TECHNICAL FIELD

This invention relates to provisioning lightpaths in an optical transmission network.

BACKGROUND

A Wavelength Switched Optical Network (WSON) supports end-to-end optical paths, called lightpaths, between nodes requiring connection in the network. Intermediate nodes in this type of network support wavelength switching and may also support wavelength conversion. WSONs can be deployed across large areas and optical paths can be routed, if possible, with minimal electrical regeneration.

Traffic recovery schemes can be implemented in the network. As an example, a traffic recovery scheme may allow traffic to be re-routed to a different path when a single or double failure occurs in part of the network, such as failure of a link or a node. Advantageously, the network should perform traffic recovery with minimal service interruption.

One issue in this type of network is the relatively long time required to set-up a lightpath. Establishing a lightpath requires operations such as: Wavelength Selective Switch (WSS) levelling, optical amplifier power setting and optimal setting of adaptive compensators for transmission impairments such as Chromatic Dispersion (CD). This can cause a delay when first establishing a lightpath, and is a particular problem during traffic recovery operations. When a fault occurs in the network, it is desirable that traffic is transferred to an alternative path as quickly as possible. Delay in setting up the alternative path can result in heavy loss of traffic.

In the case of Chromatic Dispersion, a sudden path re-rerouting can correspond to a variation of accumulated CD of the order of hundreds of ps/nm (dispersion managed systems) or tenths of ns/nm (uncompensated systems).

Examples of adaptive compensators for CD in the optical domain are tunable chromatic dispersion compensators (TCDC) that are put on-board of 40Gbit/s DPSK or DQPSK line interfaces equipped on dispersion managed systems. These components are used to minimize the BER by applying a chromatic dispersion of the order of ±1000 ps/nm to the received optical signal. Examples of adaptive compensators for CD in the electrical domain are digital filters that are implemented in 40Gbit/s or lOOGbit/s line interfaces employing coherent detection and digital signal processing (DSP) equipped on dispersion managed or unmanaged systems. These solutions minimize the BER by substantially inverting the fibre transfer function and are able to compensate for a chromatic dispersion of the order of ±50 ns/nm.

TCDC is a widely deployed solution but it is intrinsically slow because it is based on a thermo -mechanical process. The process scans all of the possible values of CD to find a range of values for which the signal is at least detectable. Eventually, the optimal operating point is identified by means of a finer search within this range. The total operation (coarse and fine search) usually takes minutes.

Coherent detection paired with DSP is extremely attractive for its ability to counteract linear transmission impairments like CD and PMD, amongst other reasons. However, the circuit that implements the CD compensator is usually split in two parts: one performs "bulk" and slow compensation and the other performs "fine" and very fast compensation. The amount of compensated CD is indicatively ±50ns/nm and ±lns/nm for the "bulk" and "fine" compensators. The "bulk" compensator may take up to 50 ms only to scan all the possible values to find the working range.

In both cases, the problem of speeding up the setting of CD compensators cannot be solved only by means of planning methods for several reasons: uncertainties of the data supplied by the cable owner (fibre length, fibre type, fibre dispersion), tolerances of the components (like DCM) combined with the intrinsic narrow range of CD tolerance of some line interfaces.

SUMMARY

An aspect of the invention provides a method of provisioning a connection across an optical transmission network, the network comprising nodes connected by optical links. The method comprises determining a value of a propagation impairment for a connection across the network between a pair of nodes using at least one measured value of the propagation impairment for a link forming part of the connection, the at least one measured value having been acquired from a previous operation of the link. The method uses knowledge of optimal (measured) values of propagation impairment settings or other physical parameters related to lightpaths already in operation when establishing new lightpaths. This has an advantage of speeding up the provisioning of a new lightpath. The impairment can be chromatic dispersion (CD), or another impairment such as optical noise (e.g. noise power spectral density or optical signal to noise ratio (OSNR)).

The measured value of the propagation impairment can be retrieved from a database, such as a Traffic Engineering Database (TED).

Advantageously, the step of determining a value of the propagation impairment uses an estimated value of the propagation impairment for the link if a measured value is unavailable.

