminate the noise contribution from the signal contribution.

The method is particularly valuable for determining the in-band noise and thus the OSNR in agile multichannel Dense Wavelength Division Multiplexing (DWDM) optical systems. In such agile systems, optical channels may be added or dropped anywhere along an optical network, after or before being optically amplified. Adding and dropping is typically performed using Optical Add Drop Multiplexers OADM) which not only filter the signal peak corresponding to the optical channel but also filter the noise. The optical noise is filtered with the useful signal peak and is consequently spectrally limited to the channel bandwidth or spectral neighborhood of the optical channel and also varies from one DWDM channel to another. The interchannel noise is therefore not representative of the in-band noise of the optical channel.

The method relies on the analysis of at least two measurements of the same input optical signal which has a useful signal contribution (corresponding to the signal peak) and a noise contribution. In the two measurements, the linear relationship between the signal contribution and the noise contribution, i.e. the observed OSNR, is different which allows the discrimination of the signal and noise contributions in the input optical signal using calculations and a comparison between the two measurements. In a first approach presented herein and which will be referred to as the Passive Polarization-Induced Discrimination (PPID) approach, the two measurements are provided by different polarization analysis of the input optical signal. At least two samples of the input optical signal are taken under different polarization analysis conditions, e.g. polarized according to a different state of polarization (SOP), the optical spectrum of which being acquired to provide the at least two measurements. In one example of this approach, the input optical signal is split using a polarization beam splitter and the two split samples are acquired to provide the two measurements. In a second approach presented herein and referred to as the Differential Resolution Bandwidth Discrimination (DRBD) approach, the input optical signal is acquired using two different Resolution Bandwidths (RBW's).

The different RBW's may be provided using a different filtering slit in the Optical Spectrum Analyzer (OSA) or a single acquired input trace may by integrated numerically to provide the two measurements.

The PPID and the DRBD approaches have some different advantages and drawbacks and can be combined into a hybrid approach where limitations of one are circumvented by using the other.

The method relies on the differential properties of the noise and signal contributions in the input optical signal to be analyzed. Firstly, the signal and noise contributions have different properties in that the signal is typically substantially polarized (or at least partially polarized) where the noise is typically unpolarized (or at least partially unpolarized). The PPID approach takes advantage of this characteristic. Accordingly, the insertion of a polarizer in the optical path of the input optical signal will have a different effect on the noise contribution than on the signal contribution. It is thus possible to provide different optical spectrum traces with different proportions of signal and noise contributions, which allows for their discrimination by assuming that the noise contribution is unpolarized while the signal contribution is polarized. Secondly, the noise contribution is typically spectrally broader than the signal contribution, and more importantly, it varies slowly where the signal varies quickly as compared to the resolution bandwidth of the OSA. The DRBD approach takes advantage of this second characteristic. Accordingly, varying the RBW in different optical spectrum traces of the input optical signal will have a different effect on the noise contribution than on the signal contribution. It is thus possible to provide different optical spectrum traces with different proportions of signal and noise contributions, which allows for their discrimination.

According to one aspect, there is provided a method for determining a noise parameter on an input optical signal having a data-carrying signal contribution and a noise contribution within an optical signal bandwidth. The method comprises the steps of: obtaining at least two optical spectrum traces from the input optical signal, the optical spectrum traces being taken under different conditions such that they show different non-zero signal-to-noise ratios; mathematically discriminating the signal contribution from the noise contribution within the optical signal bandwidth based on the optical spectrum traces; determining an in-band noise level on the input optical signal from the discriminated noise contribution; and determining the noise parameter from the determined in-band noise level, the noise parameter being indicative of the noise contribution within the optical signal bandwidth.

According to another aspect, there is provided a method for determining a noise parameter on an input optical signal having a data-carrying signal contribution and a noise contribution within an optical signal bandwidth, the signal contribution and the noise contribution having at least one of different degrees of polarization and different states of polarization from one another. The method comprises the steps of: acquiring a first and a second optical spectrum traces of the input optical signal using respectively a first and a second polarization analysis conditions, the first and second polarization analysis conditions being different from one another and each being arbitrary relative to the input optical signal, the optical spectrum traces showing different signal to noise ratios; mathematically discriminating the signal contribution from the noise contribution within the optical signal bandwidth using the optical spectrum traces; and determining an in-band noise level on the input optical signal from the discriminated noise contribution.

According to another aspect, there is provided a system for determining an in-band noise on an input optical signal within an optical signal bandwidth. The system comprises: an input for receiving an input optical signal comprising a data-carrying signal contribution and a noise contribution within the optical signal bandwidth, the signal contribution and the noise contribution having at least one of different degrees of polarization and different states of polarization from one another; a polarization optics arrangement for obtaining a first and a second sample of the input optical signal such that the first and second the samples have at least one of different states of polarization relative to the input optical signal and different degrees of polarization from one another, states of polarization of the samples being arbitrary relative to the input optical signal; an optical spectrum analyzer for acquiring a first and a second optical spectrum traces respectively of the first and second samples, the first and second optical spectrum traces showing non-zero signal to noise ratios different from one another; a spectrum processor adapted for mathematically discriminating the noise contribution in the input optical signal within the optical signal bandwidth from the optical spectrum traces using the first and the second optical spectrum traces; and a noise calculator for evaluating the in-band noise within the optical signal bandwidth from the discriminated noise contribution.

According to another aspect, there is provided a method for determining a noise parameter on an input optical signal having a data-carrying signal contribution and a noise contribution within an optical signal bandwidth, the noise contribution varying slowly in wavelength within the optical signal bandwidth compared to the signal contribution. The method comprises the steps of: obtaining a first and a second optical spectrum traces from the input optical signal, respectively corresponding to a first and a second integration widths in order for the first and the second optical spectrum traces to show different non-zero signal-to-noise ratios, the second integration width being larger than the first integration width, the first and the second optical spectrum traces comprising at least one point; mathematically discriminating the noise contribution in the input optical signal within the optical signal bandwidth using the first and the second optical spectrum traces; determining, from the discriminated noise contribution, an in-band noise level on the input optical signal within the optical signal bandwidth; and determining the noise parameter from the determined in-band noise level, the noise parameter being indicative of the noise contribution within the optical signal bandwidth.

According to another aspect, there is provided a method for determining an in-band noise level on an input optical signal having a data-carrying signal contribution and a noise contribution within an optical signal bandwidth, wherein an optical spectrum trace is obtained from the input optical signal, and wherein the noise contribution is discriminated from the signal contribution within the optical signal bandwidth in order to determine an in-band noise level on the input optical signal. The method is characterized in that: at least two optical spectrum traces are obtained from the input optical signal, the optical spectrum traces being taken under different conditions such that they show different non-zero signal to noise ratios; the noise contribution is discriminated from the signal contribution within the optical signal bandwidth using a comparison between the optical spectrum traces; and the in-band noise level on the input optical signal is determined from the discriminated noise contribution.

According to another aspect, there is provided a method for determining an in-band noise level on an input optical signal having a data-carrying signal contribution and a noise contribution within an optical signal bandwidth, wherein the signal contribution and the noise contribution have at least one of different degrees of polarization and different states of polarization from one another, wherein a first optical spectrum trace of the input optical signal is acquired using a first polarization analysis condition, and wherein the noise contribution is discriminated from the signal contribution within the optical signal bandwidth in order to determine an in-band noise level on the input optical signal. The method is characterized by the steps of: acquiring a second optical spectrum trace of the input optical signal using a second polarization analysis condition, the first and second polarization analysis conditions being different from one another and each being arbitrary relative to the input optical signal, the optical spectrum traces showing different signal to noise ratios; mathematically discriminating the noise contribution from the signal contribution within the optical signal bandwidth using a comparison between the optical spectrum traces; and determining the in-band noise level on the input optical signal from the discriminated noise contribution.

