LEARNING-BASED SEMI-PERSISTENT SCHEDULING IN WIRELESS COMMUNICATIONS

• US 20090003282A1
• Filed: 05/06/2008
• Published: 01/01/2009
• Est. Priority Date: 05/07/2007
• Status: Active Grant

• Current Assignee: Qualcomm, Inc.
• Original Assignee: Qualcomm, Inc.
• Inventors: Damnjanovic, Aleksandar, Meylan, Arnaud, Chaponniere, Etienne F.
• Granted Patent: US 8,144,589 B2
• Time in Patent Office: Days
• US Class Current: 370/331
• CPC Class Codes: H04L 47/10 Flow control; Congestion control
• CPC Class Codes: H04L 47/36 by determining packet size, e.g. maximum transfer unit [MTU]
• CPC Class Codes: H04L 47/50 Queue scheduling
• CPC Class Codes: H04L 47/522 Dynamic queue service slot or variable bandwidth allocation
• CPC Class Codes: H04L 47/54 Loss aware scheduling
• CPC Class Codes: H04L 47/56 implementing delay-aware scheduling
• CPC Class Codes: H04L 47/6255 queue load conditions, e.g. longest queue first
• CPC Class Codes: H04W 28/02 Traffic management, e.g. flow control or congestion control
• CPC Class Codes: H04W 72/535 based on resource usage policies
• CPC Class Codes: H04W 8/04 Registration at HLR or HSS [Home Subscriber Server]

Abstract

Systems and methods are provided for a learning-based determination of semi-persistent scheduling of data-packet flow wireless communication. A packetized data flow served to a wireless terminal is fully scheduled for an initial period of time in order to collect statistics associated with scheduled packet sizes (Ss) and inter-packet times (Ts). Analysis of a cumulative distribution of {S, T} pairs indicate whether a characteristic packet size (S₀) and size dispersion (D₀) are associated with the cumulative distribution. Inter-time intervals associated with the characteristic size and dispersion complete a transport format. Semi-persistent scheduling is utilized for a packetized flow when a characteristic transport format can be extracted, or learned, from the accumulated statistics. Extracted transport formats can be employed to optimize scheduling efficiency upon handover.

Claims

1. A method comprising:

scheduling in full a packet flow during a specific period of time;

collecting accumulation statistics of scheduled packet sizes (Ss) and inter-packet time intervals (Ts); and

identifying a set of peaks of highest accumulation;

when a number of {S, T} pairs contained within a tolerance size (D) of the peak is above a threshold, utilizing semi-persistent scheduling.

2. The method of claim 1, further comprising:

when the accumulation statistics match a known statistics for data packets generated by a packet-flow generator, utilizing semi-persistent scheduling.

3. The method of claim 2, utilizing semi-persistent scheduling further comprises selecting a packet format (S₀) to accommodate a largest packet size within the tolerance size D.

4. The method of claim 3, further comprising selecting as a semi-persistent scheduling time interval a smallest time interval among packets within the tolerance size.

5. The method of claim 3, further comprising selecting as a semi-persistent scheduling time interval a maximum delay tolerated by the scheduled packet flow.

6. The method of claim 3, further comprising selecting as a semi-persistent scheduling time interval a reciprocal of a maximum packet loss rate for the scheduled packet flow.

7. The method of claim 3, further comprising selecting as a semi-persistent scheduling time interval a least common multiple of a set of time interval for scheduled packet sizes within the tolerance size.

8. The method of claim 2, utilizing semi-persistent scheduling further comprises selecting a packet format (S₀) and a semi-persistent time interval (T₀) that satisfy S₀×

T₀≦

ρ

, wherein ρ

is one of an average bitrate or a guaranteed bitrate.

9. The method of claim 2, utilizing semi-persistent scheduling further comprises selecting a packet format and a semi-persistent time interval jointly across a set of fully scheduled packet flows.

10. The method of claim 9, wherein selecting a packet format and a semi-persistent time interval jointly across a set of fully scheduled packet flows includes optimizing the packet format and the time interval based at least in part upon quality of service metrics associated with the scheduled flow.

11. The method of claim 5, wherein a packet format resulting from a concatenation of S₀ and D, and τ

comprise a transport format for semi-persistent scheduling of the packet flow.

12. The method of claim 2, further comprising continuing scheduling in full a packet flow when a number of {S, T} pairs contained within a tolerance size of the peak fails to be above the threshold.

13. The method of claim 12, identifying the set of peaks includes computing a set of momenta of the {S, T} pairs accumulation statistics.

14. The method of claim 13, wherein the tolerance size equals a square root of the second momentum of the {S, T} pairs accumulation statistics.

15. The method of claim 11, further comprising:

storing a transport format; and

conveying the transport format upon handover.

16. The method of claim 15, conveying the transport format upon handover comprises communicating the transport over a backhaul communication network.

17. An apparatus that operates in a wireless communication system, the apparatus comprising:

a processor configured to schedule in full a packet flow;

to generate an accumulation distribution of scheduled packet sizes (Ss) and inter-packet time intervals (Ts); and

when a number of {S, T} pairs contained within a tolerance size (D) of a peak in the accumulation distribution is above a threshold, to implement semi-persistent scheduling; and

a memory coupled to the processor.

18. The apparatus of claim 17, when the accumulation statistics match a statistics for data packets generated by a packet-flow generator, utilizing semi-persistent scheduling, the processor further configured to implement semi-persistent scheduling.

19. The apparatus of claim 18, wherein to implement semi-persistent scheduling comprises to select a packet format and a semi-persistent time interval jointly across a set of fully scheduled packet flows includes optimizing the packet format and the time interval based at least in part upon quality of service metrics associated with the scheduled flow and conveyed in respective flow labels.

