Patent Application
20090214204
Optically transparent mesh networks are large, complex networks having many spans of optical fiber and may use dense wavelength division multiplexing (DWDM). Signal power levels in the optical channels in these networks are adjusted on a periodic basis by sophisticated control schemes that attempt to meet pre-defined power targets specified by a set of design and engineering rules. Signal and channel are used interchangeably throughout the disclosure. These control schemes also compensate for power disturbances due to environmental changes, adding and/or dropping of channels in the network, aging of various optical components, and other factors beyond a service provider's control.
Optical networks may be designed to ensure that the channel power levels fall within certain desirable ranges as the channels propagate through the network. Typically, each channel in the system will have a target power. This is to minimize effects such as noise, which will dominate if the channel power level is too small, and nonlinearities, such as four-wave mixing, if the channel power level is too high. Optical network design thus attempts to minimize noise and nonlinearity effects by specifying target power levels (per-channel and/or over the fiber) at specific physical locations, for example, the output of an amplifier node in the optical network.
However, although the self-compensating nature of the control algorithms ensures that they will strive to reach desired network performance levels, they can also mask network impairments due to, for example, hardware failures, software bugs, incorrectly installed hardware, and hardware-specific provisioning parameters used in control algorithms, e.g. fiber type, fiber length, etc. These impairments can be masked until a system margin is exhausted, at which time catastrophic failure may occur. Due to the distributed nature of the network, the failure may occur at a point in the network separated by many network elements (and a large physical distance) from the location of the impairment. Therefore, a service provider may incur significant operational costs for network outages and in localizing the fault for repair.
The health of the optical network may be affected by factors including e.g.: i) the signals propagating on the network; and ii) the underlying hardware and software components comprising the network. Existing network management solutions focus primarily on the former, and hence are expected to be inadequate at diagnosing faults associated with the underlying hardware.
Adjustments to optical amplifiers in the system may be performed to ensure the measured optical channel powers match the targeted channel powers. In this manner, a network attempts to self-correct potential errors which arise as channel powers deviate from target. To maintain this stability, adjustments are made continuously (over time) to the amplifiers within the system. In the case of a Raman amplifier, the pump powers of a set of optical pump lasers need to be changed to maintain optimal channel powers. Generally no targets are placed on the pump laser values, they are adjusted based only on optimal system performance, although minimum and maximum power levels may be specified for reliability and safety considerations.
Example embodiments seek to monitor optical networks, for example methods to estimate when a network is masking impairments and to determine the location and nature of these impairments, thus leading to robust network operation and reduced costs for network operation and deployment. Example embodiments may also have application to initial system development and debugging.
Example embodiments include a method for automatic monitoring at a monitoring station. The method includes receiving at the monitoring station, at least measured optical network data for a span of the optical network and determining whether the received optical network data is consistent with modeled optical network data determined from at least one optical network model. The optical network model models the physics of amplifiers in at least one span of the optical network. The optical network model may also model Raman amplifiers. The model may include at least one of a physics-based statistical model, referred to as an “R-beta model” and a detailed-physical model.
When an inconsistency is determined, an alarm may be activated, including alarms of an audible, visual, tactile, and/or any combination thereof type. The modeled data may also be presented to illustrate whether there is an inconsistency or not, by display, hard-copy, graph, chart, verbally, and/or any combination thereof. Example embodiments may be performed repeatedly at various intervals, e.g. continuously, every minute, hourly, daily, periodically etc.
Example embodiments will be more clearly understood from the following detailed description taken in conjunction with the accompanying drawings.
Various example embodiments will now be described more fully with reference to the accompanying drawings in which some example embodiments are illustrated.
Detailed illustrative embodiments are disclosed herein. However, specific structural and functional details disclosed herein are merely representative for purposes of describing example embodiments. This invention may also be embodied in many alternate forms and should not be construed as limited to only example embodiments set forth herein.
Accordingly, while example embodiments are capable of various modifications and alternative forms, embodiments thereof are shown by way of example in the drawings and will herein be described in detail. It should be understood, however, that there is no intent to limit example embodiments to the particular forms disclosed, but on the contrary, example embodiments are to cover all modifications, equivalents, and alternatives falling within the scope of the invention. Like numbers refer to like elements throughout the description of the figures.
It will be understood that, although the terms first, second, etc. may be used herein to describe various elements, these elements should not be limited by these terms. These terms are only used to distinguish one element from another. For example, a first element could be termed a second element, and, similarly, a second element could be termed a first element, without departing from the scope of example embodiments. As used herein, the term “and/or” includes any and all combinations of one or more of the associated listed items.
