E8 TCP. Politecnico di Milano Scuola di Ingegneria Industriale e dell Informazione

Similar documents
CSE 123: Computer Networks

Wireless Internet Exercises

cs/ee/ids 143 Communication Networks

Computer Networks ( Classroom Practice Booklet Solutions)

PIQI-RCP: Design and Analysis of Rate-Based Explicit Congestion Control

Leopold Franzens University Innsbruck. Responding to Spurious Loss Events in TCP/IP. Master Thesis. Institute of Computer Science

Reliable Data Transport: Sliding Windows

Min Congestion Control for High- Speed Heterogeneous Networks. JetMax: Scalable Max-Min

A Mathematical Model of the Skype VoIP Congestion Control Algorithm

Concise Paper: Deconstructing MPTCP Performance

Modelling Packet Delay in Ethernet and IP Networks

Modeling and Simulation NETW 707

TCP over Cognitive Radio Channels

IEEE abc-01/22r1

Impact of Cross Traffic Burstiness on the Packet-scale Paradigm An Extended Analysis

A Stochastic Model for TCP with Stationary Random Losses

Advanced Topicson Network Socket Programming

Socket Programming. Daniel Zappala. CS 360 Internet Programming Brigham Young University

Capturing Network Traffic Dynamics Small Scales. Rolf Riedi

A Quantitative View: Delay, Throughput, Loss

Lecture 7: Simulation of Markov Processes. Pasi Lassila Department of Communications and Networking

Processor Sharing Flows in the Internet

Lan Performance LAB Ethernet : CSMA/CD TOKEN RING: TOKEN

Rate adaptation, Congestion Control and Fairness: A Tutorial. JEAN-YVES LE BOUDEC Ecole Polytechnique Fédérale de Lausanne (EPFL)

A Globally Stable Adaptive Congestion Control Scheme for Internet-Style Networks with Delay 1

Stability Analysis of TCP/RED Communication Algorithms

Stability Analysis of TCP/RED Communication Algorithms

Performance Analysis of Priority Queueing Schemes in Internet Routers

Solutions to COMP9334 Week 8 Sample Problems

WiFi MAC Models David Malone

Analysis of random-access MAC schemes

A New TCP/AQM System Analysis

Analysis of TCP Westwood+ in high speed networks

CS276 Homework 1: ns-2

Chapter 5. Elementary Performance Analysis

Networked Control Systems

Modeling Impact of Delay Spikes on TCP Performance on a Low Bandwidth Link

Tuning the TCP Timeout Mechanism in Wireless Networks to Maximize Throughput via Stochastic Stopping Time Methods

Congestion Control In The Internet Part 1: Theory. JY Le Boudec 2018

Modelling an Isolated Compound TCP Connection

Window Size. Window Size. Window Size. Time. Time. Time

These are special traffic patterns that create more stress on a switch

FLAME Flexible Lightweight Active Measurement Environment

Time. Lakshmi Ganesh. (slides borrowed from Maya Haridasan, Michael George)

Markovian Model of Internetworking Flow Control

Modelling TCP with a Discrete Time Markov Chain

Appendix A Prototypes Models

Analysis on Packet Resequencing for Reliable Network Protocols

Internet Measurement and Data Analysis (7)

Unreliable Failure Detectors for Reliable Distributed Systems

Application of Link Layer FEC to DVB-S2 for Railway Scenarios

Analysis of Scalable TCP in the presence of Markovian Losses

IN THIS PAPER, we describe a design oriented modelling

A Stochastic Model of TCP/IP with Stationary Random Losses

Homework 1 - SOLUTION

Communications and Signal Processing Spring 2017 MSE Exam

NICTA Short Course. Network Analysis. Vijay Sivaraman. Day 1 Queueing Systems and Markov Chains. Network Analysis, 2008s2 1-1

Congestion Control. Topics

Performance analysis of IEEE WLANs with saturated and unsaturated sources

A model-based approach to control over packet-switching networks, with application to Industrial Ethernet

Service differentiation without prioritization in IEEE WLANs

Implementation of the IEEE 1588 Precision Time Protocol for Clock Synchronization in the Radio Detection of Ultra-High Energy Neutrinos

Modélisation d une connexion Compound TCP isolée

DYNAMIC BEHAVIOR OF TCP-LIKE WINDOW CONTROL AND THROUGHPUT PERFORMANCE

Electrical Engineering Written PhD Qualifier Exam Spring 2014

Stochastic Hybrid Systems: Applications to Communication Networks

Compound TCP with Random Losses

Outline. EECS Components and Design Techniques for Digital Systems. Lec 18 Error Coding. In the real world. Our beautiful digital world.