According to another aspect, there is provided a system for determining the in-band noise level on an input optical signal having a data-carrying signal contribution and a noise contribution within an optical signal bandwidth, wherein the signal contribution and the noise contribution have at least one of different degrees of polarization and different states of polarization from one another, the system comprising a polarization optics arrangement for polarizing at least part of the input optical signal in order to produce a first sample of the input optical signal, an optical spectrum analyzer configured for acquiring a first optical spectrum trace of the first sample, the noise contribution being discriminated from the signal contribution within the optical signal bandwidth in order to determine an in-band noise level on the input optical signal. The system is characterized in that: the first optical spectrum trace shows a non-zero signal to noise ratio; the polarization optics arrangement is further arranged to produce a second sample of the input optical signal, the first and the second samples having states of polarization relative to the input optical signal different from one another or degrees of polarization different from one another, states of polarization of the first and the second samples being arbitrary relative to the input optical signal; the optical spectrum analyzer is further configured for acquiring a second optical spectrum trace of the second sample, the second optical spectrum trace showing a non-zero signal to noise ratio different from the one of the first optical spectrum trace; the system also comprises a spectrum processor for mathematically discriminating the noise contribution from the signal contribution within the optical signal bandwidth based on the first and the second optical spectrum traces; and a noise calculator for determining the in-band noise within the optical signal bandwidth from the discriminated noise contribution.

According to another aspect, there is provided a method for estimating an in-band noise level on an input optical signal having a data-carrying signal contribution and a noise contribution within an optical signal bandwidth, wherein the noise contribution varies slowly in wavelength within the optical signal bandwidth compared to the signal contribution, and wherein a first and a second optical spectrum traces are obtained from the input optical signal respectively corresponding to a first and a second integration widths, the second integration width being larger than the first integration width. The method is characterized in that: the first and the second optical spectrum traces show different non-zero signal to noise ratios due to the different integration widths; the noise contribution is discriminated from the signal contribution within the optical signal bandwidth using numerical calculations based on the first and the second optical spectrum traces; and the in-band noise level on the input optical signal is determined from the discriminated noise contribution.

BRIEF DESCRIPTION OF THE DRAWINGS

FIG. 1 is a graph illustrating the optical spectrum of an example input optical signal along with the optical spectrum of its noise contribution and of its signal contribution;

FIG. 2 is a flow chart illustrating a general method for determining the in-band noise, the Optical Signal to Noise Ratio (OSNR) or another noise parameter on an input optical signal;

FIG. 3 is a block diagram of the main components of a system for determining a noise parameter on an input optical signal using a passive polarization-induced discrimination method;

FIG. 4A is a block diagram showing a possible polarization optics arrangements to be used in the system of FIG. 3, wherein a polarization beam splitter is used;

FIG. 4B is a block diagram showing another possible polarization optics arrangements to be used in the system of FIG. 3, wherein a 50/50 optical coupler is used along with a polarizer;

FIG. 4C is a block diagram showing another possible polarization optics arrangements to be used in the system of FIG. 3, wherein a polarizer is alternately inserted and removed from the optical path;

FIG. 4D is a block diagram showing another possible polarization optics arrangements to be used in the system of FIG. 3, wherein a polarization controller is placed before a polarizer;

FIG. 5 is a flow chart illustrating a method for determining noise parameter on an input optical signal using a passive polarization-induced discrimination approach;

FIG. 6 is a block diagram of an embodiment of a system for determining a noise parameter on an input optical signal using a passive polarization-induced discrimination method, wherein a dual channel optical spectrum analyzer is used;

FIG. 7 is a flow chart illustrating an example embodiment of the method of FIG. 5, wherein the step of discriminating the signal and noise contributions is shown in more details;

FIG. 8 is a graph illustrating the optical spectrum of an example input optical signal along with its discriminated noise and signal contributions, as calculated with a passive polarization-induced discrimination method;

FIG. 9 is a flow chart illustrating a method for determining noise parameter on an input optical signal using a differential resolution bandwidth discrimination approach;

FIG. 10 is a flow chart illustrating an example embodiment of the method of FIG. 9 with chosen integrating windows to achieve a flat-top signal peak on two optical spectrum traces;

FIG. 11 is a graph illustrating the optical spectrum of an example input optical signal to which the method of FIG. 10 is applied;

FIG. 12 is a flow chart illustrating another example embodiment of the method of FIG. 9 assuming a known shape of the signal contribution; and

FIG. 13 is a flow chart illustrating an example method for determining noise parameter on an input optical signal using a hybrid approach.

It will be noted that throughout the appended drawings, like features are identified by like reference numerals.

DETAILED DESCRIPTION

Now referring to FIG. 1, the methods and systems described herein relate to the characterization of an optical signal p which is used in optical telecommunications to transmit data over a Dense Wavelength Division Multiplexing (DWDM) optical channel. Throughout the present description, the optical signal p corresponds to one of the DWDM optical channels. In the optical channel bandwidth of interest, the optical signal p includes two components, i.e. a signal contribution s arising from the data-carrying signal, and a noise contribution n which includes all other sources of optical power within the optical channel. The noise contribution n arises mostly from the Amplified Spontaneous Emission (ASE) noise of the optical amplifiers in the optical transmission system. FIG. 1 shows the optical spectrum p(λ) of an example optical signal p, along with the optical spectrum of its signal contribution s(λ) and the optical spectrum of its noise contribution n(λ), such that:

p(λ)=s(λ)+n(λ),  (1)

and

• • p=∫_CBWp(λ),
  • s=∫_CBWs(λ),
  • n=∫_CBWn(λ),and where CBW is the Channel BandWidth of interest.

An optical spectrum trace of the optical signal p is acquired by an Optical Spectrum Analyzer (OSA) as a series of data pairs P(λ) and represents the input optical signal p convolved with the filter spectral response of the OSA h_OSA(λ) combined with any desired convolution window h_W(λ). The optical spectrum trace P(λ) is thus the spectrally-resolved optical power of the optical signal p. In a bandwidth corresponding to the channel bandwidth CBW, the optical spectrum trace P(λ) also includes a signal contribution S(λ) and a noise contribution N(λ) which are merged together and appear as the optical spectrum trace P(λ), and:

$\begin{array}{cc}\begin{array}{c}P\left(\lambda \right)=S\left(\lambda \right)+N\left(\lambda \right)\\ =h_w\left(\lambda \right)*h_{\mathrm{OSA}}\left(\lambda \right)*p\left(\lambda \right)\\ =h_w\left(\lambda \right)*h_{\mathrm{OSA}}\left(\lambda \right)*s\left(\lambda \right)+h_w\left(\lambda \right)*h_{\mathrm{OSA}}\left(\lambda \right)*n\left(\lambda \right).\end{array}& \left(2\right)\end{array}$

where h_W(λ) is a convolution window which may be applied numerically and “*” denotes the convolution operation.

The methods and systems described herein are used to discriminate the signal contribution s(λ) from the noise contribution n(λ) in the optical spectrum p(λ) using acquired optical spectrum traces P(λ) in order to determine the in-band noise on the optical signal to be characterized. The instrument noise associated with the detection system itself, namely the OSA, on the acquired optical spectrum trace P(λ) is considered to have a negligible effect compared to the optical noise contribution to be characterized.

FIG. 1 shows a single optical signal p within its corresponding optical channel but it should be noted that according to wavelength division multiplexing a plurality of optical channels shares the optical spectrum, each channel for transmitting one optical signal (not shown). It should however be kept in mind that other optical signals are typically present in the optical spectrum, spectrally on both sides of the optical signal p.

A DWDM optical channel is being defined as a spectral bandwidth, i.e. the channel bandwidth, allocated for the transmission of an optical signal in a WDM transmission scheme. The signal bandwidth is rather the actual width of the signal peak, i.e. the bandwidth over which the signal contribution is non negligible. The channel bandwidth CBW may be larger than or just as large as (over even narrower than) the signal bandwidth, depending on the density of the DWDM channels and the signal transmission rate for a given transmission scheme.