The method is particularly advantageous when applied to impairments or parameters that cumulate linearly along the signal path. The connection can comprise a plurality of links (spans) and the step of determining a value of a propagation impairment for the connection can comprise retrieving a measured value of the propagation impairment for each of the links and combining the retrieved values. Where a measured value of the propagation impairment is not available for any of the links, the method can retrieve an estimated value of the propagation impairment for those links.

The stored information can also be updated in case of events in the network, such as span length variations or fibre type replacements, or can be repeated at periodic intervals.

The method is useful in Wavelength-Switched Optical Networks (WSON), i.e. in photonic signalled networks, because it reduces the time of the re-routing process. It can also be applied to legacy static networks.

An advantage of the method is the reduction of time needed to set up an optical path in a WSON. As a consequence, faster protection mechanism can be assured and faster new services can be provisioned by the operator.

The method only requires a small quantity of messages to be exchanged with NMS (if the PCE is centralised there), or at the control plane level (e.g. WSON network with the PCE in the NE). In one embodiment, a numerical value is returned to the network management system after provisioning of a new lightpath. The method can be performed without any additional hardware. The method can be performed as part of a connection set-up, such as setting up a new connection or a recovery path for an existing connection.

Another aspect provides a method of operating a node of an optical transmission network. The node is connected by optical links to other nodes. The method comprises determining a value of a propagation impairment for a link at the node, acquired from operation of the link. The method further comprises storing the determined value for use in provisioning another connection using the link.

The determined value can be stored locally at the node. Additionally, or alternatively, the step of storing the determined value can comprise sending the determined value to another node of the network for storing at said another node. In the case of a centralised PCE, the determined value can be sent to a node which hosts the centralised PCE and Traffic Engineering Database (TED). In the case of a distributed PCE, the determined value can be sent to a peer node for storing in a local TED at the peer node.

The step of determining a value of a propagation impairment for a link can comprise causing transmission of a test signal across the optical link. The test signal can be transmitted at a lower rate than a line rate of the optical link.

Further aspects of the invention provide apparatus for performing any of the described or claimed steps.

An aspect of the invention provides apparatus for provisioning a connection across an optical transmission network. The network comprising nodes connected by optical links. The apparatus comprises a processor which is arranged to determine a value of a propagation impairment for a connection across the network between a pair of nodes using at least one measured value of the propagation impairment for a link forming part of the connection, the at least one measured value having been acquired from a previous operation of the link.

An aspect of the invention provides apparatus for a node of an optical transmission network. The node is connectable by optical links to other nodes. The apparatus comprises a processor which is arranged to determine a value of a propagation impairment for a link at the node, acquired from operation of the link. The processor is further arranged to store the determined value for use in provisioning another connection using the link. The processor can store the determined value by at least one of: storing the determined value at the node; and sending the determined value to another node of the network for storing at said another node.

The functionality described here can be implemented in hardware, software executed by a processing apparatus, or by a combination of hardware and software. The processing apparatus can comprise a computer, a processor, a state machine, a logic array or any other suitable processing apparatus. The processing apparatus can be a general-purpose processor which executes software to cause the general-purpose processor to perform the required tasks, or the processing apparatus can be dedicated to perform the required functions. Another aspect of the invention provides machine- readable instructions (software) which, when executed by a processor, perform any of the described methods. The machine-readable instructions may be stored on an electronic memory device, hard disk, optical disk or other machine-readable storage medium. The machine-readable instructions can be downloaded to the storage medium via a network connection.

BRIEF DESCRIPTION OF THE DRAWINGS

Embodiments of the invention will be described, by way of example only, with reference to the accompanying drawings in which:

Figure 1 shows an example of an optical transmission network;

Figure 2 shows a node in the network of Figure 1;

Figure 3 shows a first type of network interface which can be used in the node of Figure 2;

Figure 4 shows a second type of network interface which can be used in the node of Figure 2;

Figure 5 shows part of an optical transmission network and impairment values; Figure 6 shows a method of provisioning a lightpath which determines a value of an impairment of the lightpath. DETAILED DESCRIPTION

Figure 1 shows an example optical transmission network 2 with nodes 10. Optical transmission links 5 connect nodes 10. Traffic is carried on links 5 by wavelength channels 6, called lambdas. Connections or lightpaths are established in the network 2. Each lightpath is established between a pair (or more) of nodes 10 of ,