A Utility-Based Congestion Control Scheme for Internet-Style Networks with Delay

EECS Components and Design Techniques for Digital Systems. Lec 26 CRCs, LFSRs (and a little power)

Performance Analysis of the IEEE e Block ACK Scheme in a Noisy Channel

MPTCP is not Pareto-Optimal: Performance Issues and a Possible Solution

Giuseppe Bianchi, Ilenia Tinnirello

Understanding TCP Vegas: A Duality Model

Introduction to queuing theory

RESEARCH ON THE DISTRIBUTED PARALLEL SPATIAL INDEXING SCHEMA BASED ON R-TREE

Design and Analysis of a Propagation Delay Tolerant ALOHA Protocol for Underwater Networks

VISY-Quick Communication Protocol

Cross-layer Theoretical Analysis of NC-aided. Cooperative ARQ Protocols in Correlated Shadowed Environments (Extended Version)

Emulating Low-priority Transport at the Application Layer: A Background Transfer Service

M C. Analytical Methods for Network Congestion Control. Steven H. Low Engineering & Applied Science, Caltech

Nonlinear Instabilities in TCP-RED

Resource Allocation and Pricing. R. Srikant University of Illinois

DIMENSIONING BANDWIDTH FOR ELASTIC TRAFFIC IN HIGH-SPEED DATA NETWORKS

EP2200 Course Project 2017 Project II - Mobile Computation Offloading

384Y Project June 5, Stability of Congestion Control Algorithms Using Control Theory with an application to XCP

ATM VP-Based Ring Network Exclusive Video or Data Traffics

Internet Congestion Control: Equilibrium and Dynamics

The Analysis of Microburst (Burstiness) on Virtual Switch

Giuseppe Bianchi, Ilenia Tinnirello

Performance Effects of Two-way FAST TCP

Multiaccess Problem. How to let distributed users (efficiently) share a single broadcast channel? How to form a queue for distributed users?

Modeling and Stability of PERT

Data byte 0 Data byte 1 Data byte 2 Data byte 3 Data byte 4. 0xA Register Address MSB data byte Data byte Data byte LSB data byte

Elliptic Curves. Giulia Mauri. Politecnico di Milano website:

STABLE AND SCALABLE CONGESTION CONTROL FOR HIGH-SPEED HETEROGENEOUS NETWORKS

On the Resource Utilization and Traffic Distribution of Multipath. Transmission Control

Approximate Fairness with Quantized Congestion Notification for Multi-tenanted Data Centers

A Simple Throughput Model for TCP Veno

On Buffer Limited Congestion Window Dynamics and Packet Loss

Transcription:

E8 TP Politecnico di Milano Scuola di Ingegneria Industriale e dell Informazione

Exercises o onsider the connection in the figure A =80 kbit/s τ =0 ms R =? τ =? B o o o Host A wants to know the capacity and propagation delay of link and to this purpose sends to host B two echo messages: M length l=000 [byte], and Mlength l=500 [byte]; for each of them it measures the Round-Trip-Time (RTT) equal to 780 [ms] and 30 [ms] respectively. In the replies, B uses the same message lengths. alculate and τ assuming header legths negligible.

Solution R A B =80 kbit/s τ =0 ms =? τ =? + + + = + + + = τ τ τ τ m m RTT m m RTT

Solution A =80 Kbit/s τ =0ms R =? τ =? B 780 30 = 8000 80 000 = 80 + 0 + 8000 + τ 000 + 0 + 6000 = 0 + + τ + τ = 30 + 4000 + τ

Solution A =80 kbit/s τ =0ms R =? τ =? B τ = 80 8000 80 τ = = = 4000 8000 50 80 = + 80 3 8000 3 kbit/s = 30 ms 8000 = 8000 + 560;

Exercises o onsider the network in the figure where links are biderectional and with the same rate in both directions A τ R τ 3 o Between A and B a TP connection is active and already with a stable rate. o Assume MSS=50 [byte] and RVWND small than WND and equal to 4 segments R 3 τ 3 B

Exercises o alculate the time necessary to transfer a sequence of 04 [kbit] coming form upper layer (from transmission of first segment to reception of the last AK). Assume n Length Header IP: HIP n Length Header TP: HTP n Length Header lower layers: HLL n No error n No interfering traffic n AK length negligible o Which is the value of the window that would allow a continuous transmission on link?

