Ns Simulation Final presentation Stella Pantofel 304026354 Igor Berman 313879942 Michael Halperin 317321982.

Slides:



Advertisements
Similar presentations
Martin Suchara, Ryan Witt, Bartek Wydrowski California Institute of Technology Pasadena, U.S.A. TCP MaxNet Implementation and Experiments on the WAN in.
Advertisements

TCP Variants.
TCP Vegas: New Techniques for Congestion Detection and Control.
CSIT560 Internet Infrastructure: Switches and Routers Active Queue Management Presented By: Gary Po, Henry Hui and Kenny Chong.
24-1 Chapter 24. Congestion Control and Quality of Service (part 1) 23.1 Data Traffic 23.2 Congestion 23.3 Congestion Control 23.4 Two Examples.
Transport Layer – TCP (Part2) Dr. Sanjay P. Ahuja, Ph.D. Fidelity National Financial Distinguished Professor of CIS School of Computing, UNF.
By Arjuna Sathiaseelan Tomasz Radzik Department of Computer Science King’s College London EPDN: Explicit Packet Drop Notification and its uses.
Experimental evaluation of TCP-L June 5, 2003 Stefan Alfredsson Karlstad University.
Improving TCP Performance over Mobile Ad Hoc Networks by Exploiting Cross- Layer Information Awareness Xin Yu Department Of Computer Science New York University,
Performance Improvement of TCP in Wireless Cellular Network Based on Acknowledgement Control Osaka University Masahiro Miyoshi, Masashi Sugano, Masayuki.
Simulating Large Networks using Fluid Flow Model Yong Liu Joint work with Francesco LoPresti, Vishal Misra Don Towsley, Yu Gu.
REM : Active Queue Management Sanjeewa Athuraliya, Victor H. Li Steven H. Low, Qinghe Yin Presented by Hwangnam Kim.
The War Between Mice and Elephants LIANG GUO, IBRAHIM MATTA Computer Science Department Boston University ICNP (International Conference on Network Protocols)
Congestion Control An Overview -Jyothi Guntaka. Congestion  What is congestion ?  The aggregate demand for network resources exceeds the available capacity.
Explicit Congestion Notification (ECN) RFC 3168 Justin Yackoski DEGAS Networking Group CISC856 – TCP/IP Thanks to Namratha Hundigopal.
XCP: Congestion Control for High Bandwidth-Delay Product Network Dina Katabi, Mark Handley and Charlie Rohrs Presented by Ao-Jan Su.
School of Information Technologies TCP Congestion Control NETS3303/3603 Week 9.
The War Between Mice and Elephants Presented By Eric Wang Liang Guo and Ibrahim Matta Boston University ICNP
Explicit Congestion Notification ECN Tilo Hamann Technical University Hamburg-Harburg, Germany.
AQM for Congestion Control1 A Study of Active Queue Management for Congestion Control Victor Firoiu Marty Borden.
TCP Stability and Resource Allocation: Part I. References The Mathematics of Internet Congestion Control, Birkhauser, The web pages of –Kelly, Vinnicombe,
1 Internet Networking Spring 2003 Tutorial 11 Explicit Congestion Notification (RFC 3168) Limited Transmit (RFC 3042)
High speed TCP’s. Why high-speed TCP? Suppose that the bottleneck bandwidth is 10Gbps and RTT = 200ms. Bandwidth delay product is packets (1500.
Low Delay Marking for TCP in Wireless Ad Hoc Networks Choong-Soo Lee, Mingzhe Li Emmanuel Agu, Mark Claypool, Robert Kinicki Worcester Polytechnic Institute.
1 TCP Transport Control Protocol Reliable In-order delivery Flow control Responds to congestion “Nice” Protocol.
1 Random Early Detection Gateways for Congestion Avoidance Sally Floyd and Van Jacobson, IEEE Transactions on Networking, Vol.1, No. 4, (Aug 1993), pp
1 Internet Networking Spring 2003 Tutorial 11 Explicit Congestion Notification (RFC 3168)
A Real-Time Video Multicast Architecture for Assured Forwarding Services Ashraf Matrawy, Ioannis Lambadaris IEEE TRANSACTIONS ON MULTIMEDIA, AUGUST 2005.
1 Emulating AQM from End Hosts Presenters: Syed Zaidi Ivor Rodrigues.
1 Spring Semester 2007, Dept. of Computer Science, Technion Internet Networking recitation #8 Explicit Congestion Notification (RFC 3168) Limited Transmit.