The terminology used herein is for the purpose of describing particular embodiments only and is not intended to be limiting of example embodiments. As used herein, the singular forms “a,” “an” and “the” are intended to include the plural forms as well, unless the context clearly indicates otherwise. It will be further understood that the terms “comprises,” “comprising,” “includes” and/or “including,” when used herein, specify the presence of stated features, integers, steps, operations, elements and/or components, but do not preclude the presence or addition of one or more other features, integers, steps, operations, elements, components and/or groups thereof.
It should also be noted that in some alternative implementations, the functions/acts noted may occur out of the order noted in the figures. For example, two figures shown in succession may in fact be executed substantially concurrently or may sometimes be executed in the reverse order, depending upon the functionality/acts involved.
Unless otherwise defined, all terms (including technical and scientific terms) used herein have the same meaning as commonly understood by one of ordinary skill in the art to which example embodiments belong. It will be further understood that terms, e.g. those defined in commonly used dictionaries, should be interpreted as having a meaning that is consistent with their meaning in the context of the relevant art and will not be interpreted in an idealized or overly formal sense unless expressly so defined herein.
It should be noted that example embodiments may be performed by a computer if the method is encoded on a computer readable medium. Such computer readable mediums, may include CDs, DVDs, memory devices, ROM, RAM, Flash drives, etc. In the following description, illustrative embodiments will be described with reference to acts and symbolic representations of operations (e.g., in the form of flowcharts) that may be implemented as program modules or functional processes including routines, programs, objects, components, data structures, etc., that perform particular tasks or implement particular abstract data types and may be implemented using existing hardware at existing network elements or control nodes (e.g., a monitoring station at a network node or at a control center that is outside the network but may access the network nodes remotely). Such existing hardware may include one or more digital signal processors (DSPs), application-specific-integrated-circuits, field programmable gate arrays (FPGAs), computers, etc.
Example embodiments will now be described more fully with reference to the accompanying drawings, in which exemplary embodiments of the invention are shown. The example embodiments may, however, be embodied in many different forms and should not be construed as being limited to the embodiments set forth herein. Rather, these embodiments are provided so that this disclosure will be thorough and complete, and will fully convey the scope of the invention to one of ordinary skill in the art. In the drawings, the sizes of constitutional elements may be exaggerated for convenience of illustration.
The example embodiments assess if the signal gain across a fiber span is consistent with the amplification level of the pumps being input to the fiber span. Specifically, models of the Raman gain physics are combined with statistical estimation, which result in methods and apparatuses that allow detection of anomalous conditions. The method focuses on detecting the network behavior that is inconsistent with Raman physics, and hence may capture more complex anomalies not readily recognized by an engineering-rule-based approach.
For Raman and other amplifier types, the quantity (output-input-attenuation) and the shape of the amplifier gain, is dictated by the underlying amplifier physics. Gain may be defined as the ratio between received and transmitted channel power at specific points along a fiber. The gain may also be determined along a span, node, section, etc. If the gain profile or gain shape is inconsistent with expected system behavior, it is an indication that something in the system is affecting the measured gain shape. These changes are termed anomalies and/or inconsistencies, since they indicate in some fashion the system is behaving in an unexpected manner. These anomalies may not indicate a failure to transmit information, but they do indicate a possible future problem in the network.
Two primary mathematical models of the amplifier physics are described below: (1) a physics-based statistical model, referred to as the R-Beta model, which approximates the fiber span behavior as comprising two dominant physical effects, namely fiber attenuation and pump-signal Raman amplification; and (2) a detailed physical model, which may be a function of at least one of input channel power, input pump power, fiber type (e.g. based on manufacturer and fiber characteristics), attenuation loss, input connector losses, output connector losses, Rayleigh back-scattering, and temperature (e.g. fiber temperature).
The R-Beta model is a statistical model that accounts for pump-signal amplification by modeling the Raman gain via the path-total power over a determined span. The path total power is the integral and/or sum of pump powers in certain frequency ranges over a determined fiber length. This model may be used to detect gain deviations that are not consistent with Raman physics. Examples of such deviations are those due to power measurement errors, mis-fibering (e.g., connecting the wrong fiber to a device), various network monitor failures, and power measurement calibration errors, e.g. optical monitor (OMON) calibration errors.
