WO2003073138A2

WO2003073138A2 - Systems and methods for optical networking

Info

Publication number: WO2003073138A2
Application number: PCT/US2003/005939
Authority: WO
Inventors: Jayant Bansal; Siddharth Tandon
Original assignee: Einfinitus Technologies, Inc.
Priority date: 2002-02-26
Filing date: 2003-02-26
Publication date: 2003-09-04
Also published as: WO2003073138A3; AU2003219924A8; AU2003219924A1

Abstract

The invention teaches systems and methods for dynamically routing data in an all Optical network (Fig. 5A), in a manner that responds to changes in the network topology in real time. The invention can evolve in immediate response to changes in the network characteristics, non-limiting examples of which include changes in traffic, increase/decrease in number of network users, evolving QoS parameters, or changes in link capacities. The dynamic routing enabled by the invention supports dramatic improvements in service levels and the number of users which may be supported on the network (Fig. 5A).

Description

SYSTEMS AND METHODS FOR OPTICAL NETWORKING

BRIEF DESCRIPTION OF THE INVENTION

The present invention relates to the field of communication networks, and more particularly to systems and methods of dynamically setting up and tearing down lightpaths in an all optical network.

BACKGROUND OF INVENTION

The setting up and tearing down of light paths is commonly referred to as the Routing and Wavelength Assignment (RWA) problem. The RWA problem can be stated as follows: Given a network topology, a set of end to end lightpath requests, determine a route and wavelength(s) for the requests using the minimum possible number of wavelengths. Generalized Multi Protocol Label Switching (GMPLS) provides the mechanism for an all optical network to solve the RWA problem using techniques developed in the Multi Protocol Label Switching (MPLS) and Multi Protocol Label Switching - Traffic Engineering (MPLS-TE) protocols.

GMPLS is indeed based on the Traffic Engineering (TE) extensions to MPLS, a.k.a. MPLS-TE. This is because most of the technologies that can be used below the PSC level require some traffic engineering. The placement of Label Switched Paths (LSPs) at these levels needs in general to take several constraints into consideration (such as bandwidth, protection capability, etc) and to bypass the legacy Shortest-Path First (SPF) algorithm. Note however that this is not mandatory and that in some cases a SPF routing could be applied. In order for such a constrained-based SPF routing of LSPs to happen, the nodes performing LSP establishment need more information about the links in the network than standard intra-domain routing protocols provide. These TE attributes are distributed using the transport mechanisms already available in Interior Gateway Protocols (IGPs) and are taken into consideration by the LSP routing algorithm. Optimization of the LSP trajectories may also require some external simulations using heuristics that serve as input for the actual path calculation and LSP establishment process.

Extensions to traditional routing protocols and algorithms are needed to uniformly encode and carry TE link information, and explicit routes (e.g. source routes) are required in the signaling. In addition, the signaling must now be capable of transporting the required circuit (LSP) parameters such as the bandwidth, the type of signal, the desired protection, the position in a particular multiplex, etc. Most of these' extensions have already been defined for PSC (IP) traffic engineering with MPLS. GMPLS mainly adds additional extensions for TDM, LSC and FSC traffic engineering, by staying as generic as possible. Only a very few elements are technology specific.

GMPLS extends also two traditional intra-domain routing protocols already extended for TE, i.e. OSPF-TE and IS-IS-TE. However, if explicit routing is used, the routing algorithms used by these protocols don't need to be standardized anymore since they are now used to compute explicit routes only, and are thus not used anymore for hop-by-hop routing.

In current networks, the only technique being used is explicit setting up of lightpaths. The RWA problem is solved as a two step problem, with any of the steps being performed before the other. The first step is to find a set of routes between the source and the destination. And the second step to allocate a set of wavelengths on the chosen route between the source and the destination. The fault management for GMPLS based networks is proposed to be done using the techniques specified in Link Management Protocol (LMP)

One problem with the two step procedure outlined above to solve the RWA problem is that it results in the all-optical network becoming connection oriented. As a result fewer simultaneous users can be supported on the network, as compared to a connectionless architecture. Additionally, The currently used techniques used for providing QoS for different users are not applicable to all optical networks as there is no way one can see what traffic is flowing on a particular wavelength. As a result, it is currently not possible to provide granular QoS levels for different users on an all-optical GMPLS based network.

With GMPLS currently providing for connection oriented explicitly routed paths, a 1:1 backup is maintained to provide for fault tolerance, thereby wasting wavelengths. Additionally, since a connectionless all optical architecture does not exist, there is no provision for a technique to provide re routing in case of a failure; it simply switches the traffic to the alternate backup path.

The current solutions and techniques do not exploit the full potential of the data that is obtainable from the network. As a result, network management becomes complicated and expensive. The data about network parameters is not structured and is sub optimally analyzed, leading to poorer network management. Dynamic resource allocation is a problem across the network. There exists no technique to dynamically allocate resources like wavelengths, bandwidth etc. in real time. This results in over provisioning of resources and hence poor resource utilization. The lightpath allocation technique is offline and does not take into account the changing network parameters, thereby resulting in excessive strain on the network and suboptimal performance.

