Method, system and router providing active queue management in packet transmission systems

• US 7,139,281 B1
• Filed: 04/07/2000
• Issued: 11/21/2006
• Est. Priority Date: 04/07/1999
• Status: Expired due to Fees

• Current Assignee: Teliasonera AB
• Original Assignee: Teliasonera AB
• Inventors: Bodin, Ulf
• Primary Examiner(s): NGUYEN, STEVEN H D
• Application Number:
  US09/926,280
• Time in Patent Office: 2,419 Days
• Field of Search: None
• US Class Current: 370/412
• CPC Class Codes: H04L 47/10 Flow control; Congestion control
• CPC Class Codes: H04L 47/2408 for supporting different services, e.g. a differentiated services [DiffServ] type of service
• CPC Class Codes: H04L 47/29 using a combination of thresholds
• CPC Class Codes: H04L 47/30 in combination with information about buffer occupancy at either end or at transit nodes
• CPC Class Codes: H04L 47/32 by discarding or delaying data units, e.g. packets or frames
• CPC Class Codes: H04Q 11/0421 Circuit arrangements therefor
• CPC Class Codes: H04Q 2213/13103 Memory
• CPC Class Codes: H04Q 2213/13164 Traffic (registration, measurement,...)
• CPC Class Codes: H04Q 2213/13166 Fault prevention
• CPC Class Codes: H04Q 2213/13174 Data transmission, file transfer
• CPC Class Codes: H04Q 2213/13196 Connection circuit/link/trunk/junction, bridge, router, gateway
• CPC Class Codes: H04Q 2213/13389 LAN, internet

Abstract

A method of active queue management, for handling prioritized traffic in a packet transmission system, the method including: providing differentiation between traffic originating from rate adaptive applications that respond to packet loss, wherein traffic is assigned to one of at least two drop precedent levels in-profile and out-profile; preventing starvation of low prioritized traffic; preserving a strict hierarchy among precedence levels; providing absolute differentiation of traffic; and reclassifying a packet of the traffic of the packet transmission system, tagged as in-profile, as out-profile, when a drop probability assigned to the packet is greater than a drop probability calculated from an average queue length for in-profile packets.

Claims

1. A method of active queue management, for handling prioritized traffic in a packet transmission system, the method comprising:

providing differentiation between traffic originating from rate adaptive applications that respond to packet loss, wherein traffic is assigned to one of at least two drop precedent levels in-profile and out-profile;

preventing starvation of low prioritized traffic;

preserving a strict hierarchy among precedence levels;

providing absolute differentiation of traffic; and

reclassifying a packet of the traffic of the packet transmission system, tagged as in-profile, as out-profile, when a drop probability assigned to the packet is greater than a drop probability calculated from an average queue length for in-profile packets.

2. The method as claimed in claim 1, further comprising:

providing absolute differentiation if a prioritized traffic is fully controlled and relative differentiation if the prioritized traffic is not fully controlled.

3. The method as claimed in claim 1 further comprising:

discarding a packet, tagged as out-profile, when a drop probability assigned to the packet is greater than a drop probability calculated from an average queue length for out-profile packets.

4. The method as claimed in claim 3, wherein:

a maximum threshold value for the average queue length for out-profile packets is max_th_out,

a maximum threshold value for the average queue length for in-profile packets is max_th_in, and

max_th_out is set to a greater value than max_th_in.

5. A telecommunications system for transmission of packet data, wherein said telecommunications system employs a method of active queue management as claimed in claim 1.

6. A telecommunications system as claimed in claim 5, wherein said telecommunications system is an internet.

7. A router for use with a telecommunications system, as claimed in claim 5, wherein said router employs the method of active queue management.

8. A method of active queue management for handling prioritized traffic in a packet transmission system, the method comprising:

providing differentiation between traffic originating from rate adaptive applications that respond to packet loss, in which packets of traffic are assigned to one of a plurality of drop precedence levels,

using a modified random early detection in and out RIO to calculate RIO drop probabilities,

using a load tolerant random early detection in and out ItRIO to calculate ItRIO drop probabilities,

using a weighted random early detection WRED to calculate WRED drop probabilities,

creating a plurality of threshold levels for an average queue length, by applying the RIO, ItRIO and WRED drop probabilities to the plurality of drop precedence levels,

setting all maximum threshold levels to a same value, and

reclassifying or dropping the packets of the traffic in a queue of the transmission system by using the plurality of threshold levels and the maximum threshold levels of the queue.