The detailed physical model represents a description of various physical phenomena occurring in a Raman amplified span, e.g. a span in which the amplification takes place throughout the span. The detailed physical model may include any or all effects which contribute in a significant manner to the expected gain profile for a typical Raman amplified span of the network, e.g. both stimulated and spontaneous Raman interactions, fiber attenuation, Rayleigh back-scattering effects and connector loss effects, etc. Through simulation and direct comparison, this model may detect anomalous behavior within the optical network. For different fibers or expected optical power levels, the detailed model may be augmented with additional terms (e.g. four-wave mixing, stimulated Brillouin scattering, etc.) which may be required to obtain a quantitative agreement between simulation and measurement.
In the detailed-physical model, input powers are used to compute an expected gain profile. If the expected gain profile differs from the measured gain profile, there is the possibility of an anomaly. However, there are many effects which may cause the computed or modeled gain profile to differ from a determined gain profile. Therefore, to match the model to the determined, adjustments to one or more input parameters may be required. To the extent that adjustments are necessary, and require adjustments to parameters that may result in impossible or unphysical values, the detailed-physical model may indicate a potential anomaly.
With both models, the expected behavior of various failure scenarios can be established using large-scale simulations and corresponding alarms, the alarms being used to indicate network anomalies/inconsistencies and/or the type of anomaly/inconsistency. The proposed alarms may be used in combination with traditional rule-based approaches and data visualization of the optical network to provide a diagnostic tool for optical networks that use Raman amplifiers.
As shown, the monitoring station 100 receives network data in step S400. The received network data may be obtained using data queries sent out to the optical network and retrieving and/or receiving the requested optical data in response. The optical data may include both measured and observed data as well as network provisioned value for various parameters. The optical data may include, for example, input and output channel powers, channel wavelengths, channel frequencies, pump frequencies, pump wavelengths, input and output pump powers, and provisioned values of physical properties for a fiber span, including e.g. fiber type, fiber length, total span losses, and connector losses. The pump wavelengths/frequencies may be provisioned by the network and the pump powers may adjust over time as required by the network.
Example embodiments may include a charting tool (not shown) that produces a set of inference-assisting graphical summaries of the optical network state at the monitoring station 100. These displays are designed to highlight the data distributions from both spatial and spectral views of the network, as well as to show various indicators of the system performance and health.
The monitoring station 100 applies the received data to at least one statistical model, e.g. the R-Beta model in step S410 and/or the detailed-physical model in step S420. The R-Beta model and the detailed-physical model will be further described below.
The monitoring station 100 then compares the determined gain profile to the modeled gain profile e.g. using linear regression techniques at step S560. The flow then returns to
Alternatively, or additionally, as shown in
As shown, the collected optical network measurements are received from step S400. At step S570, the detailed-physical model determines the difference between received output data and modeled output data. The received data may include measured network data, e.g. input channel power, output channel power, input pump power, output pump power, etc., and provisioned data, e.g. connector losses, fiber attenuation, Raman gain coefficient, etc. For example, received values for input channel power and/or input and output pump powers and various provisioned parameters may be used by the detailed-physical model to predict a set of modeled output gain profiles, similar to the manner in which the R-Beta model functions. Other possible output data may include modeled output power. As indicated above, the detailed-physical model considers received data, as well as, e.g. pump powers and various provisioned parameters, e.g. connector loss, etc.
Alternatively or additionally, at step S580 the detailed-physical model may estimate provisioned parameter values. For example, the detailed-physical model may use measured input and output channel powers and input pump powers, to estimate, e.g. one of the provisioned parameters, e.g. connector losses, Raman gain coefficient, fiber attenuation, etc.
The flow from both steps S570 and/or S580 then returns to
First, the determination may be based on a determination of how well the determined gain profile fits the modeled possible set of gain profiles that are consistent with Raman physics. The determination is then compared to a threshold that may be empirically determined and if the determination is outside the threshold, then an inconsistency is determined to exist in the optical network. These inconsistencies may include measurement errors, e.g. OMON errors, mis-fibering, incorrectly calibrated pump values, software bugs, etc.
Second, the determination may be based on a determination of the difference between the provisioned parameters having a nominal value and the modeled parameters. The determination is then compared to a threshold that may be empirically determined and if the determination is outside the threshold, then an inconsistency is determined to exist in the optical network. These inconsistencies may include anomalous or incorrectly provisioned physical property information, e.g. fiber attenuation, connector losses, Raman gain coefficient.
If an inconsistency is determined at step S460, then an alarm is activated and the received and/or determined data may be presented to a user at step S480. If no inconsistency is determined at step S460, then the received data and/or the determined data may be presented to a user at step S470. The example embodiment methods may be preformed repeatedly at various intervals, e.g. continuously, every minute, hourly, daily, or at some predetermined time interval etc.
