Optical communications – Multiplex – Optical switching
Reexamination Certificate
1999-05-18
2004-06-08
Pascal, Leslie (Department: 2633)
Optical communications
Multiplex
Optical switching
C398S043000, C398S058000, C370S395410, C370S395210, C370S409000
Reexamination Certificate
active
06748170
ABSTRACT:
TECHNICAL FIELD OF THE INVENTION
The invention relates to a method for solving optimisation problems occuring during the design of a mostly optical ATM network.
DESCRIPTION OF RELATED ART
The technique of finding new solutions for designing networks based on Asynchronous Transfer Mode (ATM) or on All Optical Networks (AON) is a rapidly growing area of research in telecommunication and computer business. Both ATM and AON have own methodologies for simplifying network management and switching. Whereas ATM networks apply the VP (Virtual Path) concept, i.e many connections which have a common path in a part of the network are handled (managed, switched) together, the AONs assign the wavelengths to connections in such a way that connections of the same wavelength are handled together without any electrical processing during the transmission. Optical networks employ Wavelength Division Multiplexing (WDM) and specially designed algorithms allow a high aggregate system capacity due to spatial reuse of wavelengths.
A known idea, (Mukherjee, Ramamurthy and Banerjee, “Some principles for designing a wide-area optical network” IEEE 1994, pp. 110-119), is to combine the technology of AONs with ATM technique. By using optical amplifiers and switches in a mostly optical ATM network, the number of opto-electrical conversions can be reduced significantely, which leads to better signal quality and lower delays, but from the point of view of ATM there is a problem: the VPI/VCI modification in the cell header becomes impossible. The Wavelength Division Multiplexing (WDM) technique offers a good utilization of the fibre but the transmission capacity is divided into equal parts (channels) defined by the wavelengths used. Because of this partitioned capacity the statistical multiplexing (SM), which is another benefit of the ATM, has limits. SM can shortly be described as the gain accomplished when dynamically multiplexing together channels with bursty traffic into one channel, thereby eliminating those empty timeslots which may occur in the channels due to the burstiness. The larger the capacity, the lower the effective bandwith, therefore the resource usage is more efficient.
Combining the technology of AONs with ATM technique entails a loss of performance due to the loss of SM. The total effect though, is a gain due to the enhanced performance coming from the simplified VP management.
A use of this combined technique implies both advantages and drawbacks. The drawbacks are that the switching must be done on three different levels (instead of two), namely: 1. Pure optical switching for VPLs; 2. VP switching (Cross connects) where the switching is done on the basis of VPIs (VPIs are changed); 3. VC switching. The design procedure is computationally hard as the complexity is about the number of available wavelengths to the power of the number of nodes in the network. The main advantage though, is that many opto-electrical conversions can be avoided so that less CCs are required, thereby enhancing the performance.
There is a problem in how to make the architecture of systems like this optimal, i.e when designing backbone networks for example, one must assume full logical connectivity between the nodes in the network and try to find optimisation algorithms for best using the limited number of wavelengths to accomplish a high aggregate system capacity due to spatial reuse of wavelengths.
The architecture used in the IEEE-paper is a combination of well-known “singlehop” and “multihop” approaches as employed in many other WDM networks/proposals. The paper deals with optimising the cost, efficiency and throughput of the routing node by theorethical analysis. The optimisation is here split into two steps with the objective of minimising the average message delay caused by transfer: 1. They find and map a virtual topology into the given physical; and 2. Assign the wavelengths to links of the virtual topology. This technique brings about that the VP design is already performed in step one, meaning that the optimality criteria could be deteriorated and the delays could increase since if there are more end-to-end streams on a physical link than the allowed number of wavelengths, the paper proposes a rerouting of one of these streams. Hence, the average delay is increased as the wavelength assignment and the virtual topology forming are divided.
SUMMARY
Combining ATM networks with an All Optical Network (AON) is a rather new technique with drawbacks and advantages. Here we try to design the virtual topology of an optimal ATM network on top of a AON. The optimality criterion refers here to making the ATM network “as optical” as possible, meaning that we want as long lightpaths as possible in the network without having to change the wavelength, i.e the total number of necessary wavelength changes in the network should be as few as possible, thereby minimising the need for the lightpaths to enter Crossconnects (CCs). This idea solves the problem by, and discloses a method for, finding an optimal routing of the VPCs so that the amount of VP switching can be maximally reduced. In ATM networks is an end-to-end connection associated with a VCC (Virtual Channel Connection). One or more VCCs are established on a VP Connection (VPC) which is built of one or more VP Links (VPL) along the path of the VCC. The VPL consists of one or more VPs concatenated. Hence, the VP is a “bundle” of VCCs along one physical link.
Along a VPL the VPI/VCI is not changed. Therefore no electrical processing is needed along the VPL, which also is the objective with the invention. Since no electrical processing is needed along a VPL, the same wavelength can be assigned to all of its components, i.e VPs. In this case the VPCs using these VPLs will not have to enter in the Cross Connects (CCs). They can be switched by passive optical switches. Since the limit on the processing capacity of CCs is dependent upon the internal bus speed, applaying an ATM network on top of an AON realise the use of much larger networks. The objective is to carry the cells by the virtual topology as far as possible in the optical domain. Packet forwarding from lightpath to lightpath is performed via electronic switching when required.
When we design a backbone network, full logical connectivity must be assumed, which means that every node must be able to connect every other node in the network. In general, the design consists of three problems to be solved optimally: 1. Route the VPCs; 2. Cut them into VPLs; 3. Dimension the VPLs. These problems are dependent of each other. When our model is applied, the third problem can be relaxed, instead we have a constraint: None of the VPL capacities should exceed the partition (channel) capacity. The invention proposes a two-phase method in order to accomplish optimality as often as possible, where the first phase involves finding one (or two for reliability purposes) node-and-vertex disjoint path(s) for each node-pair and assigning them capacities according to their traffic. The objective to be minimised when routing the VPCs is the total network cost. This calls for input parameters such as position of nodes, estimated traffic demands between nodes and the costs of building physical links between them when we design the network; or estimated traffic demands between the nodes and the physical topology with capacities given when we are rerouting the VPCs. The second part of the method involves a cutting of paths between node-pairs into sequences and merging these sequences into VPLs and assigning them a wavelength. The emphasis is laid on this second part. The objective to be minimised here is the total number of cuts of a VPC or in other words, minimising the number of VPLs along a VPC. It can mathematically be expressed as a constrained discrete optimisation problem, which can be solved by stochastic optimisation (e.g Simulated Annealing or Genetic Algorithm). The novelty of the invention lies in the two step model as a whole and especially the second step with the objective of the optimisation and the constraints, i.e. the fo
Pascal Leslie
Payne David
Telefonaktiebolaget LM Ericsson (publ)
LandOfFree
Method for optimizing a mostly optical network does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Method for optimizing a mostly optical network, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Method for optimizing a mostly optical network will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3364887