6 the network 2. The terms "connection" and "lightpath" will be used interchangeably. A lightpath can pass via intermediate nodes. Each node has network interfaces for optically transmitting traffic on lambdas and for optically receiving traffic on lambdas. A node 10 connects to multiple links 5 and can comprise a wavelength selective switch (WSS). At a node 10, traffic is received at a network interface on a lambda of an ingress link 5, the traffic is forwarded to a required egress network interface, and is transmitted on a lambda of an egress link 5. A node in an optical network is typically called a Reconfigurable Optical Add Drop Multiplexer (ROADM). A node 10 can forward traffic to other nodes 10 of network 2, or can add traffic received from other nodes not forming part of network 2, or drop traffic to other nodes not forming part of network 2.

In one embodiment, a network management system (NMS) 20 can comprise an entity called a Path Computation Element (PCE) 22 which is responsible for routing lightpaths, i.e. Routing and Wavelength Assignment (RWA). The PCE uses a Traffic Engineering Database (TED) 23. The TED 23 can store information about wavelength availability to help the PCE to select an end-to-end wavelength colour. Information to verify the optical feasibility of a lightpath in an impairment aware PCE can also be stored in the TED 23. In accordance with an embodiment of the invention, the TED 23 can store values of propagation impairments received from nodes 10, where the values of propagation impairments have been acquired from operation of a link 5.

The NMS can send management information to the nodes 10 to configure operation of the nodes. The Path Computation Engine (PCE) can be centralised at a node of the network, or the functionality can be distributed among a plurality of nodes of the network. Similarly, the TED can be centralised or distributed.

Figure 2 shows one of the nodes 10 in the optical transmission network 2 of

Figure 1. Node 10 connects to optical links 51-54. Each link can support a set of lambdas, shown as wo - w _n . Each link 51-54 connects to a respective optical interface 31-34. A wavelength selective-switch 35 connects to the optical interface 30 of each link 5. Figure 2 shows node 10 connecting to four links 51-54, but it will be appreciated that the node 10 can connect to a smaller, or greater, number of links. The number of wavelength channels wo - w _n supported by each of the links 51-54 does not have to be equal. Bi-directional operation is supported by separate lambdas for forward and reverse direction, and advantageously separate links 5 are used for each direction. Each optical interface includes at least one transceiver 32 for transmitting and receiving traffic on lambdas.

Each node 10 can support transmission and reception at one or more bitrates, such as 2.5 Gbit/s, lOGbit/s, 40Gbit/s and lOOGbit/s. Future systems may use higher bitrates. Traffic can be transmitted using one or more modulation formats, such as On- Off Keying (OOK) or a phase modulation format. Possible phase modulation formats include: Differential Quadrature Phase Shift Keying (DQPSK), Dual Polarisation Quadrature Phase Shift Keying (DP-QPSK) and Quadrature Amplitude Modulation (QAM).

Node 10 has a management plane interface 61 for sending and receiving signalling messages with the management plane of the network.

Another way of establishing a connection in the network 2 is by using a distributed control plane. The network 2 can have a control plane, where nodes 10 signal to one another to perform tasks such as setting up paths, tearing down paths, and switching traffic to alternative paths when a fault occurs. Node 10 has a control plane interface 62 for sending and receiving signalling messages, for participating in control plane signalling between nodes 10. Signalling can occur between nodes 10 using a control plane technology such as Generalized Multi Protocol Label Switching (GMPLS) and signalling protocols such as ReSource Reservation Protocol-Traffic Engineering (RSVP-TE) and Open Shortest Path First Traffic Engineering (OSPF-TE). Signalling messages carry information which allows nodes 10 to indicate which wavelengths are available on links 5 between nodes 10 along the proposed lightpath and allows nodes to calculate a Quality of Transmission (QoT) metric for a proposed lightpath. This allows a node 10 to determine if a proposed lightpath will meet a required quality threshold. A memory 65 stores data used by the control plane signalling module 60. Typically, this data is called a Traffic Engineering Database (TED) 123. Embodiments of the invention also store values of impairments such as CD and optical noise. Impairment values 66 can be stored separately from the TED 123, or as part of the TED 123.

