• Save
Floor Population Metrics, presented by Chip Webb, CTO at Anue Systems
Upcoming SlideShare
Loading in...5
×

Like this? Share it with your network

Share

Floor Population Metrics, presented by Chip Webb, CTO at Anue Systems

  • 1,657 views
Uploaded on

Chip Webb, CTO from Anue Systems- a leader in test and measurement solutions, presented new ITU-T recommendations that relate to floor population metrics are emerging, including G.8260, G.8261.1 and ...

Chip Webb, CTO from Anue Systems- a leader in test and measurement solutions, presented new ITU-T recommendations that relate to floor population metrics are emerging, including G.8260, G.8261.1 and G.8263. Network limits as well as slave tolerance limits are presented.

At the Workshop on Synchronization in Telecommunications Systems (WSTS in Broomfield, Colorado) was a call for improved testing standards, new models for supporting today’s network timing devices and new methods for reducing network errors, in a series of presentations

Network and wireless providers are under intense pressure to meet the demands for reliable information access with increased network bandwidth. This increasing demand requires network providers to ensure the accuracy of packet information shared across the network. Network “noise”, increased traffic and other factors can affect the ability of the network to accurately track data, resulting in a number of issues, such as lost or missing information or dropped calls on wireless networks.

  • Full Name Full Name Comment goes here.
    Are you sure you want to
    Your message goes here
    Be the first to comment
No Downloads

Views

Total Views
1,657
On Slideshare
1,644
From Embeds
13
Number of Embeds
1

Actions

Shares
Downloads
0
Comments
0
Likes
2

Embeds 13

http://www.linkedin.com 13

Report content

Flagged as inappropriate Flag as inappropriate
Flag as inappropriate

Select your reason for flagging this presentation as inappropriate.