The methods disclosed herein rely on the fact that the properties of the signal and noise contributions within the optical channel are different. First, the signal and noise n contributions have different polarization properties. The signal contribution s is substantially polarized (or at least partially polarized) while the noise contribution n is mostly unpolarized (or at least partially unpolarized). Second, the noise contribution n is spectrally broader and varies slowly compared to the signal contribution p in a spectral slice corresponding to at least the Resolution BandWidth (RBW) defined by the filter spectral response of the OSA. It is possible to use at least one of these different properties to discriminate the signals from the noise contribution n in acquired optical spectrum traces P(λ).

FIG. 2 illustrates the general approach for determining the in-band noise, the Optical Signal to Noise Ratio (OSNR) or another noise parameter on the input optical signal p.

In step 202, at least two optical spectrum trace measurements P_A(λ) and P_B(λ) are obtained from the input optical signal p to be characterized. The two traces P_A(λ), P_B(λ) are taken on two different conditions such that the linear relationship between the signal contribution and the noise contribution, i.e. the observed OSNR, is different on both traces. This allows for the discrimination of the signal and noise contributions in the input optical signal p. As will be explained in more detail when describing the approaches below, the difference between the two traces P_A(λ), P_B(λ) arises from the difference in the optical properties of the signal and the noise contributions, combined with the different conditions on which the optical spectrum traces P_A(λ) and P_B(λ) are obtained.

In step 204, the signal contribution S and the noise contribution N are discriminated from one another in the optical bandwidth of interest using calculations and a comparison between said optical spectrum traces P_A(λ) and P_B(λ). Examples of such calculations that allow signal and noise discrimination are given herein below. Such calculations take into account the difference in properties between the signal and the noise contributions in order to eliminate one or the other in a combination or comparison between the obtained optical spectrum traces P_A(λ) and P_B(λ). It is noted that more than two traces could also be obtained and used in these calculations.

In step 206, the in-band noise level N(λ_p) under the signal peak is determined from the discriminated noise contribution N. Again, example embodiments for this step are described in more details herein below.

In step 208, the noise parameter is determined, by calculations, from the determined in-band noise N(λ_p). The noise parameter is indicative of the noise contribution n within the optical signal bandwidth SB, such as the determined in-band noise level N(λ_p) itself or the OSNR calculated from the determined in-band noise level. The noise parameter is typically output for use in monitoring, installation, commission, maintenance or troubleshooting of a DWDM optical system. It can be graphically or numerically output using a display unit or a printer for example. It can also be output by generating an electrical signal or by storing it in memory for later retrieval.

Two different approaches for determining the in-band noise or the OSNR of an input optical signal p in a DWDM optical system are provided. The first approach will be referred to as the Passive Polarization-Induced Discrimination approach (PPID) and the second as the Differential Resolution Bandwidth Discrimination (DRBD) approach. The two approaches have some different advantages and drawbacks and can be combined into a hybrid approach where deficiencies of one can be circumvented by using the other.

Passive Polarization-Induced Discrimination Approach (PPID)

Referring to FIG. 1, let p(λ) be the optical spectrum of the input optical signal p and having a signal contribution s(λ) and a noise contribution n(λ).

The PPID approach uses the differential properties between the signal contribution s(λ) and the noise contribution n(λ) in the input optical signal to be analyzed. The signal s(λ) and noise n(λ) contributions have different polarization properties in that the signal is typically substantially polarized, but at least partially polarized, where the noise is typically unpolarized, or at least partially unpolarized. In other words, the signal and the noise contributions have different degrees of polarization from one another. This last condition will be assumed for the following description. It is noted, however, that a similar method could also be used if the signal and noise contributions rather had different states of polarization from one another.

A system 10 for a noise parameter on the input optical signal p is shown on FIG. 3. The system receives the input optical signal p to be characterized. Due to the different polarization properties of the signal s and the noise n contribution, the insertion of a polarization optics arrangement 14 in the optical path of the input optical signal p has a different effect on the noise contribution n than on the signal contribution s. The polarization optics arrangement 14 is used to obtain two different samples p_A and p_B of the input optical signal p by applying two different polarization analysis conditions. Different possible polarization optics arrangements 14 are shown in FIGS. 4A-4D and are described below.

An OSA 11 acquires the optical spectrum traces P_A(λ) and P_B(λ) respectively of the two samples p_A and p_B. As a consequence of the different polarization analysis conditions between the two samples p_A and p_B, the acquired traces P_A(λ) and P_B(λ) show different OSNRs. It is noted that in the special case where the OSNR is null on one of the acquired traces, i.e. the signal is completely suppressed, the following method is also applicable.

A spectrum processor 18 receives the two traces P_A(λ), P_B(λ) and discriminates the noise contribution n and the signal contribution s. As will be described hereinbelow, the discrimination is typically performed by subtracting the traces from one another to remove the noise contribution and provide a difference optical spectrum substantially proportional to the spectrum of the signal contribution and from which the optical spectrum of the signal S(λ), and thus the optical spectrum of the noise N(λ) can be estimated. It should be noted that a linear processing, such as filtering, linear transformation into another domain, etc., can be applied to the original traces P_A(λ), P_B(λ) before applying the herein presented processing. A noise calculator 20 evaluates the in-band noise N(λ_p) from the discriminated optical noise N(λ). The OSNR is then optionally calculated by an OSNR calculator 22 using the in-band noise N(λ_p) and the discriminated signal S(λ).

FIGS. 4A to 4D show different possible polarization optics arrangements 14 that can be used. In the example arrangement depicted in FIG. 4A, a polarization beam splitter 402 splits the input optical signal p into two orthogonal polarizations p_A and p_B. The two samples p_A and p_B consequently correspond to different polarization analysis conditions. Acquisition of the two samples can then be made using a dual channel OSA or an optical switch can be used in front of a single channel OSA for alternating between the two polarization analysis conditions. It is also noted that the polarization splitter can be placed before or after the monochromator of the OSA.

The example arrangement depicted FIG. 4B comprises a 50/50 optical coupler 404 that splits the input optical signal p onto two paths A and B. A polarizer 406 is inserted on path A to polarize the part of the input optical signal p propagating of path A in order to provide the first sample p_A, while the part of the input optical signal p propagating on path B is analyzed without applying a polarization.

The example arrangement depicted FIG. 4C comprises a polarizer 408 that is alternately inserted in the optical path of the input optical signal p to provide a first sample p_A, and removed from the path to provide the second sample p_B. The alternate insertion and removal of the polarizer 408 from the optical path is controlled by a control unit 410. In FIGS. 4B and 4C, the samples p_A and p_B correspond to different polarization analysis conditions.

The example arrangement depicted FIG. 4D comprises a polarization controller 412 placed before a polarizer 414. The polarization controller 412 is controlled by a control unit 416 which commands a variation of the polarization analysis conditions between the acquisition of sample p_A and the acquisition of sample p_B.

It is also noted that although two paths are shown on FIG. 3 for p_A and p_B at the output of the polarization optics arrangement 14, when the arrangement of FIG. 4C or 4D is used p_A and p_B are actually provided successively on the same optical path.

In any of the polarization optics arrangement of FIGS. 4A, 4B, 4C and 4D, the first and second samples p_A and p_B of the input optical signal p are in different polarization analysis conditions from one another, and show different OSNRs. The alignment of the polarization optics arrangement to the input optical signal p is arbitrary and it is not required that the signal contribution be substantially suppressed on either one of the samples p_A and p_B.

