Jitterless and lossless switching fabric with network routing

• US 6,192,049 B1
• Filed: 07/15/1997
• Issued: 02/20/2001
• Est. Priority Date: 07/15/1997
• Status: Expired due to Fees

• Current Assignee: Lucent Technologies, Inc. (Nokia Corporation)
• Original Assignee: Lucent Technologies, Inc. (Nokia Corporation)
• Inventors: Sohraby, Kazem A.
• Primary Examiner(s): Olms, Douglas W.
• Assistant Examiner(s): Pizarro, Ricardo M.
• Application Number:
  US08/900,373
• Time in Patent Office: 1,316 Days
• Field of Search: 370/389, 370/395, 370/396, 370/398, 370/399, 370/468, 370/474, 370/477, 370/400, 370/401, 370/537, 370/539
• US Class Current: 370/389
• CPC Class Codes: H04L 2012/562 Routing
• CPC Class Codes: H04L 2012/5681 Buffer or queue management
• CPC Class Codes: H04L 49/256 Routing or path finding in ATM switching fabrics
• CPC Class Codes: H04L 49/3081 ATM peripheral units, e.g. policing, insertion or extraction
• CPC Class Codes: H04Q 11/0478 Provisions for broadband connections

Abstract

A switching fabric has input ports, output ports, buffers and at least one switch processor, which determines the available capacity of links at a switch and communicates this information to other switches in a network on a periodic basis. Based on this knowledge, the switch processor of the present invention determines if the necessary number of contiguous time slots in a time frame are available for the call in question. The routing method of the present method uses the periodically updated bandwidth usage to determine a route from a source switch to a destination switch. By utilizing periodic updates of bandwidth usage and assigning contiguous time slots, time slot availability becomes predictable and cell streams travel jointly along the same route, which in turn eliminates cell loss and jitter.

Claims

1. A switching fabric which provides substantially jitterless and lossless transmission from a source switch to a destination switch in a network, comprising:

at least one switch processor for determining, in response to at least one call request, if capacity exists in a link to establish a route from said source switch to said destination switch, said at least one switch processor blocking said call when no bandwidth is available;

at least one buffer for holding data packets associated with said at least one call when bandwidth is available in said link, said at least one buffer coupled to said switch processor; and

said at least one switch processor further operable to provision a required number of contiguous time slots in a specified frame when said bandwidth is available in said link, said processor further operable to route said data packets by counting said number of contiguous time slots provisioned to said data packets and cross-referencing said number of contiguous time slots to an output port and to switch said data packets in said given time frames.

2. The switching fabric according to claim 1, said at least one switch processor further operable to update said capacity determination on a periodic basis.

3. The switching fabric according to claim 2, wherein said periodic basis is at least as short as a call holding time.

4. The switching fabric according to claim 2, said at least one switch processor further operable to exchange said capacity determination on a periodic basis with other switches in a network.

5. The switching fabric according to claim 4, wherein said exchange is performed during a given time.

6. The switching fabric according to claim 5, wherein said exchange is performed during a call setup time.

7. The switching fabric according to claim 4, wherein a manner of said exchange is one selected from a group consisting of synchronous, asynchronous, distributed and centralized.

8. The switching fabric according to claim 4, said at least one switch processor further operable to internally route said data packets by pointing to a boundary of a group of contiguous time slots in a time frame.

9. The switching fabric according to claim 8, wherein said at least one buffer includes at least one constant bit rate buffer and wherein a size of said buffers holds a number of data packets receivable over a transport medium in given time frame.

10. The switching fabric according to claim 9, wherein said at least one buffer further includes at least one variable bit rate buffer.

11. The switching fabric according to claim 1 wherein said at least one switch processor further operable to note an endpoint time slot associated with said call.

12. The switching fabric according to claim 1, wherein said at least one switch processor is coupled to at least one input port and further coupled to at least one output port.

13. The switching fabric according to claim 12, wherein said at least one switch processor is integral with said input port.

14. The switching fabric according to claim 12, wherein said at least one witch processor is integral with said output port.

15. The switching fabric according to claim 12, wherein an add/drop multiplexer is coupled to said input port and said at least one buffer, said add/drop multiplexer operable to transfer said data packets within said given time frame from said at least one input port to said at least one output port.

16. The switching fabric according to claim 12, wherein a bus structure is coupled to said at least one buffer and to said output port.

17. The switching fabric according to claim 12, wherein a ring structure is coupled to said at least one buffer and to said output port.