Random Early Detection Gateways for Congestion Avoidance
Simulation of Explicit Congestion Notification and Random Early Detection Nadav Amit Teif Dmitry Yurovsky Denis.
17/10/2003TCP performance over ad-hoc mobile networks. 1 LCCN – summer 2003 Uri Silbershtein Roi Dayagi Nir Hasson.
Submitters: Stella Pantofel Michael Halperin Igor Berman
Core Stateless Fair Queueing Stoica, Shanker and Zhang - SIGCOMM 98 Rigorous fair Queueing requires per flow state: too costly in high speed core routers.
Advanced Computer Networks : RED 1 Random Early Detection Gateways for Congestion Avoidance Sally Floyd and Van Jacobson, IEEE Transactions on Networking,
TCP: flow and congestion control. Flow Control Flow Control is a technique for speed-matching of transmitter and receiver. Flow control ensures that a.
Understanding the Performance of TCP Pacing Amit Aggarwal, Stefan Savage, Thomas Anderson Department of Computer Science and Engineering University of.
Principles of Congestion Control Congestion: informally: “too many sources sending too much data too fast for network to handle” different from flow control!
B 李奕德.  Abstract  Intro  ECN in DCTCP  TDCTCP  Performance evaluation  conclusion.
ACN: RED paper1 Random Early Detection Gateways for Congestion Avoidance Sally Floyd and Van Jacobson, IEEE Transactions on Networking, Vol.1, No. 4, (Aug.
1 Lecture 14 High-speed TCP connections Wraparound Keeping the pipeline full Estimating RTT Fairness of TCP congestion control Internet resource allocation.
Chapter 12 Transmission Control Protocol (TCP)
27th, Nov 2001 GLOBECOM /16 Analysis of Dynamic Behaviors of Many TCP Connections Sharing Tail-Drop / RED Routers Go Hasegawa Osaka University, Japan.
TCP Trunking: Design, Implementation and Performance H.T. Kung and S. Y. Wang.
TCP-Cognizant Adaptive Forward Error Correction in Wireless Networks
AQM & TCP models Courtesy of Sally Floyd with ICIR Raj Jain with OSU.
1 CS 4396 Computer Networks Lab TCP – Part II. 2 Flow Control Congestion Control Retransmission Timeout TCP:
CS640: Introduction to Computer Networks Aditya Akella Lecture 15 TCP – III Reliability and Implementation Issues.
We used ns-2 network simulator [5] to evaluate RED-DT and compare its performance to RED [1], FRED [2], LQD [3], and CHOKe [4]. All simulation scenarios.
Internet Networking recitation #11
TCP continued. Discussion – TCP Throughput TCP will most likely generate the saw tooth type of traffic. – A rough estimate is that the congestion window.
Analysis and Design of an Adaptive Virtual Queue (AVQ) Algorithm for AQM By Srisankar Kunniyur & R. Srikant Presented by Hareesh Pattipati.
Transport Layer3-1 Chapter 3 outline r 3.1 Transport-layer services r 3.2 Multiplexing and demultiplexing r 3.3 Connectionless transport: UDP r 3.4 Principles.
ECEN 619, Internet Protocols and Modeling Prof. Xi Zhang Random Early Detection Gateways for Congestion Avoidance Sally Floyd and Van Jacobson, IEEE Transactions.
Congestion Avoidance Created by M Bateman, A Ruddle & C Allison As part of the TCP View project.
1 Lecture 15 Internet resource allocation and QoS Resource Reservation Protocol Integrated Services Differentiated Services.
Transmission Control Protocol (TCP) TCP Flow Control and Congestion Control CS 60008: Internet Architecture and Protocols Department of CSE, IIT Kharagpur.
Other Methods of Dealing with Congestion
Internet Networking recitation #9
Topics discussed in this section:
COMP 431 Internet Services & Protocols
Chapter 6 Congestion Avoidance
Congestion Control: The Role of the Routers
Other Methods of Dealing with Congestion
IT351: Mobile & Wireless Computing
Other Methods of Dealing with Congestion
Internet Networking recitation #10
Project-2 (20%) – DiffServ and TCP Congestion Control
Transport Layer: Congestion Control
Presentation transcript:

