Undirected Cross Connects Based on Wavelength-Selective Switches

Abstract

Undirected cross connects are provided based on wavelength-selective switches A demand d={i, j} is routed over a path pd between ports i and j in a multi-layer network based on one or more wavelength selective switches, by determining a middle layer node n in the multi-network for which there is a first path pi between the port i and the middle layer node n and a second path pi between the port j and the middle layer node n; identifying a node n′, wherein the node n′ is a first node starting from port i that path pi shares with path pj; and concatenating the path pd comprised of a subpath pi′ of the first path pi from the port i to the node n′ with a subpath of p′j of the second path pi from the node n′ to the port j. An undirected Cantor network is disclosed where the switch nodes are wavelength selective switches. An undirected Clos cross connect is also disclosed where one or more undirected switches are undirected Cantor networks having at least one wavelength selective switch

Description

FIELD OF THE INVENTION

The present invention relates to wavelength-selective switches, and more particularly, to undirected cross connects employing such wavelength-selective switches

BACKGROUND OF THE INVENTION

The nonblocking of cross connect switches in a communications network assesses the ability to route connections from input ports to output ports A cross connect C can be thought of as a directed graph where input ports are nodes with no incoming edges and output ports are nodes with no outgoing edges A request for a connection from an input port a to an output port b will be a request for a directed path from a to b For a more detailed discussion of nonblocking cross connects, see, for example, J. Y. Hui, Switching and Traffic Theory for Integrated Broadband Networks, Kluwer Academic Publishers, Norwell, Mass. (1990) and G. M. Masson et al, “A Sampler of Circuit Switching Networks,” IEEE Computer, 5:32-48 (June 1979)

For many networks, however, such as long-haul optical networks, connection requests have “bidirectional symmetry” In other words, there is a request to connect i to j if and only if there is also a request to connect j to i and moreover these two connections should be routed on the same bidirectional links. See, J Simmons et al, “Optical Crossconnects of Reduced Complexity for WDM Networks with Bidirectional Symmetry, IEEE Photonics Technology Letters, 10(6):819-821 (June 1998). This is due to the fact that optical network management is typically designed to connect transceivers in pairs A restriction to bidirectional demands may allow for simplified cross connect designs For bidirectional demands, a cross connect will not have input and output ports, instead it will just have ports

The cross connect will be an undirected network where the ports are leaf nodes and a request for a connection will be a request for an undirected path between two ports. This is of interest when one has symmetric demands (i.e., there is a demand d from node i to node j if and only if there is a demand d′ from node j to node i) and moreover demands d and d′ should be routed along the same undirected path

A cross connect is said to be strictly nonblocking if all connections from input ports to output ports can be routed without disturbing other connections A cross connect is said to be widesense nonblocking if there exists a routing algorithm such that if all previous connections have been routed using that algorithm, then that algorithm will find a route for any subsequent connection requests

A need exists for a modification to traditional cross connect designs for this undirected case A further need exists for improved cross connect designs for the undirected case that exhibit good nonblocking properties. Yet another need exists for improved cross connect designs based on various four-port switches

SUMMARY OF THE INVENTION

Generally, undirected cross connects are provided based on wavelength-selective switches According to one aspect of the invention, a method is provided for routing a demand d={i, j} over a path p_d between ports i and j in a multi-layer network based on one or more wavelength selective switches. The method comprises the steps of determining a middle layer node n in the multi-network for which there is a first path p_i between the port i and the middle layer node n and a second path p_j between the port j and the middle layer node n; identifying a node n′, wherein the node n′ is a first node starting from port i that path p_i shares with path p_j; and concatenating the path p_d, comprised of a subpath p_i′ of the first path p_i from the port i to the node n′ with a subpath of p′_j of the second path p_j from the node n′ to the port j A demand routed in accordance with the disclosed method can be considered widesense non-blocking

According to another aspect of the invention, an undirected Cantor network is disclosed that comprises k ports; at least one layer of switch nodes, wherein the switch nodes are wavelength selective switches; and k switches for connecting the k ports to the at least one layer of switch nodes The k switches can be, for example, 1×3 switches In one exemplary embodiment, the disclosed undirected Cantor network comprises half of a directed version of a Cantor network The disclosed undirected Cantor network can be simplified by replacing one of more nodes in a middle layer by one or more connections and then merging as pairs one or more nodes in a final layer that are connected by two parallel links

