Switch for an optical WDM network

i

SWITCH FOR OPTICAL TRANSMISSION NETWORK

USING WAVELENGTH MULTIPLEXING

The present invention relates to a switch intended to be implemented in an optical communication network using wavelength multiplexing.

The present invention is in the field of optical switches, or optical switching nodes, having a so-called multi-granularity architecture. Granularity is a notion which relates to predefined sets of transmission resources (typically carrier wavelengths or wavelength multiplexes), the resources of such a set being able to be considered as a whole for certain processing operations. common (typically switching). A multi-granularity architecture therefore takes into account different levels of granularity to switch the total traffic at the level of a switch. For example, part of the total traffic can be switched at the so-called fiber level, that is to say grouping together all the wavelengths likely to be conveyed by an optical fiber, which therefore corresponds to the most granular level. Student. Another part can be switched to the wavelength band level, which corresponds to an intermediate level of granularity. A last part can be switched to the wavelength level, which corresponds to the lowest level of granularity. Intermediate levels of granularity can still be defined.

The implementation of a multigranularity architecture makes it possible to limit the increase in the complexity of switches in optical networks.

Indeed, telecommunications are currently experiencing a very significant boom resulting in increased needs in data transmission. Optical fiber transmission is particularly implicated by this phenomenon and the quantity of data transmitted by optical networks is constantly increasing. This results in an increase in the number of fibers installed in the networks as well as in the number of carrier wavelengths used.

An optical fiber is now capable of transmitting up to 256 wavelengths, each wavelength being able to provide a data rate of 10 gigabits (1 Gbit = 10⁹ bits) per second. Thus, depending on the number of fibers arriving at the input of the optical switch, the total throughput to be switched may be greater than several tens of terabits (1 Tbit = 10¹² bits) per second.

An optical switch which has a multi-granularity architecture makes it possible to process such data rates by switching partly wavelengths and partly wavelength bands, that is to say respectively single-length channels wavelengths and wavelength multiplexes. The switch can further process groupings of bands. According to yet another possibility, it could also process only bands of wavelengths and groupings of bands. To simplify the description and by way of example only, we will consider in the following the case at three levels of granularity: wavelength, band and fiber, this last level corresponding to a particular case of grouping of conveyed wavelengths. by an optical fiber, shows a diagram of a node of all the bands grouping together likely to be

FIG. 1 optical switching with a multigranularity architecture, according to the prior art.

With the multi-granularity architecture, we have gone from monoblock switching nodes to switching nodes made up of a stack of sub-nodes.

Each switching sub-node is assigned a corresponding level of granularity. Thus, in the example shown, there is an FXC switching sub-node associated with the level of fiber granularity (which is a particular case of grouping of bands), a BXC switching sub-node associated with the level of band granularity, and a WXC switching sub-node associated with the wavelength granularity level.

In Figure 1, the incoming IF fibers are first sent to the IP ingress ports of the FXC switching sub-node. Among the incoming IF fibers, a few fibers are directly switched to the OF output fibers through the OP output ports of the FXC switching sub-node. An AF fiber is directly inserted from the customer to a Pi fiber insertion port_ns of the FXC switching sub-node. A DF fiber is extracted from a P fiber extraction port_ext from the FXC subnode and is sent to the client. The DF fiber must be demultiplexed in wavelengths for the customer, but the demultiplexers are not shown in the figure. F fibers_bf are inserted from the BXC switch sub-node on Pi fiber insert ports_nsof the FXC subnode. These fibers F_bf come from the band-fiber Mux LF multiplexer which multiplexes the bands coming from the output ports OP of the BXC switching sub-node. Finally, Ff fibers_bare extracted from the FXC sub-node through extraction ports and are sent to the IP input ports of the BXC sub-node after demultiplexing the fibers into bands in the fiber-band demultiplexer Demux F B.

The same switching process is found at the next lower granularity level, i.e. in the switching sub-node at the BXC band granularity level, as well as at the lowest granularity level, i.e. say in the switching subnode at the WXC wavelength granularity level.