GMPLS simply extends the MPLS based technique for setting up LSPs to the optical domain. It is able to do this as it assumes that explicit routing is the only technique to set up LSPs and that fits perfectly into the existing connection oriented optical paradigm. Since, no technique exists for hop by hop routing for GMPLS, and for dynamically allocating resources in all optical networks, therefore it is currently not possible to set up and tear down LSPs in real time.

Because of the connection-oriented nature of optical networks and explicit routing followed by GMPLS, service providers have to use complex provisioning systems to provision end to end lightpaths. This is time consuming, and hence it takes a substantial time to set up, as well as tear down, thus increasing the service provisioning time and reducing the spectrum of services that the service provider can offer. For example, service providers are now beginning to offer their customers wavelengths just like they offered lease lines. Virtual Private Optical networks are also emerging on the horizon. However, all these are extremely difficult to provision and are resource intensive on the network.

SUMMARY OF THE INVENTION

The invention teaches systems and methods for dynamically routing data in an all Optical network, in a manner that responds to changes in the network topology in real time. The invention can evolve in immediate response to changes in the network characteristics, non- limiting examples of which include changes in traffic, increase/decrease in number of network users, evolving QoS parameters, or changes in link capacities. The dynamic routing enabled by the invention supports dramatic improvements in service levels and the number of users that may be supported on the network.

In embodiments of the invention, the communications nodes in a network, such as an all Optical network, may be grouped into a plurality of clusters; the plurality of clusters may be ranked hierarchically, with a leader node is selected for each cluster. In some such embodiments, the leader node is selected on the basis of maximum connectivity to the other nodes of the cluster. The hierarchical rank may be based on measurements of one or more

Quality of Service (QoS) parameters for the nodes, such that in a given cluster, each node meets a stated QoS threshold. As the QoS characteristics of the network evolve, the clusters may merge or split accordingly to remain in compliance with the applicable thresholds. The QoS parameters may include such network metrics as jitter, delay, loss, and bandwidth availability.

In some embodiments, more than one QoS parameter is used to generate the cluster hierarchy. In some such embodiments, these QoS parameters are lexically ordered; in other embodiments, a weighted, normalized average of the QoS parameters may be taken.

In embodiments of the invention, data received at an ingress node is transmitted either to an egress node within the cluster, or, via the leader node, to another cluster of a different hierarchical rank. In some embodiments, leader nodes include routing tables populated with path information for each node in lower ranked clusters, as well as path information for a leader node in at least one higher ranked cluster. These and other embodiments are described in greater detail herein. BRIEF DESCRIPTION OF DRAWINGS

The nature of the present invention will become more apparent to those skilled in the art from a study of the following detailed description in conjunction with the appended claims and drawings, all of which form a part of this specification. In the drawings:

FIG. 1 is a block diagram of a communications node in accordance with an embodiment of the present invention;

FIG. 2 is a flow chart of a process for creating an MPLS network hierarchy in accordance with an embodiment of the present invention;

FIG. 3 is a flow chart illustrating a process for selecting a leader node for a cluster in accordance with an embodiment of the present invention;

FIG. 4 is a flow chart illustrating a process for creating routing tables for each node in the MPLS network in accordance with an embodiment of the present invention;

FIG. 5 A is a schematic block diagram of constellations comprising a first level of the hierarchy of constellations in accordance with one embodiment of the present invention;

FIG. 5B is a schematic block diagram of constellations comprising a second level of the hierarchy of constellations in accordance with one embodiment of the present invention;

FIG. 5C is a schematic block diagram of constellations comprising a third level of the hierarchy of constellations in accordance with one embodiment of the present

invention; FIG. 5D is a schematic block diagram of constellations comprising a fourth level of the hierarchy of constellations in accordance with one embodiment of the present invention;

FIG. 5E is a schematic block diagram of constellations comprising a fifth level of the hierarchy of constellations in accordance with one embodiment of the present invention;

FIG. 6 is a flow chart diagram generally illustrating a process for routing data in accordance with one embodiment of the present invention;

FIG. 7 is a schematic block diagram generally illustrating a process for re- executing a process to form new clusters in accordance with one embodiment of the present invention;

FIG. 8A is a schematic block diagram generally illustrating a group of constellations about to undergo a split in accordance with one embodiment of the present invention;

FIG. 8B is a schematic block diagram generally illustrating a group of constellations of the second level about to undergo a split in accordance with one embodiment of the present invention;

FIG. 8C is a schematic block diagram generally illustrating a group of constellations of the third level about to undergo a split in accordance with one embodiment of the present invention;

FIG. 8D is a flow chart diagram generally illustrating a process for executing a split in response to changes in network topology in accordance with one embodiment of the present invention; FIG. 8E is a schematic block diagram generally illustrating the group of constellations of FIG. 9 A about to undergo a split in accordance with one embodiment of the present invention;

FIG. 8F is a schematic block diagram generally illustrating the nodes of FIG. 9E after a split is performed in accordance with one embodiment of the present invention;

FIG. 8G is a schematic block diagram of constellations comprising the second level of the hierarchy of constellations of FIG. 9F;

FIG. 8H is a schematic block diagram of constellations comprising a third level of the hierarchy of constellations of FIG. 9 A;

FIG. 9 A is a schematic block diagram generally illustrating an exemplary group of constellations about to undergo a merger in accordance with one embodiment of the present invention;

FIG. 9B is a flow chart diagram generally illustrating a process for executing a merger of constellations in response to changes in network topology in accordance with one embodiment of the present invention; and

FIG. 9C is a schematic block diagram generally illustrating the result of a merger of the constellations illustrated in FIG. 9 A.