According to yet another aspect of the invention, an undirected Clos cross connect is disclosed that comprises k ports; n first stage p×m switches connected to the k ports; and a plurality of second stage undirected switches, wherein one or more of the undirected switches are undirected Cantor networks having at least one switch node that is a wavelength selective switch The first stage switches can be hybrid cross connects

A more complete understanding of the present invention, as well as further features and advantages of the present invention, will be obtained by reference to the following detailed description and drawings

BRIEF DESCRIPTION OF THE DRAWINGS

FIGS. 1 and 2 illustrate exemplary reflective wavelength selective switches that can be used in accordance with the present invention to construct an undirected cross connect;

FIG. 3 illustrates an exemplary 8×8 Cantor network;

FIG. 4 illustrates an alternative exemplary 8×8 Cantor network;

FIG. 5 illustrates an exemplary 8×8 undirected Cantor network in accordance with an embodiment of the present invention;

FIG. 6 is a flow chart describing an exemplary implementation of a routing algorithm, Alg, incorporating features of the present invention;

FIGS. 7 and 8 illustrate simplified versions of the exemplary 8×8 undirected Cantor network of FIG. 5;

FIG. 9 illustrates a conventional Clos cross connect design having k input ports and k output ports;

FIG. 10 illustrates an undirected Clos Cross network with k ports, in accordance with the present invention; and

FIG. 11 illustrates an exemplary implementation of a 2×3 Clos cross connect switch.

DETAILED DESCRIPTION

The present invention provides undirected (or bidirectional) widesense nonblocking cross connects based on four-port switches.

Directed Cross Connect Definitions

A k×k directed cross connect is a directed graph C with k source nodes (i.e, nodes with no incoming edges) called input nodes and k sink nodes (i.e., nodes with no outgoing edges) called output nodes. A directed demand is a request for a directed path in C from an input node to some output node The notation d=(i, j) denotes the directed demand d with input node i and output node i. A set S of directed demands is valid if all directed demands in S have distinct input nodes and distinct output nodes. A routing R_d of a directed demand d is a directed path in C from the input node to the output node of d A valid Touting of a valid set of directed demands S is a set of routings R={R_d: dεS} so that if d, d′εS, then R_d and R_d are are disjoint.

Undirected (or Bidirectional) Cross Connects Definitions

A size k (undirected) cross connect C is an undirected graph with k leaf nodes called ports. A demand is a request for a path in C between a pair of ports. The notation d−{i, j} denotes the demand d that requires a path between ports i and j A set S of demands is valid if no two demands in S have a port in common. A routing R_d of a demand d is a path in C between the ports of d A valid routing of a valid set of demands S is a set of routings R−{R_dεS} so that if d,d′εS, then R_d and R_d are edge disjoint

Hybrid Cross Connects

A hybrid cross connect is an undirected (bidirectional) network that has two sets of ports A and B A hybrid demand is a request for a path between a pair of ports in A or a request for a path between a port in A and a port in B

General Definitions

All definitions in this section are given for undirected cross connects but remain the same for directed (or hybrid) cross connects if the term “demand” is replaced by “directed demand” (or “hybrid demand”)

There are traditionally three levels of′ nonblocking from strongest to weakest, they ate:

• 1. Strictly Nonblocking—A cross connect is strictly nonblocking if for all valid demand sets S, all valid routings R of S, and any demand d∉S such that S′=S∪{d} is a valid demand set, there is a valid touting R′ of S′ such that R⊂R′.
• 2. Widesense Nonblocking—A cross connect is widesense nonblocking if there is a routing algorithm, Alg, such that for all valid demand sets S, all valid routings R of S that have been computed using algorithm Alg, and any demand d∉S such that S′∉S∪{d} is a valid demand set, algorithm A/g will find a valid routing R′ of S′ such that R⊂R′.
• 3. Rearrangeably Nonblocking—A cross connect is rearrangeably nonblocking if for any valid set of demands S there is a valid touting R of S

Wavelength Selective Switches

Although the present invention is illustrated herein in the context of optical networks and in particular, wavelength division multiplexing (WDM) networks, the disclosed cross connects are not specific to any particular kind of network e g., optical, electronic or wireless mesh networks, as would be apparent to a person of ordinary skill in the art