As shown in
Presenting the data may include using any well-known system-level analysis and well-known presentation tools that perform additional system-level diagnostics and combine the alarms reported on each polled span and the corresponding diagnostic charts into a hierarchical presentation. The system-level analysis tool provides correlation of alarms and root-cause analysis capability via easy to read display summaries of network alarms, including those generated by the model-based approaches and traditional engineering-rule based approaches.
While both the R-Beta model and the detailed-physical model may be used in parallel as shown in
Example embodiments of the R-Beta model and the detailed-physical models are further described below.
The following equation approximates the evolution of the channel power p (z, λi) at channel wavelength λi as a function of distance z along the fiber in a Raman-amplified fiber span
In (1); p(z,λj) denotes the pump power at pump at spatial location z and wavelength λj, α(λi) is the fiber attenuation coefficient at wavelength λi, R(λi,λj) is the Raman gain coefficient between the pump at wavelength λj and the signal at wavelength λi, and Npumps is the number of pumps. The relationship in (1) can be alternatively expressed by dividing by p (z,λi) and integrating both sides from z=0 to L, the fiber length, resulting in the following equation
where βj=∫0Lp(z,λj)dz represents the pump power integrated over the length of the fiber span. Denote the left hand side of (2), the attenuation-adjusted gain for channel i, by
δi≡ln p(L,λi)−ln p(0,λi)+α(λi)L, (3)
and let Δ=[δ1, . . . δ . . . ]T contain k adjusted gains. Further, with R being a k×Npumps matrix with elements Rij≡R(λi,λj), (2) can be expressed as
Δ≈Rβ (4)
Linear regression techniques can be used to fit a set of observed signal gains Y to (4), i.e.
Y=Rβ+ε (5)
Here, the quantity ε measures the gain across the Raman-amplified span that is not captured by the Rβ model in (4). The magnitude of this term, ∥ε∥ is a measure of how well the actual measured data is described by the R-Beta amplification model, independent of particular pump levels. The residual ε may include measurement noise, measurement errors, physical effects left out by the R-Beta approximation, and the effects of any potential anomalies.
The detailed-physical model may be a more comprehensive model for Raman-amplification, which modifies the R-Beta model. The detailed-physical model may include terms that account for signal-signal Raman pumping, noise generation and Rayleigh back-scattering in the Raman span as well as effects such as connector and splice losses. The detailed-physical model may be represented via the following equation:
y
s(L)=f(ys(0),Pf(0),Pb(L),α(ν),γ(ν),g(v,ζ)),Δs,ΔR), (6)
which expresses the received signal powers yS(L) as a function ƒ(·) of: the launch signal powers yS(0), the launch powers of the co- and counter-propagating Raman pumps Pƒ(0) and Pb(L), the fiber attenuation coefficient α(ν) at frequency ν, Rayleigh back-scattering coefficient γ(ν), Raman gain coefficient g(ν,) for a pair of frequencies ν and ζ, and the connector and splice losses ΔS and ΔR at the launch (z=0) and receive locations (z=L).
Some of the key physical parameters controlling the Raman span behavior are the attenuation coefficient α(ν), the Raman gain coefficient g(ν,), the fiber length L, and the connector losses ΔS, ΔR. Values for L, α(ν) and total connector loss (ΔS+ΔR) can be estimated from OTDR (Optical Time Domain Reflectometry) and total span loss measurements, while nominal values for the Raman gain coefficient g(ν,
) may be typically used.
The detailed physical model may be used to produce inconsistency alarms, for example, in the following ways: (i) measured signal powers at z=0 and pump powers at both z=0 and z=L are used to predict signal powers at z=L, and the predicted values are then compared to measured signal powers at z=L; and/or (ii) a maximum likelihood procedure is used to estimate the connector losses (ΔS,ΔR), which are compared to provisioned values. Discrepancies observed in either (i) or (ii) indicate inconsistencies.
Example embodiments describe a model-based approach for anomaly and/or inconsistency detection and monitoring in optical networks. Example embodiments model Raman amplifiers, but the methodology is readily extendable to other amplification techniques, such as Erbium-Doped Fiber Amplifier (EDFA) based devices, blockers and non-pumped DCMs (Dispersion compensating modules). As will be appreciated from the disclosure, these techniques may also be extended to examining other measures of performance, such as optical signal-to-noise ratio, bit error rate and dispersion, etc.
Example embodiments of the present invention being thus described, it will be obvious that the same may be varied in many ways. Such variations are not to be regarded as a departure from the invention, and all such modifications are intended to be included within the scope of the invention.