Solution o 04 [kbit] are equal to 3000 [byte] and to 5 segments of 50 [byte] (MSS) o Each packet on the link has a total length of n L=MSS+ HIP + HTP + HLL

Solution a) RTT = L + L L τ + + τ + + 3 τ 3 If L 4 RTT L Ttot = (5 / 4) RTT + (4 ) = 3RTT + 3 L

Solution a) Otherwise transmission is continuous: 5 5 Ttot = 5 L + RTT

Solution 3 b) The minimum value of the window that allows continuous transmission can be calculated forcing RTT to be smaller that the transmission time of the window: RTT w L w L L L L + τ + + + + τ τ 3 = 3 5 5

Exercises 3 o onsider the connection in the figure between two hosts A and B. A has to transfer a sequence of 00 segments of maximum length using TP. alculate the time necessary assuming: n MSS=000 [bit] n Header lengths at all layers negligible n onnection is open by A and the length of control packets for connection setup is negligible n AK length neglible n SSTHRESH equal to 5 MSS

Solution 3 o T=000 [bit] / [Mb/s] = [ms] o RTT = 6. [ms] + T = 7. [ms] o Transmission is discontinuous until WT < RTT, so until W=8 MSS MSS 4 MSS 5 MSS 6 MSS 7 MSS TX continuous T transfer = 6.[ms] + 6 RTT + 75 T + 6. [ms] Set up First 5 MSS Last 75 MSS =30.6 [ms] AK last MSS

Exercises 4 o At time 0 a TP connection between host A and l host B is activated. alculate the time when transmission on link becomes continuous assuming n Header lengths negligible n Bidirectional and symmetric links n RWND = 4000 [byte] and SSTHRESH = 400 [byte] n MSS = 00 [byte] n AK size = setup control packet size = 0 [byte] n onnection open by host A o How much time does it takes to transfer a file of [kbyte] (from connection setup to reception of last AK)? (Note that: byte = 8 bit, kbyte= 000 byte = 8000 bit) A =5 Kb/s τ =5ms R =50 Kb/s τ =5ms R 3 =00 Kb/s τ 3 =5ms B

Solution 4 o o o o We have: n T =00 x 8 [bit] / 5 [kbit/s] =64 ms, T =½ T =3 ms, T 3 = ½ T = 6 ms n RTT= T + T +T 3 + (τ +τ +τ 3 )+ (Tack + Tack + Tack3) = 3.ms n T setup = (Tack + Tack + Tack3) + (τ +τ +τ 3 ) =.4 ms Link is the bottleneck and transmission is continuous when: n WT > RTT so n W > RTT/ T =3,3 The time transmission becomes continuous is n T c = T setup + 3 RTT =.4 [ms] +649.6 [ms] = 75 [ms]

Solution 4 o File to transfer is [kbyte], equal to 0 MSS. o The time necessary is: n T c = T setup + 4 RTT + 3 T =.5 [s] ongestion avoidance 4 MSS

Exercises 5 o o o A TP connection is used to transfer a file of 39.5 [kbyte] using the following parameters: n MSS=500 [byte] n RTT = 500 [ms] n timeout T = *RTT. Assume the initial conditions are: n RWND = [kbyte] n SSTHRESH = 8 [kbyte] n WND = 500 [byte] Assume also that: n There is an error at time 3 [s] (all segments are lost) n At time 4,5 [s] receiver indicates RWND = [kbyte] a) Draw the evolution over time of: n WND n SSTHRESH n RWND b) alculate the time to transfer the file.

Solution 5 We have: o # segments = 39,5 [Kbyte] / 500 [byte] = 79 MSS o RWND = [Kbyte] / 500 [byte] = 4 MSS o SSTHRESH = 8 [Kbyte] / 500 [byte] = 6 MSS o Time Out = s

Solution 5 30 5 RWND TimeOut 0 SSTHRESH 6 7 8 MSS 5 0 5 0 8 4 WND 4 4 0 3 4 5 6 7 8 9 Secondi 4 4 4 4 4 o Transfer time T=8.5s

Exercises 6 o onsider the following connection A τ R τ B o A has to transfer an application message of M bytes to B using UDP a) Assuming the maximum length of UDP segments is m bytes (data only), and denoting with HLL, HIP, HUDP header length of lower layers, IP and UDP respectively, calculate the time necessary to transfer the message

Exercises 6 A τ R τ B b) As in a) but assuming on link is active a layer protocol with ARQ mechanism of stop-and-wait type (AK length negligible)

Solution 6 A τ R τ B o # of maximum length segments: M n = m o Length of last segment: l = M M m m

Solution 6 A τ R τ B a) Transfer time: if l + h m + h n m + h + l + h l + h + τ + +τ

Solution 6 A τ R τ B a) Transfer time: if l + h m + h n m + h m + h l + h + τ + + +τ

Solution 6 A τ R τ B b) Transfer time: m + h if m + h τ + m + h m + h + τ τ + l h + + n + + τ

Solution 6 A τ R τ B b) Transfer time: n if m + h m + h l + h + τ + m + h l + h + τ + +τ

Notations and Measure units o [byte] = 8 [bit] o [kbyte] = 000 [byte] = 8000 [bit] o [Mbyte] = 8 [Mbit] o [ms] = 0-3 [s] o [µs] = 0-6 [s] o [ns] = 0-9 [s]