The cross connects will be wavelength selective cross connects In other words, each demand will be a request to route a particular wavelength between two given ports

U.S. patent application Ser. No. 11/434,938, entitled, “Multiple Port Symmetric Reflective Wavelength-Selective Mesh Node,” filed May 16, 2006 (Attorney Docket No Doerr 116-27), and incorporated by reference herein, discloses reflective wavelength selective switches (WSS) used as building blocks for the undirected cross connects disclosed herein See also, U.S. patent application Ser. No 11/434,919, entitled, “Multiple Port Symmetric Transmissive Wavelength-Selective Mesh Node,” filed May 16, 2006 (Attorney Docket No Doerr 117-28), incorporated by reference herein

Generally, the disclosed reflective wavelength selective switches multiplex and demultiplex the wavelengths at each port and there is a steering mirror for each available wavelength. The switches are designed so that they are nonblocking, i e., a connection made by the switch does not have to be interrupted when it is subsequently required to form an additional connection

FIGS. 1 and 2 illustrate reflective wavelength selective switches 100, 200 that can be used in accordance with the present invention to construct an undirected cross connect In FIGS. 1 and 2, the switches 100, 200 have an open circle to indicate each port and have solid dots to indicate mirror states independently chosen for each wavelength

Degree-4 Node

The exemplary 1×6 WSS switch 100 shown in FIG. 1 is an undirected cross connect with four ports for providing one channel of a multiple wavelength channel signal with reciprocal connectivity between node ports A, B, C, and D. For a four port mesh node, a total of 4!/[2!(4−2)!] or 6 unique node port pair connections must be made by the 1×6 WSS apparatus. Note that only terminal 4 (of the seven terminals labeled left to light as 0-6) of the 1×6 WSS switch 100 is unconnected Two of the terminals (2 and 5) are directly connected to node ports B and D, respectively The node port A connects via a 1×2 directional coupler (not shown) to terminals 1 and 3 of the 1×6 WSS apparatus 100 The node port C connects via a 1×2 directional coupler (not shown) to terminals 0 and 6. The steerable mirror of the 1×6 WSS is switched to one of three positions (or states), as denoted by the left, center, and right connection dots “•”. The 1×6 WSS 100 is switchably controlled by a control signal (not shown) to enable reciprocal connections to be established between six unique pairs of node ports (i.e., AB, AC, AD, BC, BD, BC) by switching the minor, to one of the three positions Thus, as shown in table 110, in a first switching state (1) (associated with left connection dot), two simultaneous connections are made between ports A-B and C-D (since two sets of terminals 2 and 3 as well as 0 and 5 are symmetrically located around the left connection dot). In the second switching state (2) (associated with the center connection dot), a connection is made between ports A-C and B-D (since the terminals 1 and 6 as well as 2 and 5 are symmetrically located around the center connection dot) In the third switching state (3) (associated with the tight connection dot), a connection is made between ports A-D and C-B (since the terminals 3 and 5 as well as 2 and 6 are symmetrically located around the right connection dot) In this manner, a reciprocal connection is established between each unique pair of node ports by switching the mirror of the 1×6 WSS to one of the three positions While the degree-4 mesh node has been shown to include a 1×6 WSS, it should be noted that any 1×K switch can be used, where A′ is greater than or equal to 6, with the additional terminals left unconnected

Partitioned Degree-4 Node

The exemplary 1×6 WSS switch 200 shown in FIG. 2 is an undirected cross connect with four ports for providing one channel of a multiple wavelength channel signal with reciprocal connectivity between node ports A, B, C, and D FIG. 2 illustrates a hybrid design, referred to as a hybrid-4 node with six terminals As shown in table 210, the switch 200 can connect port A with port B or alternatively can connect port A with port C and port B with port D or port A with port D and port B with port C

Note that only terminal 3 (of the six terminals labeled left to right as 0-5) of the 1×5 WSS switch 200 is unconnected Three of the terminals (1, 4 and 5) are directly connected to node ports B, D and C, respectively The node port A connects via a 1×2 directional coupler (not shown) to terminals 0 and 2 of the 1×5 WSS apparatus 200 The steerable mirror of the 1×5 WSS is switched to one of three positions (or states), as denoted by the left, center, and right connection dots “•”.

