1 / 9

Project Overview

This project aims to verify and analyze the TFRC congestion control protocol, conduct experiments to check for any shortcomings, study the loss model, and develop optimizations for TFRC. A comparison of TCP, TFRC, and UDP will also be done to understand the importance of loss as a network metric.

viramontes
Download Presentation

Project Overview

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. Project Overview • Study of Tfrc • Verification, Analysis and Development Verification : Experiments. Analysis : Check for short comings if any. Study of loss model. Development : Use of Tfrc within our environment (optimizations if any) . Comparison of TCP, TFRC and UDP to understand the concept of loss and its importance as a network metric. -Mayukh, clemson university

  2. Tfrc • What is it? An equation based congestion control protocol designed to compete ‘fairly’ with TCP under varying network environments. TCP friendly : A flow that in steady state uses no more bandwidth than a conformant TCP running under comparable conditions Why does it need to be friendly ? Majority of flows are TCP flows (Reno, SACK). Any new protocol that is developed needs to keep this in mind. Tfrc is not intended to replace TCP , just complement it. Why do we need it ? Ideal for unicast streaming multimedia (there is a version of Tfrc now intended for multicast services, but that is outside the scope of this project). Any applications that require a smooth and controlled sending rate. -Mayukh, clemson university

  3. Tfrc (contd ..) • How does it work ? Uses a throughput equation to calculate the sending rate. The receiver sends feedback reports to the server. A typical feedback message has an estimation of the loss event rate (p) and the receiving rate. (X_recv). The control equation is a function of p and Round trip time R. See RFC 3448 for protocol specs. (yes its an RFC now !) Points of interest : The throughput equation used Calculation of the loss event rate. (Use of Full Averaged Loss Interval method) -Mayukh, clemson university

  4. Results There has been a lot of interesting work done on TFRC. Extensive experiments under various conditions have been conducted , below is a gist of the results. 1> Tfrc does compete fairly with TCP : This was one of the primary design objectives and it has been shown under varying conditions that Tfrc is indeed TCP friendly. 2> It has a smooth sending rate. The purpose was to make Tfrc respond more gently to congestion events (rather than cutting sending rate in half for a single packet loss like Tcp) 3> Responds slowly to congestion events. A fallout of making Tfrc generate a consistent smooth sending rate. See 1> “Modeling TCP throughput : A simple Model and its Empirical Validation ‘ 2> “Equation based congestion control for Unicast Applications “ for more details. -Mayukh, clemson university

  5. My stuff this sem. • What am I supposed to do ?? Build on the code. Get it operational in our lab. Conduct Experiments. Study the loss rate. Objective 1 : Verify that what we have maps to the ICIR code and to the protocol specs. Method : Our Tfrc Vs ICIR Tfrc Vs Tcp. Using the network emulator in our lab run a series of tests comparing performance of our Tfrc versus the rest. (Its important we have a correct version of Tfrc before we proceed!!) -Mayukh, clemson university

  6. My stuff 2 .. Objective 2 : Once we have a confirmed running version of Tfrc, we are free to conduct various other interesting experiments. Make certain that the code has been integrated into the metricTool completely. Simulated network environments : Im hoping to use Ns. Use of SURGE. To observe the behavior of Tfrc in a ‘simulated internet’ (multiple users, short lived tcp flows etc. ) -Mayukh, clemson university

  7. My stuff 3… • Objective 3 : Study of the loss model. Using the experiments as a basis and the generated data, study the loss rates . Compare loss ratios of Tfrc, Tcp and Udp. At the end of this semester should have a paper ready presenting the results of our experiments and the inferences gained. -Mayukh, clemson university

  8. Thesis work. • Long term : Possible directions : a) Continue with current project : Hopefully at the end we will have a better understanding of congestion control, and the loss model. Direct my thesis towards a better understanding of Tfrc and related loss models. Suggestions for better ways of obtaining the loss model OR congestion control mechanisms OR extensions to TFRC. -Mayukh, clemson university

  9. Thesis .. b. Extension of current project . Tie in the concept of loss with the security aspect, denial of service. Depending on my productivity this is what I envision myself to be doing a year from now. Relate the study of the loss models and tie them in with the detection of denial of service attacks. Study of patterns in loss event on a per flow basis (and aggregated flow s) along with other information (typical attack signatures) to detect and thwart typical distributed denial of service attacks. -Mayukh, clemson university

More Related