Duration limited statistical multiplexing in packet networks

• US 5,121,383 A
• Filed: 11/16/1990
• Issued: 06/09/1992
• Est. Priority Date: 11/16/1990
• Status: Expired due to Term

• Current Assignee: TTI Inventions A LLC (Intellectual Ventures LLC)
• Original Assignee: Bell Communications Research, Inc. (Telefonaktiebolaget LM Ericsson)
• Inventors: Golestani, S. Jamaloddin
• Primary Examiner(s): Olms, Douglas W.
• Assistant Examiner(s): KIZOU, HASSAN
• Application Number:
  US07/615,014
• Time in Patent Office: 571 Days
• Field of Search: 370/94.1, 370/94.2, 370/94.3, 370/60, 370/60.1, 370/61, 370/85.6, 370/79, 370/112, 340/825.5, 340/825.51
• US Class Current: 370/235
• CPC Class Codes: H04J 3/247 ATM or packet multiplexing
• CPC Class Codes: H04L 47/10 Flow control; Congestion control
• CPC Class Codes: H04L 47/15 in relation to multipoint traffic arrangements for broadcast or multicast in data networks H04L12/18
• CPC Class Codes: H04L 47/2433 Allocation of priorities to traffic types
• CPC Class Codes: H04L 47/70 Admission control; Resource allocation
• CPC Class Codes: H04L 47/805 QOS or priority aware
• CPC Class Codes: H04L 47/822 Collecting or measuring resource availability data
• CPC Class Codes: H04L 47/823 Prediction of resource usage
• CPC Class Codes: H04L 47/828 Allocation of resources per group of connections, e.g. per group of users

Abstract

A method and associated node structure (11) for the efficient and reliable transmission of delay sensitive traffic in a packet network (10) is disclosed. The method provides bounded end-to-end delay for all delay sensitive traffic and can guarantee loss-free transmission to that part of the delay sensitive traffic which has such a stringent requirement. To achieve statistical multiplexing gain, some loss is permitted for the rest of the delay sensitive traffic, with discrimination among different loss priority classes based on the corresponding degree of loss sensitivity. Bounded end-to-end delay is obtained by performing statistical multiplexing at the switching nodes (11) on a duration limited basis through use of a unique queuing discipline at the network nodes (11). This queuing discipline is instrumental in guaranteeing loss free transmission for the class of traffic with such a requirement.

Claims

1. A method for transmitting delay sensitive packets in a packet switching network comprising a plurality of nodes interconnected by links, said method comprising the steps of:

assigning each of said packets to a specific loss priority class out of a plurality of loss priority classes before the packet is transmitted between nodes in the network,

designating at each node in said network, a packet arriving in a specific arriving frame at said node as eligible for transmission during a specific department frame on an outgoing link which is adjacent to said specific arriving frame in which the packet arrived at said node, and

transmitting during said adjacent frame on said outgoing link packets eligible for transmission in an order determined by loss priority class, and dropping eligible packets which cannot be served before the expiration of said adjacent frame.

2. The method of claim 1 wherein time frames of different durations are defined on said links and different ones of said packets are transmitted in said time frames of different durations, a packet arriving at a node in an incoming time frame of a particular duration being eligible for transmission in an adjacent departing frame of the same duration.

3. The method of claim 1 wherein said packets transmitted in said network are assigned to different traffic groups, and wherein the duration of said adjacent frame on said outgoing link is divided among the packets of said different traffic groups eligible for transmission during the adjacent frame such that each traffic group receives a predetermined share of the transmission capacity of said outgoing link, within each traffic group the packets being served by the outgoing link in an order determined by loss priority class.

4. The method of claim 1 wherein packets in the highest loss priority class are guaranteed transmission without loss and with a bounded delay between a source node and a destination node.

5. The method of claim 4 wherein packets in each lower loss priority class are transmitted between a source node and a destination node with a bounded end-to-end delay and with an end-to-end loss rate probability which does not exceed a predetermined maximum.

6. The method of claim 1 wherein packets are transmitted between a source node and a destination node with a jitter delay that is bounded by a frame duration.

7. The method of claim 1 wherein packet losses in a higher loss priority class are not affected by the amount of packets in a lower loss priority class.

8. The method of claim 1 wherein each of said packets belongs to a connection which extends between a source node and a destination node in said network and wherein said method comprises the step of admitting a particular connection k to said network if on each link in a path between a source node and a destination node of the connection k a transmission bandwidth r_k is available such that the aggregate length of packets of the connection k in the highest loss priority class does not exceed r_k.Tg, where Tg is a frame duration with which the connection k is associated.

9. The method of claim 8 wherein said method comprises admitting a particular connection to said network having packets in a lower loss priority class if the end-to-end loss probability for said packets in said lower loss priority class will not exceed a predetermined maximum and if the admission of said packets in said lower loss priority class will not cause the end-to-end loss probability for an already admitted connection to exceed a maximum loss rate of the already admitted connection.

10. The method of claim 9 wherein the packets belonging to a connection preserve their order throughout the network even when the packets belong to different loss priority classes.