A transceiver control module 63 controls operation of the transceivers 32, 42 of the line interfaces. The transceiver control module 63 can control a CD compensator in the transceiver 32, 42 by instructing the CD compensator to compensate for a particular value of CD. „

o

Embodiments of the invention generally apply to any situation where there is a need to set-up or tear-down a connection or lightpath, or validate a connection or lightpath.

A WSON meshed network can be modelled as a graph where nodes are Wavelength Selective Switches (WSS) and arcs are WDM links connecting them. Lightpaths start and end at a WSS and can traverse multiple WSSs, hence a lightpath can be represented as a concatenation of links. CD introduced by WSS nodes is negligible (of the order of ps/nm) and the dominant contribution is from the fibre (transmission fibre of type ITU-T G.652, G.653, G.654, G.655 for example) and dispersion-compensation modules (DCM), if used. CD experienced by a signal through a concatenation of links is equal to the sum of the links' CD, assuming linear propagation. Linear propagation regime is usually desired for optical transport systems.

Figures 3 and 4 show optical network interfaces 31 , 33 of Figure 2 in more detail, and the position of chromatic dispersion (CD) compensators. Figure 3 shows a non-coherent line interface. A tunable chromatic dispersion compensator (TCDC) 25 is positioned in the receive path, after a power amplifier. TCDC uses a thermo- mechanical process to correct for chromatic dispersion. Figure 4 shows a coherent line interface. The CD compensator 26 is implemented as part of Digital Signal Processing (DSP) and is usually split in two parts : one part performs "bulk" and slow compensation and the other performs "fine" and (very) fast compensation. The amount of compensated CD is indicatively ±50ns/nm and ±lns/nm for the "bulk" and "fine" compensators.

In both types of cases the actual value of the CD compensator can be retrieved by means of a signalling exchange between the transceiver controller 63 and the network interface card 31 , 33 and the values are forwarded to the management plane interface 61 and/or the control plane interface 62 as needed.

Impairment settings for lightpaths

Figure 5 shows a part of a WSON with four nodes: A, B, C, D. A lightpath B-

D (i.e. B-C-D) is already equipped and a new lightpath is to be provisioned. Chromatic dispersion will be considered as an example of an impairment. A value of CD for each link of the network can be derived in various ways: - by using design assumptions or other estimation techniques, to calculate an expected value;

- by measuring an actual value used on the link.

An estimated accumulated CD [ns/nm] value is shown against each link. Measured optimal CD [ns/nm] is shown against the equipped lightpath B-D. CD setting for a new lightpath may be derived in various ways according to its route:

A-B: use the expected value;

B-C: use the expected value or use the measured value for the lightpath B-D minus the expected value of the link C-D;

C-D: use the expected value or use the measured value for the lightpath B-D minus the expected value of the link B-C;

A-D: use the expected value of link A-B plus the measured of B-D;

A-C: use the expected value of link A-B and add the expected value of link

B-C.

Figure 6 shows a method of provisioning a lightpath which determines a value of an impairment of the lightpath. Chromatic dispersion (CD) is considered as an example of an impairment. The method checks, at step 101, if a new connection has been requested. If a new connection has been requested, the method proceeds to step 102 to begin a Routing and Wavelength Assignment (RWA) task. RWA attempts to route a lightpath across the WSON and assigns a wavelength (lambda) to the lightpath. Advantageously a single wavelength is used across the RWA but it is also possible to transfer traffic between wavelengths at a node if it is not possible to find a single wavelength for an end-to-end lightpath.

Step 103 determines a value of an impairment for the lightpath. As explained above with reference to Figure 5, this step uses an actual measured value of the impairment where possible as this is likely to be close to the actual value needed. For a lightpath which comprises a plurality of links, the method attempts to use measured values for as many of the links as possible. For any links where measured values are unavailable, the method uses expected values. At step 104 the link settings determined in previous steps are sent to the relevant nodes.

At step 105 a node is instructed to apply settings to compensate for the impairment value. In a node having an interface of the type shown in Figure 3, the TCDC module applies the value of CD determined at step 103. In a node having an _{1 Q} interface of the type shown in Figure 4, the bulk and fine parts of the CD compensator apply the value of CD determined at step 103.