DETAILED DESCRIPTION OF THE INVENTION

The present invention teaches a new paradigm for efficiently and dynamically routing data in a communications network. The invention enables the network to evolve dynamically in immediate response to changes in network performance, thereby improving link utilization, maximizing users, and improving support levels. The dynamic routing enabled by the invention supports dramatic improvements in service levels and the number of users that may be supported on the network.

FIG. 1 illustrates one possible representation of a communications node in a network in accordance with one embodiment of the present invention. Communication node 102 is coupled to communication links 104, 106, 108, 110. Each communications link 104, 106, 108, 110 is operative to carry a predetermined transmission bandwidth of data, voice or video information. Each communications link 104, 106, 108, 110 has a respective transmission bandwidth, latency, and jitter and packet loss ratio associated with that link. Note that these parameters may be a function of time.

Generally node 102 may be either a redistribution point, or an end point (terminal node) for data transmissions. Terminal nodes, such as phone sets, computers, printer or fax machines, generate or use information transmitted over the network, or facilitate communications with other networks. Communications nodes may include networking equipment such as switches, routers, or gateways, which are operative to recognize and forward transmissions to other nodes.

In accordance with one embodiment of the present invention, the node 102 has a software agent residing in the control plane of a router, which is responsible for calculating node hierarchy and routing tables.

An all-optical network consists of optical fiber links between nodes with all optical switching and routing of signals at the nodes without electronic regeneration. In first generation optical networks, the physical layer was essentially a point to point optical fiber link and provided a full wavelength's worth of bandwidth to the layer above it. The most common examples of such networks are the SONET/SDH networks in use today. In the second generation optical networks, the physical layer incorporates more sophisticated mechanisms that can provide variable amounts of bandwidth between pairs of nodes. The ITU has defined a new layer for such networks called the optical layer. This definition is particularly appropriate to describe WDM networks. The optical layer provides lightpaths to the higher layer. A lightpath is an end to end connection established across the optical network, and uses a wavelength on each link in a path between the source and destination. A lightpath provides a full wavelength's worth of bandwidth to the higher layers.

Wave Division Multiplexing (WDM) is a technique used to increase capacity on an optical network. In WDM, the data is transmitted simultaneously over multiple carrier wavelengths over a fiber. These wavelengths do interfere with each other provided they are kept sufficiently apart. WDM transmission systems employing 32 wavelengths at 2.5Gbps each over a single fiber are commercially available today.

The invention assumes full duplex links and full duplex connections. In general, the topology if a wavelength routing network may be an arbitrary mesh. It consists of wavelength crossconnect (WXC) nodes, interconnected by fiber links. The network provides lightpaths between pairs of network nodes.

Creation of Hierarchically-Ordered Node Clusters

FIG. 2 illustrates a process 150 for creating a network hierarchy at the optical layer i.e. all those nodes and links which are connected via fibers carrying wavelength(s), in accordance with one embodiment of the present invention. The process 150 of creating a hierarchy of clusters of nodes is herein referred to as an X-Constellation algorithm, with the term "cluster" used interchangeably with the term "constellation" herein. The X-constellation algorithm process 150 begins 152 with all nodes meeting a parameter X depicting the number of available wavelengths between them, herein referred as the primary QoS parameter and any secondary QoS parameter thresholds grouped into clusters in the first tier. Clusters are composed of all nodes that are connected with each other via links satisfying the threshold X for available wavelengths and meeting any required secondary QoS thresholds. For example, consider 3 nodes A, B and C. Let

L_AB, LB_C, L_AC denote the links between A and B, B and C, and A and C respectively. Let S_AB, S_BC, S_AC denote the set of available wavelengths at a given time instant, on the three links. We start with node A and consider the intersection of the sets S_AB and SBC and S_AB and S_AC. Then if S_AB^ASB_C ^>=X or S_AB ^ΛSAC ^>==X then nodes A, B and C will come under the same cluster. This method can be made more flexible by specifying a set S of wavelengths which should be common. This input can be supplied either manually, or through a protocol built into the networking equipment. In alternative embodiments QoS parameters may include bandwidth, latency, jitter, packet loss ratio, topological area, policy settings, etc.

In a subsequent step 154 all nodes in the network falling below the first primary QoS threshold and meeting a second primary quality of service (QoS) parameter threshold are grouped in clusters of a second tier.

In the next step 156, all nodes falling below the second primary QoS threshold and meeting a third primary quality of service (QoS) parameter threshold are grouped into clusters of the third tier. In this non-limiting example, all remaining nodes falling into a fourth tier.

In a subsequent step 160, all of the clusters are arranged in an ascending hierarchy according to tier rank. In the current example, the fourth tier is above the third tier and so on down to the first tier. After arranging the clusters hierarchically, a leader node is selected for each cluster 162. In embodiments of the invention, leader nodes generally have greater communication capacity than the non-leader nodes within a particular cluster; this is described in greater detail infra.