FIG. 5 illustrates a method for determining the in-band noise or the OSNR of the input optical signal p using the PPID approach. In step 502, the two samples p_A and p_B are produced from the input optical signal p using different polarization analysis conditions. The two polarization analysis conditions and thus the two samples p_A and p_B are typically produced by the polarization optics arrangement 14. In step 504, the optical spectrum traces P_A(λ) and P_B(λ), respectively, of the two samples p_A and p_B are acquired, typically using the OSA 11. In step 506, the noise N and signal S contributions are discriminated using the acquired traces P_A(λ) and P_B(λ), typically in the spectrum processor 18. One embodiment of this step is described in more detail below with reference to FIG. 7. In step 508, the in-band noise level N(λ_P) is determined from N. This step is performed, for example, by the in-band noise calculator 20. In step 510, the noise parameter, i.e. the in-band noise or the OSNR, is determined using the in-band noise level N(λ_p) and is typically output. The thereby determined noise parameter is output for use in monitoring, installation, commission, maintenance or troubleshooting of a DWDM optical system. For example, the noise parameter can be output using by graphical display, by printing, by generating an electrical signal or by storing it in memory for later retrieval. The in-band noise or OSNR can also be graphically or numerically output using a display unit or a printer, along with, for example, the individual and the sum of the acquired spectrum traces (P_A(λ), P_B(λ), P(λ)). Other parameters can also be displayed or otherwise output in a graphical or numerical form. The in-band noise level may also be output for optical signal processing or for determining the noise figure of an optical amplifier for example.

FIG. 6 shows one embodiment 10′ of the system 10 generally described with reference to FIG. 3. The embodiment 10′ uses a dual-channel polarization diverse OSA 12 (see, for example, the OSA described in U.S. Pat. No. 6,636,306) to implement the functions of both the polarization optics arrangement 14 and the OSA 11. An example of the PPID approach will be described in more detail below with reference to the system 10′ of FIG. 6 and to FIG. 7. The polarization diverse OSA 12 receives the input optical signal p to be characterized on an input optical fiber 38. The polarization diverse OSA 12 comprises a polarization splitter 34 that splits the input optical signal p into two typically orthogonally polarized samples p_A and p_B, and a dual channel OSA 31. The states of polarization are arbitrarily aligned with the input optical signal p such that each polarized samples p_A, p_B typically has a non-zero signal contribution s_A, s_B and a non-zero noise contribution n_A, n_g. The dual channel OSA 31 is used to simultaneously acquire the optical spectrum traces P_A(λ), P_B(λ) of two samples p_A, p_B. The dual channel OSA 31 comprises a dual channel grating-based monochromator 15 receiving both samples p_A, p_B at inputs A and B and separating their wavelength components which are detected using an optical detector 16a, 16b respectively for each channel A, B. An acquisition unit 17 records the optical spectrum traces P_A(λ) and P_B(λ) of the two samples p_A, p_B. It will be understood that the acquisition is conditioned by the spectral response of the filter of the OSA and that the acquisition step typically comprises applying calibration factors, signal conditioning and processing to each detected trace.

The PPID approach requires that the OSNR of the two samples p_A, p_B have a non-negligible difference. It may not be the case, for example, if the noise contribution n is substantially unpolarized while at the same time the signal contribution s is substantially polarized in a state of polarization oriented at 45° relative to the states of polarization of the polarization beam splitter 34. When this situation is detected, it can be circumvented by perturbing the input optical signal p to provide a small change in its polarization condition. The perturbation can be provided manually by moving the input optical fiber 38. A polarization perturbation device 36, such as a polarization controller, can be also used to vary the polarization condition of the input optical signal p. The optical spectrum traces P_A(λ), P_B(λ) corresponding to the two samples p_A, p_B are processed by a spectrum processor 18 which compares the acquired traces P_A(λ), P_B(λ) to discriminate the signal contribution S and the noise contribution N in the acquired optical signals. The discriminated signal S and noise contribution N are used by an OSNR calculator 30 to determine the OSNR. A second iteration may be performed if required as will be explained below. The system 10′ also has a display unit 32 for displaying the determined OSNR along with, for example, the individual and the sum of the acquired spectrum (P_A(λ), P_B(λ), P(λ)). Other parameters can also be displayed or otherwise output in a graphical or numerical form.

Example 1

One example embodiment of the processing performed by the spectrum processor 18 in order to discriminate the signal S and noise N contributions is now described with reference to FIG. 7.

In the following, the optical spectrum traces are analyzed and processed in the spectral neighborhood of each of the signal peaks, or DWDM optical channels, included in the input optical signal and for which a determination of the in-band noise level is desired. The signal peaks are identified by standard known implementation techniques such that their respective central wavelengths (V are determined. The in-band noise level and the OSNR on each signal peak can be estimated using the herein described method without requiring any further optical spectrum trace acquisition.

In step 502, two samples p_A and p_B are produced from the input optical signal p using different polarization analysis conditions as described above and typically using the polarization optics arrangement 34 of FIG. 6 in this case. In step 504, the optical spectrum traces P_A(λ) and P_B(λ) of the two samples P_A and P_B are acquired, typically using the dual-channel OSA 31 of FIG. 6.

Let the optical spectrum traces of the two samples P_A(λ) and P_B(λ) be respectively described as:

P _A(λ)=S_A(λ)+N_A(λ), and  (3)

P _B(λ)=S_B(λ)+N_B(λ),  (4)

where S_A and S_B are the signal contributions to the two acquired optical spectrum traces within the optical channel and where N_A and N_B are the noise contributions to the two acquired optical spectrum traces within the optical bandwidth of interest. The global optical power spectrum P(λ), which is representative of the optical signal p, is not necessarily acquired but can also be obtained by adding the acquired optical spectrum traces P_A(λ), P_B(λ):

P(λ)=P_A(λ)+P_B(λ),  (5)

and the polarization diverse OSA 12 is calibrated such that P(λ) corresponds to the optical spectrum of the input optical signal p as would be acquired by a non-diverse OSA if it was acquired before entering the system 10′.

According to the PPID approach, two optical spectrum traces P_A(λ), P_B(λ) are obtained using a polarization diverse OSA and are compared in order to discriminate the noise and signal contributions in the acquired traces P_A(λ), P_B(λ).

As in active polarization nulling methods, the signal contribution is assumed to be at least partially polarized and the noise contribution to be somewhat unpolarized. The PPID method then provides an evaluation of the noise and signal contributions. Within the optical bandwidth of interest:

N _B(λ)=β·N_A(λ),  (6)

S _B(λ)=k·S_A(λ),  (7)

where k is a constant which depends on the polarization alignment of the input optical signal p to polarization splitter 34 when the acquisition is performed, and where β is a constant which is typically 1 in all cases if the noise contribution N is unpolarized.

The signal and noise contributions are also in different proportions on the two traces P_A(λ), P_B(λ), i.e. the OSNRs are different. It is noted, however, that the OSNRs are initially unknown and that the signal and noise contributions need to be discriminated before they are known.

The following describes a solution that is derived in the case where N_A(λ)=N_B(λ) (or β=1), which is the case when the noise is unpolarized. It should however be noted that a special case arises when the polarization of the signal contribution s is substantially at 45° from the polarization splitter 34. In this special case, the difference between S_A and S_B is lower than the acquisition resolution of the OSA and k is equal to unity. In that specific case, the OSNR is equal on the two acquired optical spectrum traces P_A(λ) and P_B(λ), and the following method cannot be directly used. Accordingly, in step 706, in the special case where k=1 and β=1, a further acquisition with a varied input polarization condition (step 708) is performed. For example, in step 708, the input optical fiber 38 is manually moved, or the polarization perturbation device 36 is used. It is however noted that when k=1 and the noise is different on both traces, i.e. N_A(λ)≠N_B(λ) (as could be the case if the optical noise is not completely unpolarized), a PPID method can still be used without requiring the additional step of perturbing the input optical signal to change its polarization condition.

According to the assumptions described above, the contribution of the noise is canceled out in the subtraction of the two optical spectrum traces P_A(λ), P_B(λ) which yields:

P _A(λ)−P_B(λ)=S_A(λ)−S_B(λ)+(N_A(λ)−N_B(λ))=(1−k)·S_A(λ)  (8)

and

P _A(λ)+P_B(λ)=S_A(λ)+S_B(λ)+N_A(λ)+N_B(λ)=(1+k)·S_A(λ)+N(λ),  (9)

where N(λ)=N_A(λ)+N_B(λ), and (N_A(λ)−N_B(λ))=0.