At step 106 an actual CD setting is acquired. In some cases, the value applied at step 105 will be at, or very close to, the actual value required and no further adjustment will be required. This should be the case if a measured value of CD was obtained at step 103. This minimises the delay before transmission begins at step 107. In other cases, some adjustment will be required. At step 107 traffic transmission begins.

At step 108 network databases are updated with the actual value of CD acquired at step 106. In this way, when another lightpath is provisioned, the method can use a more accurate value of CD and hence reduce the time required to establish the lightpath.

Steps 101-103 of Figure 6 can be performed by a PCE. Steps 105, 106 of Figure 6 can be performed by the transceiver controller 63 shown in Figure 2. The PCE can be a centralised PCE (i.e. a PCE forming part of the NMS) or it can be a distributed PCE (i.e. a PCE is located at each node). Figure 6 shows a method for an embodiment with a centralised PCE. Steps 101-103 are performed at the centralised PCE and there is communication of settings selected by the centralised PCE to a NE at step 104. The node sends any updated impairment values to the centralised PCE at step 108. In an embodiment with a distributed PCE, step 104 can be omitted as the PCE is located at the NE. However, there may be an additional step of communication between the PCE at one NE and a similar PCE module at another NE in order to establish the lightpath. With a distributed PCE, there is a signalling mechanism to distribute impairment values to other NEs. For example, this can occur after each new transmission (e.g. step 108) or at periodic intervals. A signalling protocol such as Open Shortest Path First (OSPF) or Open Shortest Path First Traffic Engineering (OSPF-TE) can be used to distribute impairment values.

The method of Figure 6 can be performed periodically. A possible timescale for a periodic check is of the order of a day.

The complexity of the problem of determining CD values scales with the number of nodes and links. One way of determining values of an impairment will now be described. It is proposed to build a path allocation p x I binary matrix M, an array _/ of CD values for each link and an array y of CD values for each lightpath. has a row for each equipped lightpath and a column for each link. M(i,j) is 1 if the lightpath i is routed through the link j, 0 otherwise.

When a new lightpath k is requested, a new array N is built consisting of / entries with the same logic as M. It is first checked that its CD setting can be obtained only by measured values. This is preferable because the estimation will be more reliable and the chances to get the optimal value are best. This can be done by taking the rows of M as basis of a linear space. If N is a linear combination of a subset of rows of M, then this is possible to infer the CD setting for the new lightpath as a linear combination of measured values. In this case, the CD setting for k is obtained by just applying the same linear combination to the corresponding entries of the array y of measured values.

If N cannot be expressed as linear combination of a subset of rows of M, the CD setting is obtained just by multiplying Nwith x.

Finally, N is appended to M and becomes its last row as soon as the measured CD value is available.

When the line interface returns the optimal value (step 106, Figure 6) it is possible update the CD database (step 108, Figure 6). As the number of lightpaths becomes large it is proposed to estimate links' CD in a least-square sense:

Formally, this can be done when M ^T M has at least rank / (i.e. it is invertible). Here is a description of the whole process: Example

An example of the matrix-based method of determining impairment values will now be described for the simple network shown in Figure 5. The first traffic in this network is B-D. There is not yet a measured value for CD. = (0 1 l), x 2.3 then j = x = (4.0)

1.7

Suppose the interface returns 4.1 as the optimal, measured, value of CD. y is set to (4.1) The second traffic is on link A-B. Its array N is (1 0 0) which is not a linear combination of rows of M. The CD setting can only be estimated by multiplying N and x and is 2.0.

The receiver returns a measured value of 1 .9, so we obtain:

The third traffic is on link B-C. Its array N is (0 1 0) which is not a linear combination of rows of M. The CD setting can only be estimated by multiplying N and x and is 2.3.

The receiver returns a measured value of 2. 1 , so we obtain:

Because ¾ is invertible and x has been updated according to Eqn. 1 .

The fourth traffic is on link C-D. Its array N is (0 0 1) which is a linear combination of rows of M (i.e. first minus the third). The CD setting is the first minus the third entry of y, that is 2.0.

The receiver returns a measured value of 1 .8, so we obtain:

M

The CD setting could also have been obtained by multiplying M and x. In this case the two methods give the same result of 2.0 for C-D. In practical cases, they could be different and a rule set can be used to select a value of CD.