Populating the Routing Tables

Upon selection of the leader nodes, routing tables are calculated for each node in the network 162. In embodiments of the invention, the algorithm for populating the table may include: • Calculating routing tables within the constellations of the actual network

• Calculating routing tables for the virtual network, i.e., for all levels above actual network level.

Once the network has been mapped hierarchically according to QoS parameters as described above, the routing algorithm is called by each node in network which calculates the routing table used by LMP and subsequently by GMPLS. In embodiments of the invention, the routing algorithm may also include the following:

For each node • Make adjacencies to each other node in the constellation

• Calculate the shortest path to each other node in the constellation

• Calculate the minimum number of available wavelengths and other QoS parameters over all links to reach all other nodes in the constellation, and store this information in the routing table with the respective destination entry

• Determine the leader in the constellation, and mark a leader bit to the destination entry in the routing table

Note that the shortest path may be calculated using Dijkstra's algorithm, or any other shortest path algorithm known to those skilled in the art.

In embodiments of the invention, records in a routing table generally includes: destination address of the packet; next hop; available wavelengths to the next hop; hierarchy of next hop; leader bit of next hop; and an adjacency bit. For non-leader nodes, the routing table includes the shortest path to each other node in the cluster and to every adjacent node regardless of cluster. In embodiments of the invention, the leader node's routing table also includes the shortest path to each node of all lower clusters, and the shortest path to the leader node of the next higher cluster. For example a non-leader node of the third cluster will have a routing table including the shortest path to each node of the third cluster including the leader node of the third cluster. A leader node's routing table will include the shortest path every node of the leader node's cluster and the shortest path to each node belonging to lower clusters, and the shortest path to the leader node of the next higher cluster.

Though the above examples deal with quality of service QoS thresholds based solely on the number of available wavelengths, it should be understood that virtually any other parameter may be used with the X-constellation algorithm to generate node cluster hierarchies. Examples of quality of service parameters that may be used to create cluster hierarchies include but are not limited to, bandwidth, latency, jitter, packet loss ratio, topological area, and policy settings. Other suitable QoS metrics shall be apparent to those skilled in the art.

Additionally more than one parameter may be used to generate X-Constellation hierarchies. In an embodiment of the invention, the primary quality of service parameter may be the cross product of one or more of the node parameters. For example, the primary QoS parameter used in steps 152 through step 158 may be the product of available wavelengths, bandwidth, latency and jitter. Alternatively both the primary and secondary quality of service parameters could be cross products of available wavelengths, bandwidth, latency, jitter, packet loss ratio, topological area, policy settings, etc. In some embodiments of the invention, a lexical ordering of one or more of the QoS parameters may be used to generate the hierarchy. In alternative embodiments, a normalized weighted average of one or more of the QoS parameters may be used to generate the hierarchy. Other alternative formula combining any basic parameters will be apparent to those skilled in the art, and any combination thereof may be used as a primary or secondary quality of service parameter in steps 152 through 158 of the X-Constellation algorithm.

FIG. 3 illustrates a process 162 for selecting a leader node for a cluster in accordance with another embodiment of the present invention. The process begins 164 by determining which nodes in a cluster have the most links. If more than one node in the cluster has the highest number of communications links 166, the number of available wavelengths of all the communications links is summed for each such node. The candidate node with the greatest total available wavelengths of communications links is then selected as the leader node for that cluster. 170. If more than one node in the cluster has the greatest total number of wavelengths of communications links, then by a rule of thumb, the node with the least binary value of its IP address is selected as the leader node of the cluster. In embodiments of the invention, the leader bit of the selected leader node is assigned a value of 1, signifying that the node is a leader node 172. If only one node of the cluster has the highest number of communications links 166 the node is selected as the leader node for that cluster 174. The leader bit of the selected node is assigned a value of 1 to signify that the node is the leader node for the cluster 176. All other nodes in the cluster will have a leader bit value of 0 to signify that they are not leader nodes.

FIG. 4 illustrates a process 163 for creating routing tables for each node in the network in accordance with one embodiment of the present invention. Initially, the leader bit of a node is read 502. If the node is not a leader, a routing table is calculated for the node 504. The routing table for a non-leader node will contain the shortest path from the non-leader node to each other network node that is either one hop from the current node or in the same cluster as the current node. For example, a non-leader node of a cluster of nodes will have a routing table including the shortest paths to any adjacent node and to each node of that cluster, including the leader node of that cluster. Routing table entries may further include a destination address, next hop, available wavelengths to the next hop, hierarchy of next hop, leader bit of next hop, and adjacency bit for each of these paths.

If a node is determined to be a leader node 502, a routing table is calculated for the leader node 512. This routing table may include the shortest path to each node that is one hop from the leader node, each node that is in the leader node's cluster, each node that is in a lower cluster, and the shortest path to the leader of the next higher cluster. For example, the leader of a cluster of nodes may have a routing table including the shortest paths to any adjacent node, each node of the second cluster, each node of the first cluster, and the leader node of the third cluster.

