1 / 36

Mitigating Egregious ACK Delays in Cellular Data Networks by Eliminating TCP ACK Clocking

Mitigating Egregious ACK Delays in Cellular Data Networks by Eliminating TCP ACK Clocking. Wai Kay Leong, Yin Xu , Ben Leong, Zixiao Wang National University of Singapore. Asymmetry in Cellular Networks. Congestion in Uplink Concurrent Upload RSFC [ Xu et al. 2012 ] Other Users.

kevyn
Télécharger la présentation

Mitigating Egregious ACK Delays in Cellular Data Networks by Eliminating TCP ACK Clocking

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript

  1. Mitigating Egregious ACK Delays in Cellular Data Networks by Eliminating TCP ACK Clocking Wai Kay Leong, Yin Xu, Ben Leong, Zixiao Wang National University of Singapore

  2. Asymmetry in Cellular Networks • Congestion in Uplink • Concurrent Upload • RSFC [Xu et al. 2012] • Other Users Uplink Downlink

  3. Egregious ACK Delays • TCP congestion control is ACK-clocked Data Data ACK

  4. Egregious ACK Delays • TCP congestion control is ACK-clocked • Congested uplink can delay ACKs • Downlink becomes idle Data Data ACK

  5. Egregious ACK Delays • TCP congestion control is ACK-clocked • Congested uplink can delay ACKs • Downlink becomes idle Data ACK

  6. Solution: Eliminate ACK Clocking Idea: If we know the bandwidth, we can send at maximum rate. cwnd Send Data at XMb/s Bandwidth: X Mb/s Data Data Data ACK Not dependent on ACK arrival

  7. Challenge 1: Estimating Bandwidth

  8. Estimating Receive Rate Mobile Receiver Sender TSval = tr0 TSval = tr1 TSval = tr0 Receive Rate ρ TSval = tr1 ΔACK/Δt=ρ tr1 – tr0 = Δt

  9. Challenges • Estimating Bandwidth • Timestamp Granularity too Coarse • Cannot estimate with high accuracy • Bandwidth variation • Have to keep updating estimation

  10. Solution Self-oscillating Feedback Loop • Estimate Receive Rate ρ • Send Rate σ How to detect congestion? No Congestion Send Faster ( σ > ρ) Send slower ( σ < ρ) Link Congested

  11. Detect Congestion • Idea: Monitor Queuing Delay • How? • TCP Timestamps • Relative Difference between sender and receiver Buffer tbuff = 0 tbuff> 0

  12. Detecting Congestion Mobile Receiver Sender ts1 x TSval = tr1 Queuing Delay (tbuff) RD – RDmin ≈ tbuff x Congestion detected when tbuff >T Min Delay (RDmin) Relative Delay (RD) A TSval = tr1 RD = tr1 – ts1

  13. Summary of Algorithm • 1. Initial Receive Rate Estimation • 2. Buffer Management Mode • 3. Monitor Mode • a) Buffer Fill State • b) Buffer Drain State tbuff > T • Probe network • Details in paper • Send 10 packets • Estimate ρ using replies • Send Faster (σ >ρ) • SendSlower (σ <ρ) TCP-RRE (Receiver-Rate Estimation) tbuff < T Significantchanges in network

  14. Parameters? • How much faster or slower to send? • What threshold T to use? • When to switch to monitor state? • See details in paper

  15. ns-2 Evaluation Measured real networks to get simulation parameters

  16. ns-2 Evaluation • Single Download with Slow Uplink • Single Download under Normal Conditions • Download with Concurrent Upload • Handling Network Fluctuation • TCP Friendliness

  17. ns-2 Evaluation • Single Download with Slow Uplink • Single Download under Normal Conditions • Download with Concurrent Upload • Handling Network Fluctuation • TCP Friendliness

  18. Download with Slow Uplink 1MB Data ACK 200 Uplink Speed (kb/s) 20 8 1 Downlink Speed (Mb/s) ns-2

  19. Download with Slow Uplink Uplink speed decreases ns-2

  20. Download under Normal Conditions ns-2

  21. Download under Normal Conditions Packet Loss 1400+ ns-2

  22. Download under Normal Conditions ns-2

  23. Download with Concurrent Upload Download 1MB Upload until download completion 3000 Uplink Speed (kb/s) 250 8 0.5 ns-2 Downlink Speed (Mb/s)

  24. Download with Concurrent Upload 2.5X ns-2

  25. Evaluation in Linux • Several Places, different ISPs • Multiple times • CDF of all experiments at each place for each ISP Download 1MBwith TCP-RRE Download 1MB with CUBIC Upload until download completion

  26. Evaluation in Linux

  27. Evaluation in Linux +50%

  28. Evaluation in Linux

  29. Evaluation in Linux +25%

  30. Conclusion

  31. Conclusion • TCP-RRE • ACK Clocking • Rate Control with Feedback Loop • Use TCP Timestamp • Estimate Receive Rate • Detect Congestion • Improves TCP • Uplink is Slow • Uplink is Congested • Keep the Delay Low • Fair to Other TCP Flows

  32. Thank You Questions

  33. Handling Network Fluctuations CUBIC

  34. Handling Network Fluctuations TCP-RRE

  35. TCP Friendliness • Run two RSFC uploads concurrently • Calculate Jain fairness index:

  36. TCP Friendliness

More Related