1 / 28

Analysis of TCP Transmission Dynamics Under Loss Probability and Timeout Models

This document explores the dynamics of TCP transmission rates in relation to packet loss probability and timeout scenarios. It presents deterministic and stationary models, highlighting the relationship between congestion window (cwnd), round-trip time (RTT), and data rates. It discusses key metrics such as the maximum drops, total packets sent, and distribution of cwnd modeled as a negative binomial distribution. Additionally, the influence of timeouts on TCP flow rates is examined, providing insights into slow-start models and their impact on network performance.

Télécharger la présentation

Analysis of TCP Transmission Dynamics Under Loss Probability and Timeout Models

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. TCP Models Objective Given the loss probability, how fast does TCP send? Deterministic model?

  2. drops wmax cwnd wmax/2 time Wmax/2* RTT Simple Stationary model Data rate = cwnd/RTT Total packet sent=

  3. drops wmax cwnd wmax/2 time Wmax/2* RTT Simple Stationary model

  4. More complicated model

  5. The PDF of the cwnd for a Simplified TCP 0.8 0.7 =1 0.6 The mth moment around the origin scales like -m/2, i.e., 0.5 =0.1 p(cwnd) 0.4 0.3 =0.05 =0.01 0.2 =0.005 0.1 C1 =1.3, C2 = 2.0, C3 = 3.5, C4 = 7.1, … 0 0 5 10 15 20 25 30 cwnd The median scales = 1.2/1/2

  6. Distribution of cwnd Cwnd is nearly distributed according to the negative binomial distribution Gamma function is factorial if argument is an integer Where: c1~sqrt(3/2) E(w²)-E(w)²=(γ/δ) γ≈0.3

  7. 1%

  8. 5% 10%

  9. Time-out model Rate of going to time-out =

  10. Timeout model If a retransmission is not dropped (only if 1 and 2 didn’t apply). Total rate of entering timeout is: = ’ + ’’

  11. Time-out . Let I₁(t), denote the rate that flows enter timeout at time t

  12. Time out prob

  13. Dynamics of cwnd if proper stochastic calculus is applied, the correct dynamics for the mean are Approximately: SDE gives Using:

  14. Models of slow-start are in the works

More Related