Cancel
    No notes for slide

Transcript

  • 1. A short introduction to the new “Floor Population” MetricsWSTS 2012Chip WebbCTO © Copyright 2002-2012 Anue Systems, Inc.
  • 2. A short introduction to the new “Floor Population” Metrics ► Introduction ► Objective ► Lucky Packets ► Metric Definitions ► Floor Window ► PDV Performance Limits ► Example Measurement ► Conclusion2 © Copyright 2002-2012 Anue Systems, Inc.
  • 3. Introduction ►What are the new metrics? • Floor Packet Count (FPC) • Floor Packet Percentage (FPP) • Floor Packet Rate (FPR) ►Where are they used? New ITU Recommendations • Defined in G.8260 • Used in G.8261.1 as a network limit (1%) • Used in G.8263 as a slave tolerance limit (1% at slave spec’d min rate)3 © Copyright 2002-2012 Anue Systems, Inc.
  • 4. Packet Timing System4 © Copyright 2002-2012 Anue Systems, Inc.
  • 5. Packet Timing System (a more abstract view) Packet Network Master Slave5 © Copyright 2002-2012 Anue Systems, Inc.
  • 6. Packet Timing System (a more abstract view) Packet Network Master Slave The Floor Population Metrics are a way to measure Packet Delay Variation (PDV) in this system (NOTE: Other metrics, such as MAFE are also possible)6 © Copyright 2002-2012 Anue Systems, Inc.
  • 7. A short introduction to the new “Floor Population” Metrics ► Introduction ► Objective ► Lucky Packets ► Metric Definitions ► Floor Window ► PDV Performance Limits ► Example Measurement ► Conclusion7 © Copyright 2002-2012 Anue Systems, Inc.
  • 8. Objective of the new metrics (from G.8260) ►To study the population of timing packets within a certain fixed cluster range starting at the observed floor delay ►To compare the population with acceptance or rejection thresholds ►To ensure that at least a minimum number of packets, or a minimum percentage of packets remains within the specified cluster range starting at the observed floor delay8 © Copyright 2002-2012 Anue Systems, Inc.
  • 9. Objective of the new metrics To measure so-called “Lucky Packets”9 © Copyright 2002-2012 Anue Systems, Inc.
  • 10. A short introduction to the new “Floor Population” Metrics ► Introduction ► Objective ► Lucky Packets ► Metric Definitions ► Floor Window ► Performance Limits ► Example Measurement ► Conclusion10 © Copyright 2002-2012 Anue Systems, Inc.
  • 11. Quiz: Which of these shows Lucky Packets?11 © Copyright 2002-2012 Anue Systems, Inc.
  • 12. Quiz: Which of these shows Lucky Packets?12 © Copyright 2002-2012 Anue Systems, Inc.
  • 13. Quiz: Which of these shows Lucky Packets?13 © Copyright 2002-2012 Anue Systems, Inc.
  • 14. Lucky Packets ►Lucky packets are the packets that experience near minimum delay • They spend little or no time waiting in queues • They are fortunate to avoid congestion in the network ►Therefore lucky packets can be selected using a “cluster range” • Anchored at the minimum delay (observed or known) • Size of the cluster range affects the sensitivity of the measurement ►Cluster range is also called “Floor Window” • More on that in a minute14 © Copyright 2002-2012 Anue Systems, Inc.
  • 15. A short introduction to the new “Floor Population” Metrics ► Introduction ► Objective ► Lucky Packets ► Metric Definitions ► Floor Window ► PDV Performance Limits ► Example Measurement ► Conclusion15 © Copyright 2002-2012 Anue Systems, Inc.
  • 16. Metric Definitions16 © Copyright 2002-2012 Anue Systems, Inc.
  • 17. Metric Definitions17 © Copyright 2002-2012 Anue Systems, Inc.
  • 18. Metric Definitions18 © Copyright 2002-2012 Anue Systems, Inc.
  • 19. Metric Definitions • Game lasted 1 minute • Three darts thrown • Two hit Bull’s Eye • 1 point for Bull’s Eye STATS • Score=2 • Percent=67% (2/3) • Rate=2/minute19 © Copyright 2002-2012 Anue Systems, Inc.
  • 20. Metric Definitions via Dart Board Analogy ►Floor Packet Count (FPC) • The number of times a dart landed in the Bull’s Eye ►Floor Packet Percentage (FPP) • The percentage of times a dart landed in the Bull’s Eye ►Floor Packet Rate (FPR) • The rate that darts land in the Bull’s Eye (e.g. per minute or hour) Note: Full ►To apply to packet timing systems: mathematical • Replace “dart” with “timing packets” definitions are • Replace “land” with “have delay” (or “are delivered”) in backup • Replace “Bull’s Eye” with “Floor Window” slides  (size of Bull’s Eye is analogous to the “cluster range”20 © Copyright 2002-2012 Anue Systems, Inc.
  • 21. A short introduction to the new “Floor Population” Metrics ► Introduction ► Objective ► Lucky Packets ► Metric Definitions ► Floor Window ► PDV Performance Limits ► Example Measurement ► Conclusion21 © Copyright 2002-2012 Anue Systems, Inc.