Along these same lines, a 3-port switch can be constructed that allows only the connection A-B or the connection A-C, as would be apparent to a person of ordinary skill in the art. This will be called a 1×2 switch In a straightforward manner, a 1×n switch can be constructed using n−1 such 1×2 switches Also, one could implement a 1×n switch using a single n+1 port WSS in a bidirectional way, as would be apparent to a person of ordinary skill in the art

The Cantor Network

The present invention recognizes that a traditional directed cross connect can be modified to provide an undirected cross connect.

Directed Cantor Network

One well-known strictly nonblocking directed cross connect design is the Cantor network See, D. Cantor; “On Construction of Nonblocking Switching Networks,” Proc of Symp on Computer-Communications Networks and Teletraffic, 253-255 (1972) The structure of a k×k Cantor network can be viewed as follows In the middle of the network are log₂k copies of k×k Ben{hacek over (e)}s networks. See, V. E Ben{hacek over (e)}s, Mathematical Theory of Connecting Networks and Telephone Traffic. Academic Press, New York, N.Y. (1935). The i^th input (output) node of the Cantor network is connected to the i^th input (output) node of each of the log₂k copies of the center Ben{hacek over (e)}s networks.

FIG. 3 illustrates an 8×8 Cantor network 300 As shown in FIG. 3, solid nodes on the left denote input nodes 310-1 through 310-8, solid nodes on the right denote output nodes 320-1 through 320-8 and solid nodes in between represent the input and output nodes of the constituent Ben{hacek over (e)}s networks The open circles are 2×2 switches, referred to as switch nodes The triangles are 1×3 switches 330-1 through 330-8 and 340-1 through 340-8 All edges are directed from left to right.

Let C(k) be a k×k Cantor network made up of Ben {hacek over (e)}s networks B₁(k), B₂(k), . . . , B_m(k), where m=log₂ k The switch nodes in all the B_i(k)'s can be thought of as being in layers (or columns in FIG. 3 ) numbered from left to right starting at 0. Then, there are 2log₂k−1 such layers with k/2 nodes in each layer. The notation L(v) denotes the number of the layer in which v lies The B_i(k) as well as C(k) are symmetric about the middle layer of the Ben{hacek over (e)}s networks, that is about the log₂k−1 layer This middle layer has a total of (k/2)log₂k nodes It is well-known that C(k) is strictly nonblocking See, for example, G M Masson et al, “A Sampler of Circuit Switching Networks,” IEEE Computer; 5:32-48 (June 1979).

FIG. 4 illustrates an alternative 8×8 Cantor network 400FIG. 4 shows more deafly the expansion aspects of the design that is used in the analysis As shown in FIG. 4, solid nodes on the left denote input nodes 410-1 through 410-8, solid nodes on the right denote output nodes 420-1 through 420-8 The open circles are 2×2 switches, referred to as switch nodes The triangles are 1×3 switches 430-1 through 430-8 and 440-1 through 440-8 The numbers 0, 1, 2 in FIG. 4 indicate the level numbers up to the middle level (level 2) To show that C(k) is strictly nonblocking, suppose S is a valid set of demands, B is a valid routing of S and S∪{d} is also a valid set of demands Then, it is desired to show that there is a routing of the demand d=(i, j) that uses no edges of the routes already used by R The notation A(m) is used to mean the lower bound on the number of nodes at the m^th layer to which there is a path from input node i that does not use any edges used by R It is easy to see that A(0)=log₂k and inductively:

$\begin{array}{cc}\begin{array}{c}A\left(m\right)=2A\left(m-1\right)-2^{m-1}\\ =2^mA\left(0\right)-m2^{m-1}\end{array}& \begin{array}{c}\left(1\right)\\ \left(2\right)\end{array}\end{array}$

This can be seen by noticing in FIG. 4 that the number of nodes that can be reached at level m from input node i is twice the number that can be reached at level m−1 (since the out-degree of each node is two and the edges from the nodes in A(m) all go to distinct nodes at level m+1) Also, 2^m−1 of the nodes that are reachable from input node i at level m+1 if no other demands are routed might in fact be blocked by demands that have been routed. Then, at the middle layer