From (8) and (9):

P _A(λ)+P_B(λ)=[(1+k)/(1−k)]·(P_A(λ)−P_B(λ))+N(λ)  (10)

S(λ)+N(λ)=K·(P_A(λ)−P_B(λ))+N(λ),  (11)

S(λ)=K·(P_A(λ)−P_B(λ)), where K=(1+k)/(1−k).  (12)

K is not known a priori but can be first estimated, in step 710, by assuming that K is a constant which does not vary in wavelength and that the noise level is low:

k _i =P _A(λ_p)/P_B(λ_p),  (13)

or letting:

K _i =[P _A(λ_p)+P_B(λ_p)]/[P_A(λ_p)−P_B(λ_p)].  (14)

This is based on the assumption that K=S_A(λ_p)/S_B(λ_p)=S_A/S_B which is true provided that the channel bandwidth CBW over which the analysis is performed is larger than the optical signal bandwidth or provided that no polarization mode dispersion affects the signal, which would thus lead to K(λ)=S_A(λ)/S_B(λ). In fact, K is an over estimation that can be related to K as follows:

K _i =K·(1+N(λ_p)/S(λ_p)).  (15)

In step 712, from this estimation K_i, a first estimation of the optical spectrum of the signal contribution S(λ), is obtained:

S(λ)_i=K_i·(P_A(λ)−P_B(λ))  (16)

In step 714, an estimation of the global optical power spectrum of the input optical signal is also obtained, namely:

P(λ)=P_A(λ)−P_B(λ).  (17)

In step 716, a first estimation of the noise contribution N(λ), is obtained using:

N(λ)_i=P(λ)−S(λ),  (18)

N(λ)_i=(P_A(λ)+P_B(λ))−K_i·(P_A(λ)−P_B(λ)).  (19)

An example of the global optical spectrum P(λ) of an input optical signal p is shown in FIG. 8 as well as a first noise estimation N(λ)₀ and a further noise estimation N(λ)₁ after one iteration. The other curves in FIG. 8 represent the acquired traces P_A(λ), P_B(λ) and the estimation of the signal contribution S(λ)₀. It can be seen that the error introduced by K₀ on the first noise estimation N(λ)₀ is large at the exact position of the signal peak, but becomes minimal on the edges of the signal peak, i.e. where the edges of the signal estimation meet the noise estimation. These crossing points c1 and c2 where S(λ)_o crosses N(λ)_o can be used to provide a first estimation of the in-band noise. An interpolation at λ_p of the estimated noise obtained at points c1 and c2, i.e. N(λ_c1)₀ and N(λ_c2)₀, provides an evaluation of the in-band noise N(λ_p)₀. At these crossing points, the error introduced by K₀ on the noise estimation is also 1+N(λ_p)/S(λ_p) which, for example, for a reasonable real life S(λ_p)/N(λ_p) of 100 (20 dB), contributes for less than 0.05 dB.

Referring back to FIG. 7, in step 718, the in-band noise N(λ_p)₀ is first evaluated using the noise estimation N(λ)_o at the edges of the signal peaks, i.e. at the crossing points c1, c2. In step 720, the OSNR₀ is calculated using the estimated in-band noise N(λ_p)₀ and the estimated signal S(λ)₀. This first estimation can be output at step 722 or iterations can be performed. Using the first estimation of the OSNR₀, a second estimation of K, i.e. K₁, is provided in step 710 from (15):

K ₁ =[P _A(λ_p)+P_B(λ_p)]/[P_A(λ_p)−P_B(λ_p)]·1/(1+1/OSNR₀).  (20)

In step 510, the noise parameter, i.e. the in-band noise, the OSNR or any other noise parameter, is determined using the determined in-band noise level N(λ_p) and is typically output.

Steps 712, 714, 716, 718, 720 and 510 are then repeated using a better estimation of K.

One skilled in the art will appreciate that the method used in step 718 to estimate the in-band noise can be varied. Referring to FIG. 8, in Example 1, the in-band noise is evaluated using an interpolation of the values of the estimated noise on points c1 and c2, i.e. N(λ_c1), and N(λ_c2)_i. In another embodiment, all points located in wavelengths in the neighborhood of c1 and c2 can be used in the interpolation.

In steps 712, 714 and 716 of Example 1, the signal contribution S(λ) is first isolated and the noise contribution N(λ) is then discriminated by subtracting the signal contribution S(λ) from the global optical spectrum P(λ). It is however noted that the calculations can be varied. For example, the noise contribution N(λ) can be isolated first and the signal contribution S(λ) discriminated by subtracting the noise contribution N(λ) from the global optical spectrum P(λ). For example, by assuming N_A(λ)=N_B(λ) in step 712, equation (16) can be replaced by:

N(λ)=2/(k_i−1)·[k_i·P_A(λ)−P_B(λ)]  (21)

It should further be appreciated that steps 712, 714 and 716 can be modified to show little sensitivity to unequal noise levels of the two acquired traces, i.e. when N_A(λ)≠N_B(λ). In the case of unequal noise, equations (8) and (9) can be rewritten in a more general form by letting:

N _B(λ)=β·N_A(λ), that is  (22)

P _A(λ)−P_B(λ)=S_A(λ)−S_B(λ)+N_A(λ)−N_B(λ)=(1−k)·S_A(λ)+(1−β)N_A(λ), and  (23)

P _A(λ)+P_B(λ)=S_A(λ)+S_B(λ)+N_A(λ)+N_B(λ)=(1+k)S_A(λ)+(1+β)N_A(λ).  (24)

In order to minimize the error introduced by the difference in noise level on P_A(λ) and P_B(λ), a multiplication factor γ can be chosen (for example by analysis of the subtraction to avoid discontinuities that can arise when negative values are obtained) and applied to P_B(λ) such that (1−β·β) is minimized to substantially 0. The difference thus obtained is proportional to the signal and we are back in a situation similar to the previously described condition where the noise was substantially unpolarized.

The OSNR of each signal channel of a DWDM system is obtained by independently performing the same calculations for each signal peak, using the acquired optical spectrum traces P_A(λ) and P_B(λ). No further acquisition is required.

As described above, the PPID approach reaches a limit in the special case where k is equal to unity, or more generally when k and β are equal within the acquisition resolution, therefore leading to identical signal-to-noise ratios in the two conditions. In this case, a further acquisition with a varied input polarization alignment of the input optical signal can be performed to circumvent this undesired condition. The input polarization alignment can be slightly changed by various simple means. For example, the input optical fiber 38 can be slightly moved, or a supplemental piece of hardware, such as the polarization controller 36, can be introduced in the system to change the polarization alignment of the input optical signal p with respect to the polarization analysis conditions in such cases.

In one embodiment, a digitally controlled polarization controller 36 is used in a system such as the one shown in FIG. 6. For each in-band noise determination to be performed, a first pair of optical spectrum traces P_A(λ), P_B(λ) is acquired. The polarization controller 36 is then used to automatically vary the input conditions, i.e. the polarization of the input optical signal, and a second pair of optical spectrum traces P_c(λ), P_o(λ) is then acquired. The spectrum processor 18 then selects which pair of traces to use, or which two traces among the four to use, for the noise and signal discrimination. The limit case where k and β are equal can then be avoided. The selection is made, for example, by selecting the two traces showing the largest difference in signal peak level from one another. Of course, more than two pairs of traces may be acquired or multiple analyses can be performed to improve the accuracy of the method.

In another embodiment, a single channel OSA is used along with the setup of FIG. 4D and multiple acquisitions (more than two) at different polarizer conditions are performed. The spectrum processor 18 then selects which traces to use for the noise and signal discrimination.

Another possibility is to use an alternative algorithm when the approach reaches the limit case where k and β are equal. An example of this is a hybrid approach as described further below in the section entitled “Hybrid Approaches”.