  • 22. The floor window (a.k.a. the Bull’s Eye)22 © Copyright 2002-2012 Anue Systems, Inc.
  • 23. The Floor Window (a.k.a. the Bull’s Eye) ►Window has width, height and vertical position • Width is defined as 200 seconds 200 seconds Window23 © Copyright 2002-2012 Anue Systems, Inc. Floor
  • 24. The Floor Window (a.k.a. the Bull’s Eye) ►Window has width, height and vertical position • Width is defined as 200 seconds 200 seconds • Height is defined as 150 microseconds Window (NOTE: Not drawn to scale) Floor 150 s24 © Copyright 2002-2012 Anue Systems, Inc.
  • 25. The Floor Window (a.k.a. the Bull’s Eye) ►Window has width, height and vertical position • Width is defined as 200 seconds 200 seconds • Height is defined as 150 microseconds • Position of window is based on minimum observed delay Window (NOTE: Not drawn to scale) Floor 150 s Minimum delay (NOTE: Anue does not take a position on the suitability of specific window parameter values)25 © Copyright 2002-2012 Anue Systems, Inc.
  • 26. A short introduction to the new “Floor Population” Metrics ► Introduction ► Objective ► Lucky Packets ► Metric Definitions ► Floor Window ► PDV Performance Limits ► Example Measurement ► Conclusion26 © Copyright 2002-2012 Anue Systems, Inc.
  • 27. Network PDV Limit (G.8261.1)27 © Copyright 2002-2012 Anue Systems, Inc.
  • 28. Network PDV Limit Packet Network Master Slave28 © Copyright 2002-2012 Anue Systems, Inc.
  • 29. Network PDV Limit FPP ≥ 1% Packet Network Master Slave NOTE: This is a relative measurement and doesn’t depend on timing packet rate29 © Copyright 2002-2012 Anue Systems, Inc.
  • 30. Network PDV Limit FPP ≥ 1% Packet Network Master Slave NOTE: This is a relative measurement and doesn’t depend on timing packet rate30 © Copyright 2002-2012 Anue Systems, Inc.
  • 31. Slave PDV Tolerance (from G.8263)31 © Copyright 2002-2012 Anue Systems, Inc.
  • 32. Slave PDV Tolerance Packet Rate Master Packet Network Slave32 © Copyright 2002-2012 Anue Systems, Inc.
  • 33. Slave PDV Tolerance FPP ≥ 1% Packet Rate Master Packet Network Slave33 © Copyright 2002-2012 Anue Systems, Inc.
  • 34. Slave PDV Tolerance FPP ≥ 1% Packet Rate Master Packet Network Slave FPC FPR NOTE: This is an absolute measurement that depends on timing packet rate34 © Copyright 2002-2012 Anue Systems, Inc.
  • 35. A short introduction to the new “Floor Population” Metrics ► Introduction ► Objective ► Lucky Packets ► Metric Definitions ► Floor Window ► PDV Performance Limits ► Example Measurement ► Conclusion35 © Copyright 2002-2012 Anue Systems, Inc.
  • 36. Example Measurement ►Packet timing system operating at 32 packets per second ►Packet Delay Variation (PDV) based on flicker noise ►Low level of random packet loss (0.01%) ►Brief network outage (80 seconds) ►Steps for calculating FPC, FPP & FPR • Find minimum delay • Draw FPC graph, explain axes • Calculate with jumping window • Calculate with sliding window • Compare jumping and sliding36 © Copyright 2002-2012 Anue Systems, Inc.
  • 37. Example: Flicker PDV and 0.01% Loss (@32 pkt/sec)37 © Copyright 2002-2012 Anue Systems, Inc.
  • 38. Search for minimum delay value38 © Copyright 2002-2012 Anue Systems, Inc.
  • 39. Draw horizontal line for minimum delay39 © Copyright 2002-2012 Anue Systems, Inc.
  • 40. Draw the Floor Window 200 Seconds 150 seconds40 © Copyright 2002-2012 Anue Systems, Inc.
  • 41. Add the FPC graph41 © Copyright 2002-2012 Anue Systems, Inc.
  • 42. Add the FPC graph42 © Copyright 2002-2012 Anue Systems, Inc.
  • 43. Look at just the FPC Axis43 © Copyright 2002-2012 Anue Systems, Inc.
  • 44. Look at just the FPC Axis Smallest countable value is one packet44 © Copyright 2002-2012 Anue Systems, Inc.
  • 45. Look at just the FPC Axis Largest possible value is 6400 pkt. (32pkt/sec * 200 sec) Smallest countable value is one packet45 © Copyright 2002-2012 Anue Systems, Inc.
  • 46. Look at just the FPC Axis Largest possible value is 6400 pkt. (32pkt/sec * 200 sec) Axis has logarithmic scale Smallest countable value is one packet46 © Copyright 2002-2012 Anue Systems, Inc.
  • 47. Compare FPC to FPP Largest possible value is 6400 pkt. (32pkt/sec * 200 sec) Smallest countable value is one packet47 © Copyright 2002-2012 Anue Systems, Inc.
  • 48. Compare FPC to FPP Largest possible 6400 pkt value is 6400 pkt. is 100% (32pkt/sec * 200 sec) Smallest countable value is one packet48 © Copyright 2002-2012 Anue Systems, Inc.
  • 49. Compare FPC to FPP Largest possible 6400 pkt value is 6400 pkt. is 100% (32pkt/sec * 200 sec) 64 pkt is 1% Smallest countable value is one packet49 © Copyright 2002-2012 Anue Systems, Inc.