Ns Simulation Final presentation Stella Pantofel Igor Berman Michael Halperin

Simulation of TCP Reno vs. TCP SACK over the network containing congested satellite or usual link, Using: Explicit Congestion Notification(ECN) and Random Early Detection(RED)

The outline of the project Using NS-2 simulator, we simulate a congested network that uses RED with ECN with following options: –TCP Reno / TCP SACK –Different RED threshold parameters. –Variable number of connections. –The network will have usual links with one of the links congested (either satellite or usual one). We will supply results (graphs) that compare the throughputs.

TCP SACK Overview Selective acknowledgement (SACK) is a TCP option designed for adopting the sender rate to the actual network conditions better. –Main purpose is increasing the throughput when multiple packets are lost from the same window. –Main idea: receiver tells the sender not only the next in-sequence expected byte, but also a range of bytes received out-of-order. It replies with a bit-mask of packets received rather than the last segment that was received in a continuity. SACK uses the Option field in the TCP header. –It is invoked only if both sides support it. Number of timeouts with SACK experienced during the connection is dramatically reduced.

Active Queue Management Queue management algorithms ’ purpose –manage the length of packet queues by dropping packets –detect congestion before the router queue overflows. Example: Random Early Detection(RED) –Drops packets when the buffer is full –Uses probabilistic drops of packets. It monitors the average queue size and randomly chooses connections to notify of the congestion –(average queue size) no packets are dropped. – (average queue size) > ( maximal threshold) => every arriving packet is dropped. –(minimal threshold) each arriving packet is dropped with a probability that is a function of the average queue size.

Explicit Congestion Notification (ECN) Routers may mark packets instead of dropping them. ECN-Capable Transport bit is set by the data sender to indicate that the end points of the transport protocol are ECN capable. CE (Congestion Experienced) bit is set by the router to indicate congestion to the end points. Upon the receipt of a single CE packet the congestion control algorithms followed at the end systems are the same as the response to the loss of a single packet.

Simulated Network 1.N sources permanently send ftp data to terminal2. 2.From each source there is one flow. 3.Single simulation lasts 60 seconds. 4.N is increased for next simulation, until it is 60.

Topology Definitions 1.When simulating congestion over regular links, delay of Terminal1- Satellite and Satellite-Terminal2 links is 5ms (instead of 125 ms). 2.Buffer capacity of Terminal packets. 3.Buffer capacity of Bridge – 100 packets. 4.Starting time for each connection varies uniformly between 0 and 1 seconds. 5.Bandwidth of each source’s link to Bridge is random and varies between 3 and 6Mb. 6.Same about delay – it varies between 3 and 6 ms.

Simulating satellite links in NS2 version 2.1b7a The available version of NS2 simulator (2.1b7a) doesn’t allow combining satellite links with regular links in one simulation. Therefore, we are trying to simulate satellite links using wired links with long propagation delay and high rate of loss error (if needed). To check validity of the above simulation we conducted several experiments over this topology.

Satellite link vs. Regular link with long delay With DropTail With RED

Satellite link vs. Regular link with long delay (con.) These graphs describe the throughputs of the satellite link vs. a regular link as a function of time. As we can see the throughput is almost identical. From this we conclude that despite minor differences, the overall behavior is very similar in both cases. Based on the results we believe that the use of regular link with long delay as a replacement for geo-stationary satellite link is valid, and that is what we did in our simulations.