Differential Resolution Bandwidth Discrimination Approach (DRBD)

As discussed hereinabove, a further method of discriminating the optical noise contribution from the optical signal contribution in the input optical signal is to rely on the differences in their spectral properties, namely that the noise contribution n(λ) varies more slowly in wavelength than the signal contribution s(λ). The Differential Resolution Bandwidth Discrimination (DRBD) approach uses these differential spectral properties between the signal contribution s(λ) and the noise contribution n(λ) in the input optical signal to be analyzed.

FIG. 9 illustrates a method for determining the in-band noise, the OSNR or any other noise parameter to be determined on the input optical signal p using the DRBD approach in general. In step 902, at least two optical spectrum traces P₁(λ), P₂(λ) of the input optical signal p are obtained using different integration widths. In order to obtain the two traces, the RBW slit of the OSA can be used to adjust the RBW to two different values. The two traces are then directly acquired using the OSA. The two traces may also be obtained by numerical integration using sliding windows with different widths, which is the mathematical equivalent to varying the RBW slit of the OSA. In other words,

P ₁(λ)=h₁(λ)*h_OSA1(λ)*p(λ)=H₁(λ)*P(λ),  (25)

P ₂(λ)=h₂(λ)*h_OSA2(λ)*p(λ)=H₂(λ)*p(λ),  (26)

where h_i(λ), i=1 or 2, represents the numerical integration function and h_OSAi(λ) represents the OSA integration width caused by hardware, i.e. the RBW slit of the OSA. Either h_i(λ) or h_OSAi(λ) can be varied to obtain optical spectrum traces P₁(λ), P₂(λ) with different integration widths. The convolution window H_i(λ), resultant from the numerical integration function h_i(λ) combined with the OSA integration width h_OSAi(λ), is typically selected such that H₂(λ) is larger than H₁(λ) and H₂(λ) is narrower than the region where the noise contribution to be determined is present. H₂(λ) is thus typically narrower than the channel bandwidth. The resulting traces P₁(λ) and P₂(λ) have different levels of noise and signal contributions and, consequently, different signal-to-noise ratios:

P ₁(λ)=S₁(λ)N₁(λ)=H₁(λ)*s(λ)H₁(λ)*n(λ)=h₁(λ)*S(λ)+h₁(λ)*N(λ),  (27)

P ₂(λ)=S₂(λ)+N₂(λ)=H₂(λ)*s(λ)+H₂(λ)*n(λ)=h₂(λ)*S(λ)+h₂(λ)*N(λ).  (28)

As will be discussed in more details below, some assumptions or knowledge on the signal and noise contributions allows for their discrimination.

In step 904, the noise N and signal S contributions are mathematically discriminated using the optical spectrum traces P₁(λ) and P₂(λ). Embodiments of this step are described in more detail below with reference to FIGS. 10 to 12. In step 906, the in-band noise level N(λ_p) is determined from N. In step 508, the noise parameter, i.e. the in-band noise, the OSNR or any other noise parameter, is determined using the determined in-band noise level N(λ_p) and is typically output.

Example 2

One example embodiment of a method according to the DRBD approach is now described with reference FIGS. 10 and 11.

In this example, the numerical integration functions h_i(λ) and h₂(λ) are rectangular convolution windows of width RBW₁ and RBW₂ respectively. In step 1002, the two optical spectrum traces P₁(λ) and P₂(λ) are obtained by first acquiring an input optical spectrum trace:

P(λ)=h_OSA(λ)*p(λ)=S(λ)+N(λ),  (29)

that is then integrated with the convolution windows h₁(λ) and h₂(λ) to obtain P₁(λ) and P₂(λ) respectively (see equations (25) and (26) where h_OSA(λ)=h_OSA1(λ)=h_OSA2(λ) in this case). The two optical spectrum traces P₁(λ) and P₂(λ) are obtained with h₁(λ) being chosen such that the resulting signal peak in S₁(λ) is flat, as shown in FIG. 11, or the signal peak in S₁(λ) exhibits a constant slope, i.e. in the case of a non-flat noise, in the immediate region of the peak, i.e. over λ_p±δλ, where δλ is greater than the acquisition resolution. In other words:

S ₁(λ_p)=S₂(λ_p), or

h ₁(λ)*S(λ)|λ_p=h₂(λ)*S(λ)|λ_p,  (30)

where |λ_p denotes an evaluation of the preceding expression at λ_p.

In this example, RBW₂ is larger than RBW₁ and is chosen as the optical channel bandwidth, or the defined multiplexer bandwidth for a given network configuration, the multiplexer bandwidth typically corresponding to the bandwidth to which the noise is limited. It is noted that the precision of the in-noise level to be obtained is a function of the power level resolution of the acquired optical spectrum traces and of the number of data points in the bandwidth defined by RBW₂ RBW₁.

As can be seen on FIG. 11, when both RBW₁ and RBW₂ are selected larger than the width of the signal peak (the optical signal bandwidth), as the width of the convolution window is enlarged from RBW₁ to RBW₂ centered on the signal peak, the peak level (i.e. at λ_p) of the signal contribution does not substantially vary from S₁(λ) to S₂(λ) in the integrated optical spectrum traces P₁(λ) and P₂(λ). The variation of power level from P₁(λ_p) to P₂(λ_p) comes from the noise contribution which is integrated over a larger convolution window in P₂(λ) than in P₁(λ). This variation of the power level at λ_p can then be used to evaluate the noise contribution at λ_p, i.e. N(λ_p), with the assumption that the noise in the wavelength range between RBW₁ and RBW₂ is representative of the noise at and then applying the desired normalizing factor to account for the chosen integration widths and OSA filter spectral response. Accordingly, in step 1004, the in-band noise is estimated by subtracting the two optical spectrum traces evaluated at λ_p.

P ₂(λ_P)−P₁(λ_P)=[h₂(λ)−h₁(λ)]*N(λ)|λ_P,  (31)

the in-band noise corresponding to the noise level at the peak central wavelength λ_p.

In step 1006, equation (31) is solved to find the in-band noise at λ_P (N(λ_P)) using known mathematical techniques.

In step 1008, the noise parameter, i.e. the in-band noise, the OSNR or any other noise parameter, is determined using the determined in-band noise level N(λ_P) and is typically output.

This method provides for an adaptive noise estimation technique where h₂(λ) and h₁(λ) are optimally selected according to the input optical signal without requiring an experimented user's judgment.

One will understand that the shape of the convolution windows h₁(λ) and h₂(λ) can be varied. Gaussian shape windows, for example, can be used.

The method of FIG. 10 shows some limitations when the signal is spectrally large and the noise is substantially limited to the optical signal bandwidth, forcing the choice of an inappropriate RBW₁ and RBW₂ pair and leading to estimation errors in the noise. This, however, can be remedied when properties of the shapes of the signal contribution s(λ) or of the noise contribution n(λ) are known. FIG. 12 illustrates such a method allowing the determination of an in-band noise parameter when shape properties of the signal contribution s(λ) are known.

Example 3

When the signal peak is nearly as wide as the width of the noise contribution in the optical channel, the assumption made in the method of FIG. 10 that the signal peak power integrated over a wider bandwidth RBW₂ is equal to the signal peak power when integrated over RBW₁ becomes unsuitable. However, when the spectral shape of the signal contribution is known, it is no longer necessary to choose a first convolution window h₁(λ) that entirely contains all of the signal power. For example, the spectral shape of the signal contribution s(λ) can be found by recognizing the modulation spectrum of the signal, or relying on the polarization properties as described above, which would be an example of a hybrid approach as described below. The spectral shape of the signal contribution s(λ) can also be found by applying filtering techniques assuming that the noise contribution n(λ) spectrally varies slowly compared to the signal contribution s(λ) over the optical signal bandwidth. Spectrally faster components are then identified on the acquired optical spectrum trace and the signal contribution s(λ) is assumed to correspond to those.