Of the bands arriving at the IP input ports of the BXC subnode, a few are switched to the OP output ports of the BXC subnode. An AB tape is directly inserted from the client to an insert port of the BXC subnode. A DB tape is extracted through a P extraction port_ext of the FXC subnode and is sent to the client. The DB band must be demultiplexed in wavelengths for the customer, but the demultiplexers are not shown in the figure. B bands_b are inserted from the WXC switch sub-node on Pi insert ports_ns of the BXC sub-node. These B bands_bcome from the multiplexer Mux B, which provides band multiplexing of the wavelengths coming from the output ports OP of the switching sub-node BXC. Finally, Bb bands are extracted from the BXC sub-node through extraction ports and are sent to the IP input ports of the WXC sub-node after demultiplexing the bands in wavelengths in the band demultiplexer - length d 'demux wave B

The same switching process is found again in the WXC subnode. Of the wavelengths arriving at the IP input ports of the WXC subnode, a few are switched to the OP output ports of the WXC subnode. Wavelengths A are directly inserted from the customer on insertion ports P_ins of the WXC subnode. D wavelengths are fetched through pull ports of the WXC subnode and are sent to the client.

This architecture according to the prior art, as it has just been described in conjunction with FIG. 1, implements switching matrices (typically based on optical crossbar switches) separated for each level of granularity. The fiber granularity level is processed in the FXC switching matrix, the band granularity level is processed in the BXC switching matrix, and the wavelength granularity level is processed in the WXC switching matrix. We therefore have a specific switching matrix by granularity. For given numbers of input ports assigned respectively to the three levels of granularity, this solution is that which makes it possible to limit the complexity and the size of the assembly as much as possible.

However, the number of input / output ports of each of the switching matrices allocated to each level of granularity being fixed, this becomes a drawback if we plan to modify this architecture to adapt it to changes in traffic over time. .

Let us take a concrete example for such an architecture with 10 Gbit / s throughput per wavelength, 16 wavelengths per band and 10 bands per fiber. It may be necessary to switch:

in an initial step: 500 wavelengths, no band, no fiber, which represents a total throughput of 5 Tbit / s;

<td>- in</td><td>a second step:</td><td> 250</td><td>lengths</td><td>wave, 250</td>

<td>bands,</td><td>no fiber, which</td><td colspan="2">represents a</td><td>total flow</td>

<td>from 42.5</td><td>Tbit / s;</td><td></td><td></td><td></td>

<td>- in</td><td>a third step:</td><td> 100</td><td>lengths</td><td>wave, 400</td>

<td>bands,</td><td>no fiber, which</td><td colspan="2">represents a</td><td>total flow</td>

<td colspan="2">of 65 Tbit / s;</td><td></td><td></td><td></td>

<td>- in</td><td>a fourth step:</td><td> 100</td><td>lengths</td><td>wave, 300</td>

bands, 100 fibers, which represents a total speed of 209 Tbit / s;

- in a fifth step: no wavelength, 200 bands, 300 fibers, which represents a total throughput of 512 Tbit / s;

During the first step, a WXC 500 * 500 switching matrix must be provided (which means a number of matrix states equal to 500 * 500) for the wavelength granularity. However, this WXC switch fabric will not be fully utilized in the following steps.

In the third step, it is necessary to provide a BXC 400 * 400 switching matrix for the band granularity. But, during the fifth step, only half of the input / output ports of this matrix will be used.

Finally, in the fifth step, an FXC 300 * 300 switch matrix for fiber granularity is needed. Again, this switching matrix is ​​underused in the other steps.

Thus, according to the previous example of evolution, with the architecture of the prior art, the total number of input ports to be provided in the optical switch is equal to 1200, and will only be partially used.

Also, the aim of the present invention is to implement an architecture making it possible to switch different levels of granularity, while avoiding the drawbacks of the prior art, that is to say by providing an architecture which is optimal not at a given stage of the evolution of the traffic to be switched but for a set of configurations adapted throughout this evolution.