$\begin{array}{cc}\begin{array}{c}A\left({\log }_2k-1\right)=\frac{k{\log }_2k}{2}-\frac{\left({\log }_2k-1\right)k}{4}\\ =\frac{k({\log }_2k+1)}{4}\end{array}& \begin{array}{c}\begin{array}{c}\left(3\right)\\ \end{array}\\ \left(4\right)\end{array}\end{array}$

It can also be shown that there are at least the same number of nodes at the middle layer from which there is a path to output node j Thus, the total of both of these lower bounds is

$\frac{k({\log }_2k+1)}{2}$

and this is greater than the total number

$\frac{k{\log }_2k}{2}$

of nodes in the middle layer (level 2). This implies that there must be at least one node n in the middle layer for which there is a path from i to n and a path from n to j such that no edge of either path is used by any path in R That is, there is a valid path for the demand d=(i, j).

It should be noted that the above argument actually shows the following somewhat stronger result Let A be a subset of the input nodes and for aεA let p_a be a path from input node a to some node b at some level m_a where m_a>log₂k−1 Suppose i, j∉A Then, there are paths p_i and p_j starting at i and j respectively, and both ending at the same node at level log₂k−1 so that neither p_i not p_j shares any edge with any path p_a, where aεA. (Note that p_i and p_j might have edges in common) This can be used to show that a modified version of the Cantor network in accordance with the present invention is widesense nonblocking

Undirected Cantor Network

The present invention considers widesense nonblocking cross connects rather than strictly nonblocking cross connects Consider the Cantor network 300, 400 of FIGS. 3 and 4. In the directed case, there is a unique directed path from any input node to any given node in the middle layer and similarly from any middle layer node to any given output node Thus, the only real choice is which middle layer node a demand should be routed through Once that middle layer node has been chosen, the path is unique However, in the undirected case, there are paths from an input node to a middle layer node that could move back and forth and use up most of the edges in the graph Thus, allowing arbitrary routings tends to be much more difficult in the undirected setting

FIG. 5 illustrates an exemplary 8×8 undirected Cantor network 500 in accordance with an embodiment of the present invention, where the number of leaf nodes 510-1 through 510-8 is A=8 The open circles are 2×2 switches, referred to as switch nodes The triangles are 1×3 switches In general, the construction is just the half of a directed version of a Cantor network 400 from the input nodes 510 to the middle layer nodes 520-1 through 520-12 (level 2). The middle layer nodes 520 are marked with dashed circles in FIG. 5 The input or output nodes of the Cantor network 500 and those of the constituent Ben{hacek over (e)}s networks are not shown The undirected Cantor network 500 is widesense nonblocking for symmetric demands

FIG. 6 is a flow chart describing an exemplary implementation of a routing algorithm 600, Alg, incorporating features of the present invention. Suppose S is a valid set of demands where P is a valid routing of S as computed by Alg Let d−{i, j} be a demand such that S∪{d} is valid (step 610) Find a middle layer node n during step 620 such that there is a path p_i between i and n and a path p_j between j and n where p_i and p_i share no edges of the routes of P Also, Alg is restricted during step 630 so that p_i (and p_j) so that the layer number of nodes increases as the path moves along p_i (and p_i) from i (and j) to n. During step 640, n′ is the first node starting from i that path p_i shares with path p_j Then, during step 650, the subpath p′_i of p from i to n′ is concatenated with the subpath of p′_j of p_j from n′ to j. This path is p_d It can be shown that an undirected Cantor network is widesense nonblocking using algorithm Alg

The route chosen for the demand {i, j} by Alg consists of edge disjoint paths from i and j to some common node along p_i and p_j This common node might be n or it might be some node at some lower level. Thus, C(k) is widesense nonblocking.