Routing Data Through the Node Clusters

FIG. 6 illustrates a process for routing data 700 in accordance with one embodiment of the present invention. An ingress node of the optical network 702 receives a lambda. An ingress node is a node communicatively coupled to systems outside the network. The lambda includes a header containing information indicating the destination. The ingress node will determine the lambda's egress node based on the destination. The ingress node will compare the egress node with the ingress node's routing tables, (which was generated in the process of FIG. 2). If the intended egress node for the lambda is in the same cluster as the ingress node (has an entry in the node's routing tables (see FIG. 2) for a non-leader ingress node) or is at a distance of one-hop from the ingress node in step 704 the process continues to step 706. At 706 the ingress node transmits the lambda to the egress node by a method of lambda switching. The new lambda being created from the information base created in the process of FIG. 4.

If the lambda is intended to leave the optical network, the egress node transmits the lambda to a destination outside the network in step 708, or the egress node may use the lambda in some manner. If the ingress node determines that the egress node is not within the ingress node's cluster 704

(for e.g., if the ingress node is not a leader node, and there is entry in its routing table for the egress node), the ingress node transmits the lambda to the leader node of the ingress node's cluster 710.

If the egress node is in a cluster below the cluster of the leader node 712, the leader node makes this determination by consulting its routing table 714. The leader node subsequently switches the lambda to the egress node. As discussed with reference to FIG. 2, the leader node has routing table entries for every node of every lower cluster. In step 716 the egress node uses the lambda or forwards it to a final destination beyond the optical network.

If the egress node is not below the cluster of the leader node 712, the QoS demand for the lambda is compared to the QoS values of the next higher level 717. If the next higher level of the optical network meets the quality of service requirement for the packet, the leader node switches the lambda to the leader node of the next higher cluster 718. If the leader node is also the leader node of the above cluster, it simply keeps the lambda.

The leader node that receives the packet determines whether the egress node is in the receiving leader node's cluster or a lower cluster 720. If the egress node is in the current cluster or a lower cluster, the leader node will switch the lambda to the egress node 724. The egress node either utilizes the lambda, or proceeds to forward it out of the network 726.

If the egress node is not in the receiving leader node's cluster or a lower cluster 720, the process determines whether a higher cluster exists 722. If a higher cluster exists, the leader node switches the lambda to the leader node of the next higher cluster 718. If no higher clusters exist 722, the lambda is dropped 728. This occurs because there is no viable path to the egress node. In embodiments of the invention, a detailed error message may be returned in response.

Thus lambda will continually travel to higher and higher clusters until they reach a leader node that knows a path to the egress nodes. This greatly increases the efficiency of routing in all optical networks by reducing the information management required by any single node when compared to a purely hop by hop routing method. ι

For example, a lambda entering the system at node 2 (FIG. 5 A) with a final destination of node 6 (or destination at some point outside the MPLS network via node 6), will be transmitted to node 1 (the lead node of the first cluster 602) 710. Since node 6 is not in a cluster below node 1, the QoS of the second level of the network is compared to the lambda's QoS requirements 717.

If the QoS requirement is met, the packet is sent to the leader node of the next higher level above the current leader node 718. In the present example, since node 1 is the leader above node 1 at the second level, the packet remains at node 1, but node 1 is considered to be at the second level. Node 1 checks its routing tables to determine if node 6 is below it 720. Since node 6 is below node 1 at the second level, node 1 switches the lambda to node 6. The routing tables of node 1 will contain a shortest path to node 6, since node 6 is on a lower level than lead node 1 at the second level.

Evolution of Clusters FIG. 7 illustrates a process at 800 for re-executing the X-constellation process of FIG. 2 in order to form new clusters in accordance with one embodiment of the present invention. The optical network initially executes the X-constellation algorithm 802 as illustrated in FIG. 2. A new routing table is populated according to the X-Constellation algorithm. Upon receipt of a lambda 804 the lambda travels through the network 808 via a path. The sum of the wavelengths used along this path is then determined. If the sum of wavelengths used, across the optical network, exceeds a predetermined threshold, the X-constellation grouping algorithm of FIG. 2 is re- executed 812, with the current capacities of nodes and links within the optical system used as input.

In embodiments of the invention, once a flow is allocated on a particular path, the dynamic unused wavelength entry corresponding to each link of the path is updated to reflect the change in the number of used wavelengths over that path, caused by allocating the flow. If the change in number exceeds the threshold value, then re-clustering is undertaken. Thus, with allocation and de-allocation of demands, existing clustering may split or merge.

Figs 8A-8H illustrate an example of an optical network hierarchy undergoing a split in accordance with an embodiment of the present invention. FIG. 8 A illustrates an initial group of constellations about to undergo a split 920. The nodes are connected via communications links; those links illustrated in FIG. 8 A without a specified transmission capacity have at least 100 unused wavelengths. The first level of constellations include constellations 922, 924, 926, 928, 930, 932 and 934. In this example, a node 14 has been selected as the leader node of the first constellation 922.