To this end, the invention proposes to use one and the same switching matrix to switch all the levels of granularity at the same time. The three separate switching matrices according to the prior art, corresponding respectively to a level of fiber, band and wavelength granularity, are replaced by a single switching matrix which processes all the granularities. According to needs, that is to say according to the traffic to be switched, adapted numbers of ports of the single matrix will be assigned respectively at a low level of granularity (the wavelengths), at an intermediate level of granularity ( wavelength bands), and finally at a high level of granularity (fibers).

The invention therefore relates to an optical switch for an optical network using wavelength multiplexing, comprising:

pl input ports respectively receiving pl wavelengths, p2 output ports, and first routing means able to direct the wavelengths received on said pl input ports selectively towards said p2 output ports, and / or - ql input ports receiving respectively ql wavelength bands, q2 output ports, and second routing means capable of routing the bands of wavelengths received on said ql input ports selectively towards said q2 p output ports, and / or r1 input ports respectively receiving r groups of bands, r2 output ports, and third routing means able to direct the groups of bands received on said r1 input ports selectively towards said r2 output ports, said switch comprising at least two of said first, second and third routing means, characterized in that said first, second and third routing means consist of a single switching matrix capable of coupling any one of said pl + ql + rl input ports to l 'any of said p2 + q2 + r2 output ports.

Other characteristics and advantages of the invention will emerge more clearly on reading the following description of a particular embodiment with reference to the figures in which: FIG. 1 is a diagram of an optical switch implementing an architecture multigranularity according to the prior art, as described in the preamble above;

FIG. 2 is a diagram of an optical switch according to the present invention.

In the preferred embodiment of the invention described below with reference to FIG. 2, the number of granularity levels is equal to three: the wavelength, the wavelength band and the fiber. The invention can nevertheless be implemented with a number of levels of granularity equal to two or greater than three.

Incoming fibers IF are received at the input of switch 1 which delivers outgoing fibers OF at the output. A series of wavelengths A are also added from the customer's premises to switch 1, while a series of wavelengths D are taken from switch 1 to the customer.

Switch 1 implements a single multi-granularity switch matrix MXC which comprises a first series of IP input ports, equal in number to p1, assigned at the wavelength granularity level to respectively receive p1 lengths. wavelength, a second series of input ports IBP, of a number equal to ql, assigned to the level of granularity ίο wavelength band to receive respectively ql wavelength bands and a third series of ports entrance

IFP, of a number equal to r1, assigned to the level of fiber granularity to respectively receive r1 fibers.

The single MXC matrix also includes, in correspondence with the input ports, a first series of output ports OP, a number equal to p2, assigned to the wavelength granularity level, a second series of output ports OBP, with a number equal to q2, assigned to the wavelength band granularity level, and a third series of OFP output ports, with a number equal to r2, assigned to the fiber granularity level.

At the input of switch 1, the heart of which is delimited by a dotted line in FIG. 2, there is an input interface made up of a set of fiber-band Demux demultiplexers. _P__B and band-wavelength demultiplexers Demux _Β-λ · At the output of switch 1, there is an output interface made up of a set of wavelength - MUX band multiplexers χ._Β and MUX band-fiber multiplexers _B._P. The output interface can also include wavelength converters, band converters and / or regenerators, not shown in FIG. 2. However, their presence is optional.

In the heart of switch 1 there is an internal rearrangement zone consisting on the one hand of a set of Demux fiber-band demultiplexers _P._B and band-wavelength demultiplexers Demux _Β_χ, of the same type as those described above and, on the other hand, of a set of MUX band wavelength multiplexers χ__Β and MUX band-fiber multiplexers _B._F, also of the same type as those described above.

Among the incoming IF fibers, some, after having first been demultiplexed into bands through Demux fiber-band demultiplexers _F._B, are again demultiplexed in wavelength through Demux band-wavelength demultiplexers _Β-χ then are sent to the pl input ports assigned to the wavelengths ΐλΡ of the single matrix MXC. The wavelengths are then switched, by means of first routing means, to the p2 output ports assigned to the wavelengths ΟλΡ of the matrix MXC. The first switching means consist of the single switching matrix MXC and make it possible to switch the wavelengths received on the pl input ports assigned to the wavelengths selectively to the p2 output ports assigned to the lengths d 'wave.