FIGS. 7 and 8 illustrate simplified versions 700, 800 of the 8×8 undirected Cantor network 500 of FIG. 5 The open circles are 2×2 switches, referred to as switch nodes. The triangles are 1×3 switches. It should be noted that the middle level nodes of FIG. 5 ate not really necessary since they are degree two That is, the middle level nodes can be replaced with simple connections 720, as shown in FIG. 7 Then, the 12 nodes at the final layer 730 in FIG. 7 that are connected by two parallel links can be merged as pairs into six degree-4 nodes 830, as shown in FIG. 8 The complexity of the undirected Cantor network 500, 700, 800 can be viewed as follows Each initial triangle node is a 1×log₂k switch which can be constructed from a single 1+log₂k WSS as mentioned above The design of′ degree-4 nodes can be used in each of the nodes in the final stage of the undirected Cantor network. The other nodes in the design can each be realized using a single hybrid-4 node as described above There are k log₂k such nodes in each of log₂k−2 levels and k log₂k/2 at the final level. Thus, the total complexity of the disclosed undirected Cantor networks is roughly k log₂²k

Clos Cross Connect

Consider a design based on a traditional directed strictly nonblocking Clos cross connect

Directed Clos Cross Connect

FIG. 9 illustrates a conventional (3-stage) Clos cross connect design 900 for k input ports 910-1 through 910-k and k output ports 920-1 through 920-k Again, the actual graph is considered to be a directed graph with edges directed from left to right It is well-known that if m, the number of center stage switches 915 is at least 2 p−1, then the directed Clos cross connect is strictly nonblocking

Undirected Clos Cross Connect

As with the Cantor network, the present invention recognizes that a widesense nonblocking undirected cross connect can be constructed with k ports based on the Clos design. FIG. 10 illustrates an undirected Clos Cross network 1000 with k ports, in accordance with the present invention This general design has been investigated in early work that was interested in determining the minimum number of second stage switches for a rearrangeably nonblocking design See, for example, W Kabacinski, “Comment: Two-stage Non-Blocking Bidirectional Switch Networks,” Electronics Letters, 25(17):1198 (August 1989) and P Raby and P Banks, “Two-Stage Nonblocking Bidirectional Switch Networks,” Electronics Letters, 24(6):362-363, March 1988. However, they only allowed demands between ports on different first stage switches whereas the present invention allows such demands plus demands where both ports are on the same first stage switch.

The k ports are connected to n first stage p×m switches 1010 The second stage undirected switches 1020 with n ports can be implemented using, for example, the undirected Cantor network 500, 700, 800 described above Thus, it is assumed that the second stage switches 1020 are all widesense nonblocking

The first stage switches 1010 ale hybrid cross connects. That is, if A is the set of p ports on the left side and B is the set of m ports on the right side, then any set of demands must be supported where demands have either both end ports in A or one in A and one in B Given that the second stage switches 1020 are widesense nonblocking, then if the hybrid cross connects at the first stage 1010 are strictly or widesense nonblocking and m≧2 p−1 then the resulting undirected Clos Cross connect will be widesense nonblocking

If p=2 and m=3, then a first stage switch 1010 can be built, as indicated in FIG. 11FIG. 11 illustrates an exemplary implementation of a 2×3 switch 1100 As shown in FIG. 11, the black circles 1130 indicate optical couplers (i e, power combiner/splitters) and the rectangles 1110, 1120 represent 5-port WSSs using the notation described above. Thus, the total complexity of the k/2 first stage switches 1110, 1120 would be k/2 If undirected k/2×k/2 Cantor cross connects were used for the three second stage switches 1130, each with roughly k log₂²k/2 complexity, then the total complexity would be on the order of 3 k log₂²/2.

System and Article of Manufacture Details

As is known in the art, the methods and apparatus discussed herein may be distributed as an article of manufacture that itself comprises a computer readable medium having computer readable code means embodied thereon The computer readable program code means is operable, in conjunction with a computer system, to carry out all or some of the steps to perform the methods or create the apparatuses discussed herein The computer readable medium may be a recordable medium (e g, floppy disks, hard drives, compact disks, memory cards, semiconductor devices, chips, application specific integrated circuits (ASICs)) or may be a transmission medium (e.g, a network comprising fiber-optics, the world-wide web, cables, or a wireless channel using time-division multiple access, code-division multiple access, or other radio-frequency channel) Any medium known or developed that can store information suitable for use with a computer system may be used The computer-readable code means is any mechanism for allowing a computer to read instructions and data, such as magnetic variations on a magnetic media of height variations on the surface of a compact disk.