  • 50. Compare FPC and FPP to FPR Largest possible value is 6400 pkt. (32pkt/sec * 200 sec)Smallest countablevalue is one packet50 © Copyright 2002-2012 Anue Systems, Inc.
  • 51. Compare FPC and FPP to FPR Largest possible 32 pkt/sec is max value is 6400 pkt. (same as 100% FPP) (32pkt/sec * 200 sec)Smallest countablevalue is one packet51 © Copyright 2002-2012 Anue Systems, Inc.
  • 52. Compare FPC and FPP to FPR Largest possible 32 pkt/sec is max value is 6400 pkt. (same as 100% FPP) (32pkt/sec * 200 sec) 1% is 0.32pkt/sec and is 64 pktSmallest countablevalue is one packet52 © Copyright 2002-2012 Anue Systems, Inc.
  • 53. Draw the 1% FPP Limit Line53 © Copyright 2002-2012 Anue Systems, Inc.
  • 54. Draw the 1% FPP Limit Line54 © Copyright 2002-2012 Anue Systems, Inc.
  • 55. Calculate Floor Population with Jumping Window55 © Copyright 2002-2012 Anue Systems, Inc.
  • 56. Calculate Floor Population with Jumping Window56 © Copyright 2002-2012 Anue Systems, Inc.
  • 57. Calculate Floor Population with Sliding Window57 © Copyright 2002-2012 Anue Systems, Inc.
  • 58. Calculate Floor Population with Sliding Window58 © Copyright 2002-2012 Anue Systems, Inc.
  • 59. Compare: Jumping/Sliding59 © Copyright 2002-2012 Anue Systems, Inc.
  • 60. 60 © Copyright 2002-2012 Anue Systems, Inc.
  • 61. Conclusion ►Three new related metrics for “Lucky Packets” • FPC (Floor Packet Count) [How many darts hit the Bull’s Eye?] • FPP (Floor Packet Percent) [What percent of darts hit the Bull’s Eye?] • FPR (Floor Packet Rate) [How often do the darts hit the Bull’s Eye?] ►PDV Limit is FPP ≥ 1% • Window width is 200 sec. • Window height is 150us ►Two ways to calculate (depending on amount of window overlap) • Jumping windows • Sliding windows61 © Copyright 2002-2012 Anue Systems, Inc.
  • 62. Thank You! Questions? cwebb@anuesystems.com62 © Copyright 2002-2012 Anue Systems, Inc.
  • 63. Backup Slides63 © Copyright 2002-2012 Anue Systems, Inc.
  • 64. Formal Mathematical Definition of the new Metrics ►x[i] is the measured latency of timing packet i, • 0 ≤ i < N. (i.e. there are N packets in the data set ► Pis the nominal time between timing packets ► is the cluster range (vertical window height) ►W represent the window interval (horizontal window width) • It can also be expressed as K samples, K = W/ P. Note: It is assumed that the packet rate of the timing flow is nominally constant. The case for a variable rate of packet transmission is for further study.64 © Copyright 2002-2012 Anue Systems, Inc.
  • 65. Mathematical Definition of the Metrics ►Step 1: Find the minimum delay packet ►Step 2: Calculate the indicator function65 © Copyright 2002-2012 Anue Systems, Inc.
  • 66. Mathematical Definition of the Metrics (cont.) ►Step 3: Count the packets in the window (FPC) ►Step 4: Express this result as a packet rate (FPR) ►Step 5: Also express as a percentage (FPP):66 © Copyright 2002-2012 Anue Systems, Inc.
  • 67. Absolute and Relative Metrics ►FPP is a relative metric • Calculation does not depend on the timing packets rate • Relative means that the metric tells us what has changed between reference planes. ►FPC and FPR are absolute metrics • Calculation depends on the rate at which timing packets are sent ►Network performance is best measured as a relative limit • FPP compares the network output relative to its input • Since the network doesn’t create the packets, can’t be absolute ►Slave performance is best measured with an absolute limit • FPC or FPR • But G.8263 refers to G.8261.1 limit at a given packet rate (still absolute)67 © Copyright 2002-2012 Anue Systems, Inc.
  • 68. Slave PDV Tolerance Limit is actually an absolute level Pkt Rate PEC-S-F Packet Rate68 © Copyright 2002-2012 Anue Systems, Inc.
  • 69. Slave PDV Tolerance Limit is actually an absolute level Pkt Rate PEC-S-F Packet Rate 99% Actual PEC-S-F 1% FPR Limit69 © Copyright 2002-2012 Anue Systems, Inc.
  • 70. Slave PDV Tolerance Limit is actually an absolute level Pkt Rate e.g. 32 pkt/sec. PEC-S-F The range of Packet Rate acceptable packet rates is implementa- tion dependent. 99% Actual PEC-S-F 1% FPR Limit e.g. 0.32 pkt/sec.70 © Copyright 2002-2012 Anue Systems, Inc.
  • 71. Slave PDV Tolerance Limit is actually an absolute level Pkt Rate e.g. 32 pkt/sec. PEC-S-F The range of Packet Rate acceptable packet rates is implementa- tion dependent. 99% Actual PEC-S-F 1% FPR Limit NOTE: This is an e.g. 0.32 pkt/sec. absolute measurement NOTE: But it allows different because it depends on PEC-S-F implementations to timing packet rate. have different limits.71 © Copyright 2002-2012 Anue Systems, Inc.