Impact on Control/Management Plane

The method is fully compliant with the WSON structure. The x array of CD values per link has a structure equivalent to the TED (the traffic engineering database) stored in each NE and, adding proprietary protocol extensions, can be signalled across the network as well as other TED pertaining information. ^

The information relevant to measured CD for lightpath is partially available in the NE (i.e. for a lightpath which is generated or terminated in the NE) and fully available in NMS.

In the method shown in Figure 6, actual measured values of an impairment (e.g. CD) are acquired when a lightpath is established across the network. The method can also comprise a step of performing pre-evaluation of impairment (e.g. CD) settings by using test or probe signals. The test or probe signals do not have to carry real traffic. These probe signals can be generated by spare line interfaces used to fill the optical spectrum for other purposes, such as power balancing. These probe signals can be at the same line rate, or at much lower line rates as traffic-carrying line interfaces. Transmitting the probe signals at a lower line rate can reduce the operating cost. A lower line rate can be used because chromatic dispersion is a physical parameter and does not depend on the kind of signal being transmitted.

The method can be extended to take into account spectral dependency of the CD that may be significant in very long uncompensated links. For example, in a 2000km link over G.655 eLEAF fibre, channels at the extremes of the C-band would see a CD difference of the order of 2000 km * 0.084 ps/(nm ^A 2 km) * 30 nm ~ 5ns/nm.

One value of CD can be estimated/measured/stored per link. It is also possible to store a set of values of CD for a link. A value of CD can be stored per wavelength. It is known that the value of CD vs. wavelength can be well approximated by a 2nd order polynomial. Accordingly, an alternative to storing a value of CD per wavelength is to store a set of three (or more) values acquired at different wavelengths and to perform a calculation using the polynomial to determine the value at a particular wavelength of interest.

The methods described above can be applied to other kinds of propagation impairments, and especially those that accumulate linearly, such as optical noise. In the case of OSNR, it would be possible to estimate the OSNR degradation introduced by each link, given the OSNR at the receiver end point. Considerations made about average CD hold for average OSNR (taken in linear units). The OSNR degradation introduced by a WSS can also be included in the database of impairment values. OSNR degradation introduced by a WSS is known by design, and is not a variable that needs to be estimated. An average value of OSNR can be stored, representing an average value across a range of wavelengths used on a link. Alternatively, individual ^ wavelength values could be stored. A value of OSNR can be associated with the kind of traffic/modulation scheme.

The speed of the method can be further increased by reporting measured values of an impairment parameter obtained from lightpaths on a per-link basis. In this way, as soon as a new path request is raised, the computation to estimate the CD of the new lightpath is limited to multiplying N with x, where the x is no more relevant to only theoretical values for CD per link but, where possible, CD per link is calculated starting from measured values. The new x array calculation can be performed without any speed constraint for example as soon as a new lightpath is set up and measured or on a regular schedule or whatever strategy the network operator wants to implement.

Another additional advantage is the exploitation of live measurements from equipped interfaces to evaluate the feasibility of provisioning new lightpaths: for example adding non-coherent interfaces in uncompensated network (by pushing their CD tolerance to the limit) or upgrading the network to 100G or 400G (by pushing its OSNR tolerance to the limit). The network operator benefits from the accurate knowledge of the CD to exploit as much as possible the CD tolerance of the line interfaces. For example, the operator is able to operate close to the CD tolerance of the line interface because the CD is known rather than allowing some safety margin for uncertainty. Similarly, in the case of OSNR, knowing an actual value of OSNR can allow an operator to operate closely to an OSNR limit, rather than allowing an uncertainty margin.

When used for OSNR estimation, the method may be used for fault localisation, i.e. to identify links where an unexpected drop of OSNR occurs (i.e. unexpected span loss increase or shortfall of amplifier output power).

Modifications and other embodiments of the disclosed invention will come to mind to one skilled in the art having the benefit of the teachings presented in the foregoing descriptions and the associated drawings. Therefore, it is to be understood that the invention is not to be limited to the specific embodiments disclosed and that modifications and other embodiments are intended to be included within the scope of this disclosure. Although specific terms may be employed herein, they are used in a generic and descriptive sense only and not for purposes of limitation.