Requests for wavelength allocation may occur between two nodes either within a constellation or between two nodes of different constellations. Allocations occurring within a constellation may result in a split. An allocation of wavelength on a link, wherein the remaining number of unused wavelengths still meets the level number, does not result in a merger. The change is simply propagated to the nodes of the present constellation as well as in the tree of all leader nodes presiding over the link undergoing allocation. For example, link 936 is allocated 10 wavelengths of data transmission, leaving 110 wavelengths. Since the capacity of link 936 still exceeds the 100 threshold of the first level, no split is required. The new number of unused wavelengths is communicated to node 14 (the leader of the present constellation), the leader node 14, all nodes within the present constellation, and to any leader nodes in linked constellations which are higher in the hierarchy. FIG. 8B is a schematic block diagram of constellations at 940 comprising a second level of the hierarchy of constellations of FIG. 8 A. Second level constellations are comprised of nodes with links having a number of unused wavelengths of less than 100 and at least 75. A first constellation 942 of the second level includes nodes 5, 10, 12 and 14, with node 14 being the leader of the constellation 942. As is shown in FIG. 8B, nodes 7 and 8 are not members of constellations at the second level of the hierarchy of constellations because they have no links with unused wavelengths between 100 and 75, though they are members of constellations at level one (FIG. 8A). Instead nodes 7 and 8 are represented as independent nodes at the second level.

FIG. 8C is a schematic block diagram of constellations at 946 comprising a third level of the hierarchy of constellations of FIG. 8 A. Third level constellations include all nodes with links having an unused wavelength capacity of less than 75 and at least 50. A first constellation 948 of the third level includes nodes 7 and 14, with node 14 being the leader of the constellation. Only a single constellation of the third level exists in this exemplary embodiment.

FIG. 8D illustrates a process 850 for re-executing the X-constellation process in order to split constellations in response to changes in network topology in accordance with one embodiment of the present invention. A set of transmission wavelengths 852 is allocated to a link 936 within the optical network. The remaining unused wavelengths of the link 936 are compared to the wavelength requirement of the link 854. If the resulting unused wavelengths retained by the link are insufficient for the level the link is currently on, an alternative path is checked 856. If no alternative path of sufficient wavelengths is available, a split is executed, creating two new constellations in step 858 separated by the link 936. New leaders are then selected for each new constellation as described with regard to FIG. 3. The routing tables of all nodes of both new constellations and the leaders above each constellation are modified 860 to reflect the change in the network. If, the resulting number of wavelengths of link 936 still qualified for the link's current level within the hierarchy of levels 854, the routing tables of all nodes within the constellation to which the link 936 is contained would be modified 862 to reflect the reduced wavelength capacity of link 962. The routing tables of the leader nodes above the constellation level would also be modified to reflect the reduced wavelength capacity of link 962. Similarly, if an alternate path exists 856, the routing tables are modified accordingly 862. Note that though the process of FIG. 8D refers solely number of wavelength requirements, the quality of service requirements may include many other parameters or a combination of parameters, as has been discussed previously herein and will be apparent to those skilled in the art.

FIG. 8E illustrates the group of constellations of FIG. 8A about to undergo a split at 949 comprising a first level of an optical network hierarchy in accordance with one embodiment of the present invention. The nodes are connected via communications links. Links illustrated in FIG. 8 A without a specified transmission capacity have at least 100 available wavelengths.

Nodes of a first constellation 922 are connected contiguously via links of at least 100 unused wavelength capacity. As illustrated, link 936 has a reduced unused wavelength capacity of 80. Since no alternate route with at least 100 unused wavelengths exists between node 9 and node 1, a split is performed. FIG. 8F illustrates the nodes of FIG. 8E after a split is performed in accordance with one embodiment of the present invention. Constellation 922 from FIG. 8E is partitioned into constellations 951 and 952. Constellation 951 includes nodes 9, 13 and 14. Constellation 952 includes nodes 1, 2, 3 and 4. New leaders are selected for constellations 951 and 952. Node 1 is selected for constellation 952 and node 14 is selected for constellation 951. The routing tables of each member of constellations 951 and 952 are modified to reflect the topology of the new constellations.

FIG. 8G is a schematic block diagram of constellations 970 comprising the second level of the hierarchy of constellations of FIG. 8F. Second level constellations are comprised of nodes with links having a wavelength capacity of less than 100 and at least 75. A first constellation 972 of the second level includes nodes 1, 5, 10, 12 and 14, with node 14 being the leader of the constellation 942. The routing tables of the leader node 14 of constellation 972 are modified to reflect the changes in network topology caused by the split of FIG. 8F.

On any level above the first level, the links depicted may represent a path comprising many individual links and nodes. For example, link 936A is in fact a virtual link comprised of multiple links with the stated minimum bandwidth capacity.

FIG. 8H is a schematic block diagram of constellations 946 comprising a third level of the hierarchy of constellations of FIG. 8A. Third level constellations in this example are comprised of nodes with links having a wavelength capacity of less than 75 and at least 50. A first constellation 948 of the third level includes nodes 7, 9 and 14, with node 14 being the leader of the constellation 942.

Whenever there is a split or merge of clusters, entries are created or destroyed from the tables of a router. For instance, when the constellations merge, new paths are formed within the constellation by setting up new LSP's and entries added to the routing table. When there is a split, LSP's are destroyed and entries are deleted from the routing tables. Thus, the routing tables keep changing with allocation and de-allocation of demands.