TCP SACK vs. TCP Reno over satellite link and regular link with default RED parameters

TCP SACK vs. TCP Reno over satellite link and regular link with default RED parameters Results 1.The average improvement, over a satellite link, achieved by using TCP Sack vs. TCP Reno is around 5%. 2.The average improvement, over a regular link, achieved by using TCP Sack vs. TCP Reno is less than 0.5 %. 3.The average throughput over a regular link in both cases is higher by approximately 20%.

Red parameters 1.Formulas used in RED’s queue management algorithms Average queue size: Computation of probability for packet to be dropped/marked 2.The RED parameters we are playing with: min th (threshold_)- the minimum queue size threshold, default – 5. max th (maxthreshold_) - the maximum queue size threshold – default 15. w q (q_weight_) - the weight factor in computing the the average queue size - default max p (1/linterm) - the probability to be dropped when the average queue size equals max th – default 0.1 (linterm = 10)

Proposed Settings For RED Parameters 1.We increased the w q parameter relatively to its default value, so that it follows more closely the current queue size 2.max th was set to a considerably high value compared to queue buffer size. Once there is a heavy congestion, we are trying to absorb it, because sources will be aware of it only in a very long time as a result of a considerably large delay on the links. 3.min th was set to a considerably low value – increasing the chance for early “unforced” drop and slowing down some of connections before real congestion is experienced. 4.max p was decreased relatively to the default value, so that the rate of drops when the average queue size is between min th and max th is not too high.

Improving throughput of TCP Reno over satellite link by changing RED parameters

Improving throughput of TCP SACK over satellite link by changing RED parameters

TCP Reno vs. TCP Sack with default RED and changed parameters over satellite link

How errors harm the throughput?

1.Low loss rate of % doesn’t influence the throughput significantly, as only a few packets are dropped during the simulation (if any). 2.Medium loss rate of 0.1% causes the deterioration of the throughput when the number of sources is small. As the number of sources increases so is the throughput, because when any connection is affected by error, its load on the links is reduced and others can use available bandwidth. 3.High loss rate of 1% results in enormous degradation of throughput, and even increasing number of sources doesn’t allow to reach previous level.

Error of % Sack improves Reno by factor of This error rate means that on average only 1 packet out of is lost, and this is negligible when calculating average throughput over 60- second simulation. Thus,the results we got are very similar to the results without errors.

Error of 0.1% Sack improves Reno by factor of Compared to % loss rate, we can see that only at the end we are getting the same throughput, i.e. the more sources we have, the better they could backup each other in a case of losses.

Error of 1% Sack improves Reno by factor of In a case of 1% error rate we can observe severe decrease of throughput by more than 50%. However the pattern of improving as number of sources grows still exists.

Errors Tuning RED parameters helps even when we have errors on the links,however, when the loss rate is extremely high, changing Red parameters doesn’t help “much”. At the same time, as it can be observed from the last graph, relative improvement achieved by using SACK is slightly higher than it is usually. This is probably due to the fact that SACK allows to recover from several packet drops from the same window, and since we have 1% rate of losses there is a good chance that several packets from the same window will be actually dropped.

Conclusions From the presented graphs we can draw several conclusions: 1.TCP Sack improves the throughput and is preferable over TCP Reno. However, we must remember that there is a trade off when using TCP Sack: the processing time of a packet is longer than that of TCP Reno. 2.Red parameters: We observed that for given network it is worthwhile to change the default parameters of RED so they will suite the specific configuration of the network. It produces visible improvements in throughput. However, determining optimal values is not an easy task and depends on the characterization of the network traffic as well as on the physical characteristics of the network. 3.The best (by far) results are achieved while combining TCP SACK with Red parameters configured specifically for the given network. Not only RED and TCP SACK don’t interfere with each other but they are even improving each other.