Now referring to FIG. 12, in step 1204, the shape of the signal contribution s(λ) can be either previously known or can be found using various techniques. Assuming that the shape of the signal contribution s(λ) is known, the relation factor γ between s₂(λ_p) and s₁(λ_p) is also known. Let

γ=s₂(λ_p)/s₁(λ_p)  (32)

from equations (27) and (28),

P ₂(λ_p)−γP₁(λ_p)=N₂(λ)−γN₁(λ)|λ_p  (33)

P ₂(λ_p)−γP₁(λ_p)=H₂(λ)*n(λ)−γH₁(λ)*n(λ)|λ_p  (34)

P ₂(λ_p)−γP₁(λ_p)=[H₂(λ)−γH₁(λ)]*n(λ)|λ_p.  (35)

In step 1208, since H₂(λ), H₁(λ) and γ are known, the noise contribution n(λ) at λ_p can be solved for in equation (35) using known mathematical techniques. For example, if again rectangular convolution windows are chosen, step 1208 is equivalent to determining the noise within a region corresponding to H₂(λ_p)−H₁(λ_p), and assuming that this noise is representative of the noise contribution n(λ) at λ_p. The in-band noise n(λ_p) can be obtained.

In step 1210, the noise parameter, i.e. the in-band noise, the OSNR or any other noise parameter, is determined using the determined in-band noise level and is typically output.

It is noted that when determining, in step 1204, the shape of the signal contribution s(λ) using filtering techniques, it can be difficult to properly discriminate the noise contribution n(λ) from the signal contribution s(λ) when the signal contribution has a slowly varying spectral component that would be filtered when filtering the slowly varying noise contribution n(λ). However, proper selection of filters and evaluation in specific regions of the spectrum, for example on fast rising or falling edges, may allow determination of the shape of the signal contribution with sufficient precision to identify how the signal contribution should vary when applying different convolution windows.

It is noted that one skilled in the art will recognize that changes may be made to the DRBD approach described herein. For example, instead of performing numerical convolutions to provide two optical spectrum traces of the input optical signal, the RBW of the OSA can be varied using its variable RBW slit such that the optical spectrum traces acquired with these new parameters provide the two integration traces. It should still be noted that numerical convolutions are less sensitive to the calibration of the OSA.

Hybrid Approaches

The PPID approach and the DRBD approach each have some different advantages and drawbacks and can be combined in a hybrid approach which circumvents the limits of each approach. A proper combination of both approaches can be chosen for adapting the method to cases where the acquisition conditions limit the performance of one or the other.

For example, as described above, the PPID approach is limited when the signal-to-noise ratio is the same on the two acquired traces, e.g. when the signal level is the same on P_A(λ) and P_B(λ), and N_A(λ)=N_B(λ) (i.e. the input optical signal has a polarization state that is aligned at 45° with the polarization beam splitter 34 in the example of FIG. 6). The PPID approach can then be replaced by the DRBD approach.

Furthermore, it should be noted that the DRBD approach is particularly effective in cases where the OSNR is small (i.e. when the signal peak is significantly suppressed). One can benefit from this advantage to improve the PPID approach in cases where, on one of the acquired trace (P_A(λ) or P_B(λ)), the signal peak is substantially suppressed. This can be useful when, for example, the shape of the signal contribution contains artifacts introducing errors in the subtraction process. Applying the DRBD to such a trace provides a more accurate estimation of the noise level in that polarization analysis condition. The DRBD approach may also be used in an active polarization nulling method (see J. H. Lee et al., “OSNR Monitoring Technique Using Polarization-Nulling Method”, IEEE Photonics Technology Letters, Vol. 13, No. 1, January 2001) to reduce the requirements on the suppression of the signal. A measurement having a residual signal contribution may then be used to estimate the noise.

There are multiple ways to combine the two approaches in the manner that is most suitable for the condition of acquisition that is available. For example, it is noted that the PPID approach is dependent on the degree of polarization (DOP) of the noise and signal contributions. When the polarization mode dispersion or the polarization dependent loss of the system reaches a point where the PPID approach becomes inaccurate, the DRBD approach can be used as an alternative.

Example 4

One hybrid approach that combines both the PPID and DRBD approaches in order to improve their performances is now described with reference to FIG. 13.

In step 1302, two samples p_A and p_B are produced from the input optical signal p using different polarization analysis conditions as described above according to the PPID approach. The two samples p_A and p_B are typically obtained using a polarization optics arrangement such as the polarization beam splitter 34 of FIG. 6.

In step 1304, the optical spectrum traces P_A(λ) and P_B(λ) of the two samples p_A and p_B are acquired, typically using the dual-channel OSA 31 of FIG. 6. As mentioned previously:

P _A(λ)=S_A(λ)+N_A(λ), and  (36)

P _B(λ)=S_B(λ)+N_B(λ).  (37)

In step 1306, two different integration widths, RBW₁ and RBW₂ (RBW₂ larger than RBW₁), are selected such that the signal power is substantially contained within RBW₁, and RBW₂ is still narrower than the spectral width of the noise contribution. The acquired traces P_A(λ) and P_B(λ) are then numerically integrated by performing a convolution respectively with a rectangular window h₁(λ) and h₂(λ) of width RBW₁ and RBW₂ respectively. This yields the following results:

P _A1(λ)=P_A(λ)*h_i(λ)=S_A1(λ)+N_A1(λ),  (38)

P _B1(λ)=P_B(λ)*h₁(λ)=S_B1(λ)+N_B1(λ),  (39)

P _A2(λ)=P_A(λ)*h₂(λ)=S_A2(λ)+N_A2(λ),  (40)

P _B2(λ)=P_B(λ)*h₂(λ)=S_B2(λ)+N_B2(λ).  (41)

In step 1308, the noise and signal contributions are discriminated by applying PPID method on the optical spectrum traces P_A1(λ), P_B1(λ), P_A2(λ) and P_B2(λ). Let:

K ₁ =[P _A1(λ_p)+P_B1(λ_p)]/[P_A1(λ_p)−P_B1(λ_p)],  (42)

we have:

[P_A2(λ_p)+P_B2(λ_p)]−K₁ [P_A2(λ_p)−P_B2(λ_p)]=[S_A2(λ)+N_A2(λ)+S_B2(λ)+N_B2(λ)]−K₁·[S_A2(λ)+N_A2(λ)−S_B2(λ)−N_B2(λ)].  (43)

Then, inserting K₁ from equation (42) into (43), noting that S_B1(λ)=S_B2(λ) and S_A1(λ)=S_A2(λ) as a consequence of the choice of RBW₁ and RBW₂ in step 1306 above, and assuming that N_A1(λ)=N_B1(λ) and N_A2(λ)=N_B2(λ), which is the case when the noise contribution is essentially depolarized, then

$\begin{array}{cc}\begin{array}{c}\left[P_{A2}\left({\lambda }_p\right)+P_{B2}\left({\lambda }_p\right)\right]-K_1·\left[P_{A2}\left({\lambda }_p\right)-P_{B2}\left({\lambda }_p\right)\right]=N_{A2}\left(\lambda \right)+N_{B2}\left(\lambda \right)-\\ \left[N_{A1}\left(\lambda \right)+N_{B1}\left(\lambda \right)\right]\\ =N\left(\Delta RBW\right)\end{array}& \left(44\right)\end{array}$

where N(ΔRBW) is the noise contribution integrated in the spectral region corresponding to RBW₂−RBW₁. N(ΔRBW) therefore corresponds to the noise that would be measured by an OSA having a resolution bandwidth equal to the difference of integration widths, i.e. RBW₂−RBW₁.

Other integration widths RBW₁ and RBW₂ can then be used by going back to step 1306, in order to estimate the noise contribution in other resolution bandwidths RBW₂−RBW₁, or simply to obtain a more accurate measurement of the noise contribution in the same resolution bandwidth RBW₂−RBW₁. Adaptively, other iterations can be performed until it is determined that RBW₁ and RBW₂ were properly selected, i.e. RBW₁ and RBW₂ are wider than most of the signal power, but still narrower than the noise contribution width.