FIG. 9 A illustrates an example group of constellations 1000 that are about to undergo a merger. Constellations 1002 and 1004 are connected via links 936, 1006 and 1008. Link 936 has an initial wavelength capacity of 80. As an illustrative example, traffic originally assigned to link 936 is de-allocated resulting in link 936 having a wavelength capacity of 120. As wavelengths are generally freed when transmissions of previously assigned data are completed, and the number of unused wavelengths of link 936 would exceed the threshold of the first level hierarchy (100), this would result in a merger between constellations 1002 and 1004

FIG. 9B illustrates a process 1050 for re-executing the X-constellation process in order to merge constellations in response to changes in network topology in accordance with one embodiment of the present invention. At the outset, the wavelengths of link 936 are de-allocated 1052. In step 1054 the resulting number of unused wavelength of the link 936 is compared to the threshold of the current level of the link. If the resulting wavelength capacity retained by the link is sufficient for a lower level within the hierarchy of levels, a merger is executed 1056, combining the constellations at each end of link 936 in order to form a new constellation. A new leader is then selected for the new constellation 1058. The routing tables of all nodes of the new constellation and the leaders above the new constellation are then modified 1060 to reflect the change in the network.

If the resulting number of unused wavelengths of link 936 does not qualify for a lower level within the hierarchy of levels 1054, then the routing tables of all nodes within the constellation to which the link 936 is contained would be modified 1062 to reflect the reduced available wavelength capacity of link 962. The routing tables of the leader nodes above the constellation level would also be modified to reflect the increased wavelength capacity of link 936.

FIG. 9C illustrates the result of constellations 1002 and 1004 merging at 1090 to form constellation 1092 due to the deallocation of wavelengths originally being used in link 936, leaving link 936 with a wavelength capacity of 120. The merger process is performed in much the same way as the splitting process of FIG. 8. After the new constellation 1012 is formed, the routing tables of each node of the constellation 1092, and the leader nodes directly above constellation 1092 are modified to reflect changes in network topology. In addition to the above mentioned examples, various other modifications and alterations of the invention may be made without departing from the invention. Accordingly, the above disclosure is not to be considered as limiting and the appended claims are to be interpreted as encompassing the true spirit and the entire scope of the invention, and is defined by the following claims.

What is claimed is:

Claims

1. A method for dynamically routing data in an all-optical network, the network including a plurality of nodes, each node operative to receive and transmit data, the method comprising: grouping the nodes into a first plurality of clusters, wherein each cluster has a hierarchical rank based on a measurement of one or more QoS metrics for each of the first plurality of clusters; selecting a leader node for each cluster in the first plurality of clusters, wherein the leader node is coupled at least to a second cluster in the first plurality of clusters; receiving data in the form of wavelengths (lambda) at an ingress node, wherein the ingress node is a member of one of the clusters; switching the lambda to the leader node of the cluster associated with the ingress node; from the leader node, switching the lambda to an egress node in the optical network; in response to transmitting the data to the egress node, taking a second measurement of the one or more QoS metrics for the plurality of nodes; regrouping the plurality of nodes into a second plurality of clusters, such that the second plurality of clusters has a hierarchical rank based on the second measurement of the one or more QoS metrics.

2. The method of claim 1, wherein the one or more QoS metrics include one or more of the group consisting of unused wavelengths between nodes, jitter, delay, loss, bandwidth, packet loss ratio.

3. The method of claim 2, wherein the one or more QoS metrics are arranged as a vector.

4. The method of claim of 2, wherein the one or more QoS metrics are ordered lexically.

5 The method of claim 2, wherein the hierarchical ranks based on the first and second measurements of the one or more QoS metrics are based on a weighted average on the one or more QoS metrics.

6. The method of claim 2, wherein the selecting the leader node for each cluster in the first plurality of clusters further includes determining a most-linked node for each cluster in the first plurality of clusters.

7. The method of claim 2, wherein the data includes a threshold for the one or more QoS metrics.

8. The method of claim 2, wherein regrouping the plurality of nodes into a second plurality of clusters further includes merging two or more clusters from the first plurality of clusters.

9. The method of claim 8, wherein the merging to the two or more clusters is in response to deallocating wavelengths within the two or more clusters.

10. The method of claim 2, wherein regrouping the plurality of nodes into a second plurality of clusters further includes splitting one or more clusters from the first plurality of clusters.

11. The method of claim 10, wherein the splitting one or more clusters from the first plurality of clusters is in response to allocating wavelengths within the one or more clusters.

12. The method of claim 2, wherein regrouping the plurality of nodes into a second plurality of clusters further includes selecting a leader node for each cluster in the second plurality of clusters.

13. The method of claim 2, wherein grouping the nodes into the first plurality of clusters further includes populating a routing table for each node in the network.

14. The method of claim 13, wherein the routing table includes a path to a leader node associated with a next higher cluster in the hierarchy of clusters.

15. The method of claim 13, wherein regrouping the nodes into the second plurality of clusters further includes populating a revised routing table for each node in the network.