Other incoming IF fibers are demultiplexed into wavelength bands through Demux fiber-to-band demultiplexers _F._B and are sent to the ql input ports assigned to the IBP wavelength bands of the MXC matrix. The wavelength bands are then switched, via second routing means, to the q2 output ports assigned to the OBP wavelength bands of the MXC matrix. The second routing means consist of the single switching matrix MXC and make it possible to direct the wavelength bands received on the q1 input ports selectively towards the q2 output ports.

Finally, certain incoming IF fibers are sent directly to the r1 input ports assigned to the IFP fibers of the MXC matrix to be switched, via third routing means, to the r2 output ports assigned to the OFP fibers of the MXC matrix. These third switching means consist of the single switching matrix MXC and make it possible to route the fibers received on the r1 input ports selectively towards the r2 output ports.

At the output ports assigned to the wavelength granularity OP, certain wavelengths are directed to the output interface and are then multiplexed into bands and then into fibers via the MUX band wavelength multiplexers χ__Β and tape - MUX fiber _b-f- Other wavelengths may be directed to the internal rearrangement area. These wavelengths can then be looped back to the MXC switching matrix on the one hand, at the input ports assigned to the IBP band granularity by means of the wavelength-band MUX χ multiplexers._Β located in the internal rearrangement zone and, on the other hand, at the level of the input ports assigned to the IFP fiber granularity via the wavelength multiplexers - MUX band χ-_Β and multiplexers

<img file="FR2814902A1_D0001.tif" />

At the output ports assigned to the OBP band granularity, some bands are routed directly to the output interface and are fiber-multiplexed via MUX tape-fiber multiplexers _B__P. Other bands can be directed to the internal rearrangement area. These bands can then be looped back to the MXC switching matrix on the one hand, at the input ports assigned to the IP wavelength granularity via the band - wavelength demultiplexers Demux_Β-λ located in the internal rearrangement zone and, on the other hand, at the level of the input ports assigned to the IFP fiber granularity via the band - MUX fiber multiplexers _b-fAt the output ports assigned to OFP fiber granularity, some fibers are routed directly to the output interface. Other fibers can be directed to the internal rearrangement zone. These fibers can then be looped back to the MXC switching matrix on the one hand, at the input ports assigned to the IBP band granularity via the Demux fiber-band demultiplexers._f-b located in the internal rearrangement zone and, on the other hand, at the level of the input ports assigned to the IP wavelength granularity via the fiber-band Demux demultiplexers _F._B and band-wavelength demultiplexers Demux _B_x.

According to an advantage of the invention, this single-matrix architecture makes it possible to switch all the granularities at the same time and makes it possible to be more flexible as a function of the evolution of the traffic to be switched in the optical switch. The single switching matrix implemented in the switch according to the invention is provided to be able to couple any one of the p1 + ql + rl input ports to any one of the p2 + q2 + r2 output ports, even if for a given multi-granularity configuration all the possible switch states are generally not used.

Thus, input / output ports which were assigned to wavelengths can subsequently be assigned to wavelength bands so as to increase the capacity of the matrix in terms of throughput. This increase in capacity is achieved without changing the space switch but by intervening at the input and output interfaces, as well as at the level of the internal rearrangement zone. That is to say by modifying the connections at the level of demultiplexers and multiplexers. Such a switching matrix therefore has the ability to adapt to increasingly high data rates by being able to change the capacity at will.

This single MXC switching matrix implements the same technology as the fiber granularity level switching matrices used in the architectures according to the prior art implementing a separate switching matrix for each granularity.

According to a particular embodiment of the invention, the input interface and the output interface as described above are omitted. The demultiplexing at the input of the switch and the multiplexing at the switch 1 puts internal rearrangement, the output is therefore removed and only the area formed by all the multiplexers and demultiplexers located in the architecture is implemented. In this particular mode, only the fiber granularity is entered in switch 1.

are then directed internally to be band granularity,

Interior realization level of

The fibers towards the demultiplexed zone then at the level of rearrangement at the level of wavelength granularity.