Multiplex communications – Pathfinding or routing – Through a circuit switch
Reexamination Certificate
1999-07-16
2002-09-17
Ngo, Ricky (Department: 2664)
Multiplex communications
Pathfinding or routing
Through a circuit switch
C370S217000, C370S395100
Reexamination Certificate
active
06452926
ABSTRACT:
The present invention relates to an ATM-switch, in particular a large ATM-switch.
BACKGROUND
The use of ATM-networks in particular as transport networks is more and more expanded owing to their versatile capability of transmitting information at varying speeds and with different priorities, where the information can be derived from sources having very different characteristics and needs. In large such networks there is naturally a need for very fast switching which must be performed in the switching nodes of the network since the amount of information that is to be handled in each switching node is significant. Also, in such large networks each node can be connected to a large number of other nodes and there will then be need for fast switches which are also capable of switching data cell streams between a large number of inputs and a large number of outputs. Such switches must also be reliable and it must be ensured that cells are not lost.
Switches or switching units used in or constituting a node of a network may be constructed as a single switch element or switch core. A general problem with a switch containing a single switch core is that if some part of the switch core is faulty, all traffic flowing through the switch will be affected. A common way of solving this problem is to provide at least two identical switch cores, which in this case often are called parallel planes, to which identical cell streams are fed, and make a selection of correctly transferred cells at the egress side of the combination of parallel planes, a preferred one of the parallel plane then being used from which cells to be forwarded are taken. If one plane is faulty, one of the other parallel planes is selected as the preferred plane. The problem of using this principle in a large switch having a large number of input ports and output ports is that the probability of the occurrence of a fault is relatively high, due to the large amount of hardware required in each switch plane, and that the selection of a preferred plane from which correct cells are to be taken can not be performed without disturbing all traffic, unless some very sophisticated mechanism is used.
Another method is to use load sharing between a plurality of, say n+1, identical planes, and in such a case, if one of the load-sharing planes is faulty, the traffic is automatically distributed among the other n load-sharing planes. The switch should then be designed so that the capacity of these remaining n load-sharing planes will be sufficient to switch the full traffic through the switch.
An ATM switching system is disclosed in U.S. Pat. No. 5,471,460 having e.g. a three-stage construction comprising stages having a plurality of switch elements. A self-routing switching network is disclosed in U.S. Pat. No. 5,237,565 having a plurality of stages each having at least two switch elements. In a first stage or in first stages the incoming traffic can be evenly distributed to balance to the cell load over the different outlets. However, this distribution is only made for a special command, a “distribution” mode signal DI, in the cells.
SUMMARY
It is an object of the present invention to provide a switching unit or switch to be used in primarily an ATM-network allowing a reliable switching at a high speed and having a large number of inputs and outputs, the switching having redundancy to be used in the case of errors or failures of hardware components of the switching unit or switch.
It is another object of the invention to provide a composite, large and high speed switch built of switch elements which are commercially available.
The problem solved by the invention is thus how to provide a large switch that can operate at a high speed and in a reliable way.
Thus, a composite switch is built of a plurality of switch elements, the switch elements being arranged in at least three stages. The stages are connected serially so that a cell passing through the composite switch has to pass successively all the stages. The switch elements in a first stage work in a distributed mode switching at random received cells to outport ports of the switch elements of this stage. The switch elements of the first stage are duplicated to provide redundancy in the shape of redundant parallel planes. A cell incoming to the composite switch are by input circuits distributed to both redundant planes in the first stage and is transferred in identical copies over the planes to output ports which correspond to each other, i.e. have the same order number. A plane selection in order to select one of the identical cell copies is made after each pair of switch elements in the first stage, at each pair of corresponding output ports. This selection is made independently of each other for the different pairs of switches in the first stage. This means that if a switch element is faulty, which is the preferred switch element from which cells are taken on the output side of the pair to which this switch element belongs, a switch-over is made to the other switch element of this pair, the switch-over being made only for this pair of switch elements, so that this other switch element now is the preferred one, from which cells are taken. Thus only the traffic of cells through this pair of switch elements is affected by the switch-over and by the faulty switch element.
The switch elements in the first stage work in the distribution mode, which means, as indicated above, that an incoming cell will be switched at random to any of the outputs of a considered switch element. In second and third stages of the switch the switch elements of these stages all work in an ordinary switching mode. If a switch element in the second stage fails, the traffic will only pass through the other switch elements in the second stage. This is possible by the random distribution of cells to output ports in the first stage and by connecting a switch element in the first stage to all switch elements in the second stage, so that all switch elements in the second stage will receive cells from each switch element in the first stage. If the number of switch elements in the second stage is high, say 32, there will for a failure of a switch element in the second stage still be 31/32 of the capacity left, which is a sufficient capacity in most of the possible applications of the switch.
The second stage of the composite switch is designed as a load-sharing stage in which the switch elements work in a conventional switching mode, as indicated above, and will direct cells to the correct switch element in the third stage of the switch as controlled by some routing information such as a routing tag carried in each cell. The third stage has like the first stage duplicated switch elements or planes forming redundant parallel planes but working in a conventional switching mode like the switch elements in the second stage. In the case where one of the switch elements in a pair of parallel switch elements in the third stage fails, a switch-over is made to the other parallel plane for this pair of switch elements only, in the same way as for the switch elements of the first stage.
The principle as outlined above can generally be used in a composite switch comprising at least three stages of switch elements. If more stages are used, one or more of the middle stages, i.e. of those which are not directly connected at the ingress side or at the egress side of the composite switch, can be treated as one stage operating in a load-sharing mode, i.e. such a switching unit is still considered to be basically built of three stages. In a practical case the composite switch can thus comprise five elementary stages, the first and fifth stages having redundant planes and the second, third and fourth stages being load-sharing. Also, the input stage and the output stage can be built of substages, each working in a redundant mode having redundant parallel planes. The composite switch can then for example consist of five elementary stages, the first and second stages having redundant planes, the third plane working in a load-sharing man
Ngo Ricky
Nixon & Vanderhye P.C.
Telefonaktiebolaget LM Ericsson (publ)
LandOfFree
Reliable and robust atm-switch does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Reliable and robust atm-switch, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Reliable and robust atm-switch will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-2853990