9. The method as claimed in claim 8, wherein,

a maximum threshold value for the average queue length for in-profile packets is max_th_in,

a minimum threshold value for the average queue length for in-profile packets is min_th_in, and

a maximum drop probability for packets marked as in-profile is max_p_in,

said method further comprising:

using a set of threshold parameters, including max_th_in, min_th_in, and max_p_in, instead of random early detection RED parameters, to determine whether an in-profile packet should be tagged as out-profile.

10. The method as claimed in claim 9, said method further comprising:

setting a plurality of maximum threshold parameters max_th#, including max_th_in and max_th_out, to a same value.

11. The method as claimed in claim 9, wherein there are three levels of drop precedence, said method further including:

calculating an average queue length for each level of drop precedence based on packets tagged with a corresponding level and packets tagged with a higher level of drop precedence.

12. The method as claimed in claim 11, further comprising:

assigning a unique threshold to each of the two highest prioritized precedence levels, said unique threshold being used to determine when a packet is to be tagged with a lower precedence level, and

providing a relative differentiation among said three levels when the average queue lengths for the two highest precedence levels exceeds both thresholds.

13. The method as claimed in claim 12, further comprising:

providing more than three drop precedence levels, and

employing an average queue length parameter for each drop precedence level with associated minimum threshold parameters min_th#s and maximum drop probability values max_p#s.

14. The method as claimed in claim 13, wherein there are eight drop precedence levels.

15. The method as claimed in claim 13, wherein there is a single minimum threshold th_in, for all precedence levels such that no packets are dropped if the average queue length is less than th_in.

16. A method of active queue management for handling prioritized traffic in a packet transmission system, configured to provide differentiation between traffic originating from rate adaptive applications that respond to packet loss, wherein traffic is assigned one of at least a first and second drop precedent level, namely in-profile and out-profile, said method including of:

calculating an average queue length avg_ql;

assigning minimum thresholds min_th_in and min_th_out, for in-profile packets and out-profile packets respectively, and a maximum threshold max_th;

retaining all packets with their initially assigned drop precedent levels while the avg_ql is less than, or equal to, a threshold th_in;

assigning a drop probability to each packet, determined from the average queue length;

retaining all packets while the avg_ql is less than the th_in; and

dropping packets in accordance with their assigned drop probability;

wherein in the assigning the drop probability to each packet, maximum drop probabilities max_p_in and max_p_out for in-profile packets and out-profile packets are not exceeded; and

wherein

max_p_out is greater than max_p_in, max_p_out being the maximum drop probability of packets marked as out-profile and max_p_in being the maximum drop probability for packets marked as in-profile.

17. The method as claimed in claim 16, further comprising:

applying said method to a FIFO queue.

18. The method as claimed in claim 16, further comprising:

dropping a packet if avg_ql is >

max_th, when a packet arrives;

calculating an average queue length for a packet tagged as in-profile avg_ql_in, and, if avg_ql_in >

th_in and min_th_in <

avg_ql, calculating a probability of dropping a packet tagged as in-profile Pin and dropping or retaining said in-profile packet in accordance with a value of Pin;

calculating a probability of dropping a packet tagged as out-profile Pout if min_th_out <

avg_ql, and dropping or retaining said out-profile packet in accordance with a value of Pout.

19. The method as claimed in claim 16, further comprising:

employing a plurality of drop precedence levels, greater than two, and deriving an average queue length for each drop precedence level.

20. The method as claimed in claim 18, further comprising:

setting max_th for each drop precedence level to the same value.

21. The method as claimed in claim 19, wherein there are three levels of drop precedence, further comprising:

calculating an average queue length for each level of drop precedence based on packets tagged with the corresponding level and packets tagged with a higher level of drop precedence.

22. The method as claimed in claim 21, further comprising:

assigning a unique threshold to each of the two highest prioritized precedence levels, said unique thresholds being used to determine when a packet is to be tagged with a lower precedence level, and

providing a relative differentiation among said three levels when the average queue lengths for the two highest precedence levels exceeds both thresholds.

23. The method as claimed in claim 22, further comprising:

providing more than three drop precedence levels; and

employing an average queue length parameter for each drop precedence level with associated thresholds min_th#s and max_p#s.

24. The method as claimed in claim 23, wherein there are eight drop precedence levels.

25. The method as claimed in claim 23, wherein there is a single minimum threshold, th_in, for all precedence levels such that no packets are dropped if the average queue length is less than th_in.