In step 1310, the in-band noise level N(λ_P) is estimated from N(ΔRBW) obtained in step 1308, assuming that the noise contribution integrated in the spectral region corresponding to RBW₂−RBW₁ is representative of the noise contribution at the signal peak Δ_p.

Finally, in step 1312, the noise parameter, i.e. the in-band noise, the OSNR or any other noise parameter, is determined using the estimated in-band noise level N(λ_P) and is typically output.

The embodiments of the invention described above are intended to be exemplary only. The scope of the invention is therefore intended to be limited solely by the scope of the appended claims.

Claims

1. (canceled)

2. (canceled)

3. (canceled)

4. (canceled)

5. (canceled)

6. (canceled)

7. (canceled)

8. (canceled)

9. (canceled)

10. (canceled)

11. (canceled)

12. (canceled)

13. (canceled)

14. (canceled)

15. (canceled)

16. (canceled)

17. (canceled)

18. (canceled)

19. (canceled)

20. (canceled)

21. (canceled)

22. (canceled)

23. (canceled)

24. (canceled)

25. (canceled)

26. (canceled)

27. (canceled)

28. (canceled)

29. (canceled)

30. (canceled)

31. (canceled)

32. (canceled)

33. (canceled)

34. (canceled)

35. (canceled)

36. (canceled)

37. (canceled)

38. (canceled)

39. (canceled)

40. (canceled)

41. (canceled)

42. (canceled)

43. (canceled)

44. (canceled)

45. (canceled)

46. A method for determining a noise parameter on an input optical signal having a data-carrying signal contribution and a noise contribution within an optical signal bandwidth, the method comprising: obtaining at least two optical spectrum traces from said input optical signal, said optical spectrum traces being taken under different conditions such that they show different non-zero signal-to-noise ratios;mathematically discriminating said signal contribution from said noise contribution within said optical signal bandwidth based on a mathematical combination of said optical spectrum traces;determining an in-band noise level on said input optical signal from the discriminated noise contribution; anddetermining the noise parameter from the determined in-band noise level, the noise parameter being indicative of the noise contribution within the optical signal bandwidth.

47. The method as claimed in claim 46, wherein the signal contribution and the noise contribution have at least one of different degrees of polarization and different states of polarization from one another and wherein said obtaining at least two optical spectrum traces comprises:producing at least two samples of the input optical signal, under different polarization analysis conditions, states of polarization of the two samples being arbitrary relative to said input optical signal; andacquiring an optical spectrum of each one of the first and the second samples to obtain said optical spectrum traces.

48. The method as claimed in claim 47, wherein said polarizing at least part of the input optical signal comprises polarization beam splitting said input optical signal into said samples, said samples having states of polarization orthogonal from one another.

49. The method as claimed in claim 48, wherein said discriminating comprises: subtracting said optical spectrum traces from one another to obtain a difference optical spectrum substantially proportional to an optical spectrum of said signal contribution;estimating an optical spectrum of said input optical signal using said difference optical spectrum;determining an optical spectrum of said input optical signal from at least one of the first and second optical spectrum traces; anddetermining a level of said optical noise by subtracting the estimated optical spectrum of said signal contribution for the determined optical spectrum of said input optical signal.

50. The method as claimed in claim 46, wherein said noise contribution within said optical bandwidth varies slowly in wavelength compared to said data-carrying signal contribution, and optical spectrum traces are obtained using different integration widths to provide said different conditions, a second one of the integration widths being larger than a first one of the integration widths.

51. The method as claimed in claim 50, wherein said obtaining comprises: acquiring an optical spectrum measurement of said input optical signal;convoluting the acquired measurement with a first convolution window having a width corresponding to the first integration width to obtain a first optical spectrum trace, and convoluting the acquired measurement with a second convolution window having a width corresponding to the second integration width to obtain the second optical spectrum trace.

52. The method as claimed in claim 51, wherein said discriminating comprises subtracting said optical spectrum traces from one another to obtain a difference trace, a value of said difference trace at a central wavelength of said optical signal being representative of said optical noise.

53. The method as claimed in claim 46, wherein said noise parameter comprises an optical signal to noise ratio of the input optical signal, determined using the determined in-band noise.

54. The method as claimed in claim 46, further comprising outputting the determined noise parameter.

55. The method as claimed in claim 49, wherein said optical spectrum traces each have a signal contribution and a noise contribution and wherein said performing calculations further comprises estimating a factor K related to a proportion between the signal contributions of said optical spectrum traces for use in said estimating an optical spectrum of said input optical signal.

56. The method as claimed in claim 55, wherein said calculating further comprises: re-estimating said factor K considering the evaluated in-band noise; andcalculating the optical spectrum of said optical noise using the re-estimated factor K.

57. The method as claimed in claim 56, wherein said acquiring comprises: if said optical spectrum traces show equal signal to noise ratios, disturbing a state of polarization of said input optical signal to vary the signal to noise ratio on at least one of said optical spectrum traces.

58. The method as claimed in claim 57, wherein said noise parameter comprises an optical signal to noise ratio of the input optical signal, determined using the determined in-band noise level.

59. The method as claimed in claim 47, further comprising: convoluting said first optical spectrum trace with a first convolution window and a second convolution window to respectively obtain a first and a second integrated traces, said first convolution window corresponding to a first integration width and said second convolution window corresponding to a second integration width larger than said first integration width;convoluting said second optical spectrum trace with said first convolution window and said second convolution window to respectively obtain a third and a fourth integrated traces; andwherein said discriminating is performed using said first, second, third and fourth integrated traces.

60. The method as claimed in claim 51, wherein a spectral shape of one of said signal contribution and said noise contribution is known within said optical signal bandwidth, said discriminating is performed by assuming said spectral shape.

61. The method as claimed in claim 51, wherein a spectral shape of said signal contribution is known within said optical signal bandwidth, said discriminating comprises: determining a factor related to a ratio between the signal contributions of said second to said first optical spectrum trace;multiplying said first optical spectrum trace with said factor; andsubtracting the multiplied optical spectrum trace from said second optical spectrum trace to obtain a difference trace indicative of said noise contribution.

62. A system for determining an in-band noise on an input optical signal within an optical signal bandwidth, the system comprising: an input for receiving an input optical signal comprising a data-carrying signal contribution and a noise contribution within said optical signal bandwidth, said signal contribution and said noise contribution having at least one of different degrees of polarization and different states of polarization from one another;a polarization optics arrangement for obtaining a first and a second sample of the input optical signal such that the first and second samples have at least one of different states of polarization relative to the input optical signal and different degrees of polarization from one another, states of polarization of the samples being arbitrary relative to said input optical signal;an optical spectrum analyzer for acquiring a first and a second optical spectrum traces respectively of the first and second samples, the first and second optical spectrum traces showing non-zero signal to noise ratios different from one another;a spectrum processor adapted for mathematically discriminating said noise contribution in said input optical signal within said optical signal bandwidth from said optical spectrum traces using said first and said second optical spectrum traces; anda noise calculator for evaluating said in-band noise within the optical signal bandwidth from the discriminated noise contribution.

63. The system as claimed in claim 62, wherein said spectrum processor comprises a differentiator for calculating, from said optical spectrum traces, a difference optical spectrum substantially indicative of said signal contribution; and a noise solver for solving for said noise contribution using calculations involving said optical spectrum traces and said difference optical spectrum.

64. The system as claimed in claim 62, wherein said polarization optics arrangement comprise a polarization beam splitter for splitting said input optical signal into said first and second samples, said samples having orthogonal states of polarization from one another.

65. The system as claimed in claim 62, further comprising a polarization disturbing device for disturbing a state of polarization of said input optical signal to vary the signal to noise ratio on at least one of said optical spectrum traces such that the first and second optical spectrum traces show different signal to noise ratios.