The computer systems and servers described herein each contain a memory that will configure associated processors to implement the methods, steps, and functions disclosed herein The memories could be distributed or local and the processors could be distributed or singular The memories could be implemented as an electrical, magnetic or optical memory, or any combination of these or other types of storage devices Moreover, the term “memory” should be construed broadly enough to encompass any information able to be read from or written to an address in the addressable space accessed by an associated processor With this definition, information on a network is still within a memory because the associated processor can retrieve the information from the network

It is to be understood that the embodiments and variations shown and described herein are merely illustrative of the principles of this invention and that various modifications may be implemented by those skilled in the art without departing from the scope and spirit of the invention

Claims

1. A method for touting a demand d={i, j} over a path pd between ports i and j in a multi-layer network having one or more wavelength selective switches, said method comprising the steps of: determining a middle layer node n in said multi-network for which there is a first path pi between said port i and said middle layer node n and a second path pj between said port j and said middle layer node n;identifying a node n′, wherein said node n′ is a first node starting from port i that path pi shares with path pj; andconcatenating said path pd, comprised of a subpath p′i of said first path pi from said port i to said node n′ with a subpath of p′i of said second path pj from said node n′ to said port j.

2. The method of claim 1, wherein said first and second paths pi and pj share no edges of a route.

3. The method of claim 1, wherein a layer number of nodes increases along said first path pi from said port i to said node n

4. The method of claim 1, wherein a layer number of nodes increases along said second path pj from said port j to said node n

5. The method of claim 1, wherein a demand touted in accordance with said method is widesense non-blocking.

6. An undirected Cantor network, comprising: k ports;at least one layer of switch nodes, wherein said switch nodes are wavelength selective switches; andk switches for connecting said k ports to said at least one layer of switch nodes

7. The undirected Cantor network of claim 6, wherein said k switches are 1×3 switches.

8. The undirected Cantor network of claim 6, wherein said undirected Cantor network comprises half of a directed version of a Cantor network

9. The undirected Cantor network of claim 6, wherein one or more nodes in a middle layer are replaced by one or more connections.

10. The undirected Cantor network of claim 9, wherein one or more nodes in a final layer that are connected by two parallel links ate merged as pairs into degree-4 nodes

11. The undirected Cantor network of claim 6, wherein said wavelength selective switches comprise node port connection means for providing reciprocal connectivity between each of N node ports and terminals of a wavelength selective switch, so that each switch state activated by a control signal provides a set of at least one unique node port-pair connections

12. The undirected Cantor network of claim 6, wherein said undirected Cantor network is widesense non-blocking

13. An undirected Clos cross connect, comprising: k ports;n first stage p×m switches connected to said k ports; anda plurality of second stage undirected switches, wherein one or more of said undirected switches are undirected Cantor networks having at least one switch node that is a wavelength selective switch

14. The undirected Clos cross connect of claim 13, wherein said first stage switches ate hybrid cross connects.

15. The undirected Clos cross connect of claim 13, wherein said wavelength selective switch comprises node port connection means for providing reciprocal connectivity between each of N node ports and terminals of said wavelength selective switch, so that each switch state activated by a control signal provides a set of at least one unique node port-pair connections

16. The undirected Clos cross connect of claim 13, wherein said undirected Clos cross connect is widesense non-blocking

17. An apparatus for routing a demand d={i, j} over a path pd between ports i and j in a multi-layer network having one or mote wavelength selective switches, the apparatus comprising: a memory; andat least one processor, coupled to the memory, operative to:determining a middle layer node n in said multi-network for which there is a first path pi between said port i and said middle layer node n and a second path pj between said port j and said middle layer node n;identifying a node n′, wherein said node n′ is a first node starting from port i that path pj shares with path pj; andconcatenating said path pd comprised of a subpath pi ′ of said first path pi from said port i to said node n′ with a subpath of pj′ of said second path pj from said node n′ to said port j

18. The apparatus of claim 17, wherein said first and second paths pi and pj share no edges of a route

19. The apparatus of claim 17, wherein a layer number of nodes increases along said first path pi from said port i to said node n.

20. The apparatus of claim 17, wherein a layer number of nodes increases along said second path pj from said port j to said node n

21. An undirected cross connect node comprised of at least a first stage of directed wavelength-selective elements and at least one additional stage of undirected wavelength-selective switch elements.