18. The switching fabric according to claim 1, wherein said at least one switch processor determines said capacity for a constant bit rate call by using a fixed bandwidth value associated with said constant bit rate call.

19. The switching fabric according to claim 18, wherein said at least one switch processor determines said capacity for a variable bit rate call by using a bounding parameter selected from the group consisting of a peak rate and a minimum bandwidth.

20. The switching fabric according to claim 18, said at least one switch processor further operable to update said capacity determination on a periodic basis.

21. The switching fabric according to claim 18, said at least one switch processor further operable to exchange said capacity determination on a periodic basis with other switches in a network.

22. The switching fabric according to claim 18, said at least one switch processor further operable to internally route said data packets by referencing in a group manner a number of contiguous time slots allocated to said data packets associated with said call to an output port, and wherein said at least one switch processor switches said data packets in given time frames.

23. A method for providing substantially jitterless and lossless transmission from a source switch to a destination switch, said method comprising the steps of:

determining a set of candidate routes, in response to at least one call request, by calculating link capacity between said source switch and said destination switch;

comparing a required bandwidth of said call request with a maximum available bandwidth of a link to select a call route;

selecting a direct link for said call route if said capacity of said direct link is available;

selecting an alternate link for said call route if said direct link is unavailable and said alternate link has available bandwidth, said alternate link including at least one switch;

blocking said call request when said required bandwidth is unavailable;

updating said link capacity on a periodic basis at said source switch, said destination switch and other switches to permit dynamic changing of said call route during a call

assigning a contiguous set of time slots in a given time frame to accommodate said call request;

internally routing data packets by cross-referencing a number of contiguous time slots assigned for said call request with an output port; and

switching said data packets associated with said call request within said given time frame.

24. The method according to claim 23, further including the steps of:

assigning a contiguous set of time slots in a given time frame to accommodate said call request; and

buffering data packets associated with said call request contiguously in a buffer.

25. The method according to claim 24, wherein a size of said buffer is structured to hold a number of data packets receivable over a transport medium in said given time frame.

26. The method according to claim 24, wherein said buffer is a partitioned memory device for holding a number of data packets receivable over a transport medium.

27. A network including a switching fabric which provides substantially jitterless and lossless transmission from a source switch to a destination switch, comprising:

at least one switch processor for determining, in response to at least one call request, if capacity exists in a link to establish a call route from said source switch to said destination switch, said at least one switch processor blocking said call request when said capacity is unavailable;

at least one buffer for holding data packets associated with said at least one call when bandwidth is available in said link, said at least one buffer coupled to said at least one switch processor; and

said at least one switch processor further operable to provision a required number of contiguous time slots in a given time frame and to route said data packets by cross-referencing said number of contiguous time slots provisioned to said data packets associated with said call an output port and switch said data packets in given time frames.

28. The network according to claim 27, said at least one switch processor further operable to update said capacity determination on a periodic basis and further operable to exchange said capacity determination on said periodic basis with other switches in said network.

29. The network according to claim 27, wherein said at least one buffer includes at least one constant bit rate buffer and wherein a size of said at least one buffer is structured to hold a number of data packets receivable over a transport medium in said given time frame.

30. The network according to claim 29, wherein said at least one buffer further includes at least one variable bit rate buffer.

31. The network according to claim 29, wherein said at least one switch processor is coupled to at least one input port and further coupled to at least one output port, and wherein said at least one switch processor is integral with one of said input port and said output port.

32. The network according to claim 27, wherein said at least one switch processor is a centralized processor.

33. The network according to claim 27, wherein said at least one switch processor is a distributed processor.

34. The network according to claim 27, further including at least one add/drop multiplexer for transferring said data packets in said given time frame.

35. A switching fabric for providing substantially jitterless and lossless switching of data packets, comprising;

at least one switch processor for determining bandwidth availability by checking periodically updated bandwidth availability from other switches in response to a call request, said at least one switch processor rejecting said call request when a required bandwidth is unavailable;

said at least one switch processor assigning contiguous time slots in a given time frame to said data packets when said required bandwidth is available;

at least one input unit for processing said data packets when said required bandwidth is available wherein said input unit includes at least one buffer for holding said data packets, said buffer sized to hold a number of data packets corresponding to said given time frame and a speed of a transport medium; and

at least one output unit for receiving processed data packets from said at least one input unit and transmitting said processed data packets to a next switch.

36. The switching fabric according to claim 35, wherein said at least one input unit includes at least one add/drop multiplexer for transferring said data packets.