16. The method of claim 2, wherein each node in the network includes at least one of a router, a switch, and a terminal.

17 The method of claim 2, wherein the optical network includes one or more internetworks.

18. A lambda-switched communications network comprising: plurality of communications nodes^'; a plurality of communications links connecting the plurality of communications nodes; plurality of node clusters, each cluster including one or more nodes from the plurality of communications nodes, wherein the plurality of node clusters are arranged in a hierarchical order defined by one or more QoS parameters, such that for each node cluster, the one or more nodes in the node cluster meet one or more thresholds for the one or more QoS parameters; wherein the plurality of node clusters gain or lose nodes in response to variations in the one or more QoS parameters.

19. The lambda-switched network of claim 18, wherein each cluster of the plurality of node clusters includes a leader node, wherein the leader node has a greatest number of links in the node cluster.

20. The lamb da- switched network of claim 19, wherein for each cluster, the leader node is in communication with at least one other cluster from the plurality of node clusters.

21. The lamb da- switched network of claim 20, wherein each node includes a routing table.

22. The lambda-switched network of claim 21, wherein for each leader node, the routing table includes a path to all other nodes in the node cluster.

23. The lambda-switched network of claim 22, wherein for each leader node, the routing table includes routing information for all node clusters from the plurality of node clusters which have lower rank in the hierarchical order.

24. The lamb da- switched network of claim 18, wherein the one or more QoS parameters are from the group consisting of latency, jitter, loss, available bandwidth.

25. The lambda-switched network of claim 24, wherein the one or more parameters are arranged as a vector.

26. The lambda-switched network of claim 24, wherein the one or more QoS parameters are arranged in lexical order.

27. The lambda-switched network of claim 24, wherein the one or more QoS parameters are translated to a weighted average.

28. The lambda-switched network of claim 18, wherein the traffic is routed between the nodes via the links via Generalized Multi Protocol Label Switching.

29. The lambda-switched network of claim 18, wherein the packet-switched network is at least partially an all-optical network.

30. A method of routing a packet in a network, the network including a plurality of nodes arranged in a hierarchical plurality of node clusters, the method comprising: at a first node in the network, determining an egress node for the packet, wherein the first node belongs to a first node cluster from the plurality of node clusters; determining whether the egress node is within the first node cluster; if the egress node is not in the first node cluster, transferring the packet to a leader node in the first node cluster; if the egress node resides in a node cluster which is ranked lower in the hierarchical plurality of node clusters, forwarding the packet to the lower-ranked node cluster; if the egress node resides outside the first node cluster and the egress node does not reside lower in the hierarchical plurality of node clusters, determining a QoS threshold for the packet; if the QoS threshold is met by a node cluster ranked higher in the hierarchical plurality of node clusters, forwarding the packet to a lead node of a next higher node cluster in the plurality of node clusters.

31. The method of claim 30, wherein the QoS threshold is contained within a header for the packet.

32. The method of claim 30, wherein the plurality of nodes includes a plurality of routers, switches, and terminals.

33. The method of claim 30, wherein the network supports Multi Protocol Label Switching.

34. The method of claim 30, wherein the QoS tlireshold measures one or more of the parameters consisting of delay, jitter, loss, available bandwidth.

35. The method of claim 34, wherein the QoS threshold is a vector representation of the one or more of the parameters.

36. The method of claim 35, wherein the hierarchical plurality of node clusters is arranged according to the one or more parameters.

37. The method of claim 36, wherein the one or more parameters are in lexical order.

38. The method of claim 30, wherein the leader node includes a routing table, the routing table including a path for each node in the first cluster.

39. The method of claim 38, wherein the routing table includes a path for each node in the node cluster which is ranked lower in the hierarchical plurality of node clusters.

40. The method of claim 39, wherein forwarding the lambda to the lower-ranked node cluster further includes identifying the egress node in the routing table for the leader node in the first cluster.

41. A method of populating a routing table for a communications node in a network, wherein the network is organized as a plurality of hierarchically-arranged node clusters, such that the node is a member of a node cluster in the plurality of node clusters, the method comprising: determining a shortest path to each node in a plurality of other nodes within the cluster; entering the shortest path to each of the plurality of other nodes within the cluster in the routing table; determining a minimum bandwidth link for each node in the plurality of other nodes in the cluster; entering the minimum bandwidth link in the routing table; selecting exactly one leader node for the cluster, selecting the leader node further including selecting a node from the node cluster with a greatest number of links to the plurality of other nodes; in the leader node, determining a shortest path to each node in a plurality of node clusters of lower rank, and entering the shortest path to each node in the lower ranked clusters in the routing table

42. The method of claim 41, wherein selecting exactly one leader node further includes: if there is more than one node in the node cluster with the greatest number of links, summing the minimum bandwidth link for each of the more than one nodes.

43. The method of claim 41, further including: in the leader node, calculating a shortest path to a leader node of a node cluster of next highest rank.

44. The method of claim 41, wherein determining the shortest path to each node in a plurality of other nodes within the cluster further includes applying Dijkstra's Shortest Path algorithm.

45. The method of claim 41 , wherein the network is at least partially a GMPLS-based network.

46. The method of claim 41, wherein the plurality of hierarchically arranged node clusters are ranked according to one or more QoS parameters for the network. 4 / . 1 ne metnod of claim 41 , wherein the QoS parameters include one or more of the group consisting of jitter, delay, loss, available bandwidth, available wavelength