20. The apparatus of claim 18, wherein to implement semi-persistent scheduling comprises to select a packet format (S₀) to accommodate a largest packet size within the tolerance size D.

21. The apparatus of claim 20, wherein to implement semi-persistent scheduling further comprises to select as a semi-persistent scheduling time interval a smallest time interval among packets within the tolerance size.

22. The apparatus of claim 20, wherein to implement semi-persistent scheduling further comprises to select as a semi-persistent scheduling time interval a least common multiple of a set of time interval for scheduled packet sizes within the tolerance size.

23. The apparatus of claim 20, wherein to implement semi-persistent scheduling comprises to select as a semi-persistent scheduling time interval one of a maximum delay tolerated by the scheduled packet flow or a maximum packet loss rate for the scheduled packet flow.

24. The apparatus of claim 23, wherein to implement semi-persistent scheduling comprises to select a packet format (S₁) and a semi-persistent time interval (τ

) that satisfy S₁×

τ

≦

R, wherein R is one of an average bitrate or a guaranteed bitrate.

25. The apparatus of claim 24, wherein a packet format resulting from a concatenation of S₀ and D, and the semi-persistent time interval comprise a transport format for semi-persistent scheduling of the packet flow.

26. The apparatus of claim 18, the processor further configured to compute a set of momenta of the {S, T} pairs accumulation distribution to determine the distribution shape.

27. The apparatus of claim 26, wherein the memory stores the accumulation distribution of {S, T} pairs.

28. The apparatus of claim 18, wherein the memory stores a set of statistics for a set of data-packet flows generated by a set of data packet generators.

29. The apparatus of claim 26, the memory stores a set of transport formats for semi-persistent scheduling.

30. The apparatus of claim 29, the processor further configured to convey a stored transport format upon handover.

31. The apparatus of claim 30, wherein the stored transport format is conveyed over a backhaul communication network.

32. A wireless communication device comprising:

means for accumulating distribution of fully scheduled packet sizes (Ss) and inter-packet time intervals (Ts);

means for utilizing semi-persistent scheduling when a number of {S, T} pairs contained within a tolerance size (D) of a peak in the accumulation distribution is above a threshold; and

means for utilizing semi-persistent scheduling when the accumulation statistics match a known statistics for data packets generated by a packet-flow generator, utilizing semi-persistent scheduling.

33. The device of claim 32, further comprising means for scheduling in full a packetized data flow.

34. The device of claim 32, wherein means for utilizing semi-persistent scheduling comprises means for selecting a packet format (S₀) to accommodate a largest packet size within the tolerance size D.

35. The device of claim 32, wherein means for utilizing semi-persistent scheduling comprises means for selecting a packet format and a semi-persistent time interval jointly across a set of fully scheduled packet flows includes optimizing the packet format and the time interval based at least in part upon quality of service metrics associated with the scheduled flow and conveyed in respective flow labels.

36. The device of claim 34, wherein means to for implementing semi-persistent scheduling further comprises means for selecting as a semi-persistent scheduling time interval (τ

) one of a smallest time interval among packets within the tolerance size, a least common multiple of a set of time interval for scheduled packet sizes within the tolerance size, a maximum delay tolerated by the scheduled packet flow or a maximum packet loss rate for the scheduled packet flow.

37. The device of claim 34, wherein a packet format resulting from a concatenation of S₀ and D, and τ

comprise a transport format for semi-persistent scheduling of the packet flow.

38. The device of claim 33, further comprising means for computing a set of momenta of the {S, T} pairs accumulation distribution to determine the distribution shape.

39. The device of claim 38, means for storing a set of transport formats for semi-persistent scheduling.

40. The device of claim 39, further comprising means for conveying a set of stored transport formats upon handover.

41. A computer program product including a computer-readable medium comprising:

code for causing a computer to schedule in full a packet flow;

code for causing a computer to collect accumulation statistics over a specific period of time of fully scheduled packet sizes (Ss) and inter-packet time intervals (Ts);

code for causing a computer to identify a set of peaks of highest accumulation;

code for causing a computer to implement semi-persistent scheduling when a number of {S, T} pairs contained within a tolerance size (D) of the peak is above a threshold.

42. The computer program product of claim 41, wherein the computer-readable medium further includes code for causing a computer to implement semi-persistent scheduling when the accumulation statistics match a known statistics for data packets generated by a packet-flow generator, utilizing semi-persistent scheduling.

43. The computer program product of claim 42, wherein code for causing a computer to implement semi-persistent scheduling includes code for causing the computer to select a semi-persistent packet format (S₀) and a semi-persistent scheduling time interval (T₀).

44. The computer program product of claim 42, wherein S₀ accommodates a largest packet size within the tolerance size D.

45. The computer program product of claim 42, wherein S₀×

T₀≦

R, with R being one of an average bitrate or a guaranteed bitrate.

46. The computer program product of claim 42, wherein T₀ is a smallest time interval among packets within the tolerance size.

47. The computer program product of claim 42, wherein T₀ is a maximum delay tolerated by the scheduled packet flow, wherein the maximum delay is conveyed by a flow label.

48. The computer program product of claim 47, wherein T₀ is a reciprocal of a maximum packet loss rate for the scheduled packet flow, wherein the maximum packet loss rate is conveyed by a flow label.

49. The computer program product of claim 42, wherein the computer-readable medium further comprises code for causing a computer to concatenate S₀ and D.

51. The computer program product of claim 41, code for causing a computer to identify a set of peaks of highest accumulation includes code for causing the computer to compute a set of momenta of the {S, T} pairs accumulation statistics to determine an accumulation distribution shape.

50. The computer program product of claim 50, wherein the concatenation of S₀ and D, and τ

comprise a transport format for semi-persistent scheduling of the packet flow.