11. The method of claim 9 wherein successive packet losses in a connection are substantially uncorrelated.

12. A method for transmitting delay sensitive packets belonging to particular connections between a source node and a destination node in a packet switching network comprising an array of nodes interconnected by links, each of said connections belonging to a particular traffic group g. g=1, . . . ,G and each of said packets belonging to a particular loss priority class p_g out of a plurality of loss priority classes p_g =1, . . . ,P_g associated with each traffic group, each of said traffic groups being allocated a predetermined fraction of the transmission capacity on each link, said method comprising the steps of:

on each link in said network, defining a sequence of time frames for each traffic group, the time frames on each link for the traffic group g having a duration Tg,

admitting a connection k to the network in a traffic group g only if a transmission bandwidth r_k is available for the traffic group g on each link in a path between a source node and a destination node for the connection k such that the aggregate length of packets of the connection k in the highest loss priority class P_g does not exceed r_k Tg in a frame of duration Tg, and

delaying a packet belonging to the traffic group g and arriving at a node of said network on an incoming link during a specific arriving frame of duration Tg and addressed for transmission by an outgoing link at least until the beginning of an adjacent departing frame of duration Tg on the outgoing link, and dropping any packet which cannot be served by an outgoing link in its corresponding adjacent departing frame.

13. The method of claim 12 wherein within each traffic group packets are served by an outgoing link according to loss priority class.

14. The method of claim 12 wherein excess packet traffic in one traffic group does not affect the transmission of packets in another traffic group.

15. The method of claim 12 wherein all of said connections belong to a single traffic group, and only a single sequence of time frames of a single duration T is defined on each link of said network.

16. The method of claim 12 where said admitting step comprises admitting connections with packets in a lower loss priority class only if the end-to-end loss rate probability for such packets will not exceed a predetermined maximum and if the admission of such packets will not cause the end-to-end loss rate probability of packets of an already existing connection in said network to exceed a predetermined maximum for the already existing connection.

17. The method of claim 12 wherein packets are served in an adjacent frame of an outgoing link in accordance with loss priority class and said dropping step comprises dropping packets only in the lower loss priority classes.

18. The method of claim 12 wherein a new connection is admitted to the network only if its admission will not modify the packet loss rate of an already existing connection in such a way that the packet loss rate of an existing connection will be increased beyond a predetermined tolerance level.

19. A node for use in a packet network for routing delay sensitive packets, each of said delay sensitive packets belonging to a particular connection in a particular traffic group and belonging to a particular loss priority class, said node comprising

a plurality of incoming and outgoing links,

receiving means for receiving said delay sensitive packets via said incoming links, said receiving means receiving each of said packets belonging to a particular connection of a particular traffic group in arriving frames of a particular duration associated with the particular traffic group,

storage means associated with each of said outgoing links for storing said packets addressed for transmission on each of said outgoing links, and

controller means associated with said storage means for controlling the transmission of said received packets on said outgoing links by designating each of said received packets being received in an arriving frame of a particular duration as eligible for transmission in a first departing frame of said particular duration on the outgoing link to which said each received packet is addressed which starts at or after the end of the frame of said particular duration in which said each received packet arrived at the node and by enabling each outgoing link to serve the packets designated as eligible for service by it during a department frame in an order determined by the loss priority class of the eligible packets, whereby packets which cannot be served by the outgoing link to which they are addressed in the departing frame for which they are designated as eligible are dropped.

20. The node of claim 19 wherein said receiving means comprises a switch.

21. The node of claim 19 wherein said storage means associated with each outgoing link comprises a set of FIFO queues for each traffic group, each set of FIFO queues containing a single FIFO queue for each loss priority class in the corresponding traffic group.

22. The node of claim 21 wherein said node belongs to a network which carries packets belonging to a single traffic group in frames of duration T and wherein said storage means associated with each outgoing link comprises a single set of FIFO queues, the set of FIFO queues associated with each outgoing link comprising one FIFO queue for each loss priority class in said single traffic group.

23. The node of claim 19 wherein said node comprises additional storage means associated with each outgoing link for storing non-delay sensitive packets.

24. The node of claim 19 wherein said controller means always causes packets in the highest loss priority of each traffic group to be transmitted by the desired outgoing link in the frame in which said packets in the highest loss priority class are designated as eligible for transmission.

25. The node of claim 19 wherein there is only one traffic group and all of said frames are of duration T.

26. The node of claim 19 wherein said controller means controls the transmission of said packets on said outgoing links based on the aggregate number of packets in each of said loss priority classes seeking service by each outgoing link, and without having to monitor the packets of each connection individually.

27. A method of transmitting delay sensitive packets in a packet switching network comprising a plurality of nodes interconnected by links, said method being CHARACTERIZED BY the steps of:

assigning each of said packets to a specific loss priority class out of a plurality of loss priority classes before the packet is transmitted between nodes in the network,

designating at each node in said network, a packet arriving in a specific arriving frame at said node as eligible for transmission during a specific departing frame on an outgoing link which is adjacent to said specific arriving frame in which the packet arrived at said node, and

transmitting during said adjacent frame packets eligible for transmission in an order determined by loss priority class, and dropping eligible packets which cannot be served before the expiration of said adjacent frame.