Dynamic versus Static Traffic Policing: A New Approach for Videoconference Traffic over Wireless Cellular Networks Author: Polychronis Koutsakis Anukrati.

Slides:



Advertisements
Similar presentations
Capacity of wireless ad-hoc networks By Kumar Manvendra October 31,2002.
Advertisements

1 Adaptive Bandwidth Allocation in TDD-CDMA Systems Derek J Corbett & Prof. David Everitt The University of Sydney.
© NOKIAProduced as informative material for 3GPP RAN WG1 meeting No. 2 Downlink Shared Channel - DSCH DSCH associated with a dedicated channel (DCH) Downlink.
A Centralized Scheduling Algorithm based on Multi-path Routing in WiMax Mesh Network Yang Cao, Zhimin Liu and Yi Yang International Conference on Wireless.
Simulation Study on the Effect of the trTCM Parameters Hakyong KIM R&D Center, Corecess Inc. Tel : PCS : Mail :
1 CONGESTION CONTROL. 2 Congestion Control When one part of the subnet (e.g. one or more routers in an area) becomes overloaded, congestion results. Because.
TELE202 Lecture 8 Congestion control 1 Lecturer Dr Z. Huang Overview ¥Last Lecture »X.25 »Source: chapter 10 ¥This Lecture »Congestion control »Source:
William Stallings Data and Computer Communications 7 th Edition Chapter 13 Congestion in Data Networks.
Technical Seminar Presentation Presented by :- Plaban Kumar Rout (EC ) “Wireless ATM & Congestion Control ” Presented by Plaban Kumar Rout.
Traffic Shaping Why traffic shaping? Isochronous shaping
CS Spring 2009 CS 414 – Multimedia Systems Design Lecture 16 – Multimedia Transport Subsystem (Part 3) Klara Nahrstedt Spring 2009.
1.  Congestion Control Congestion Control  Factors that Cause Congestion Factors that Cause Congestion  Congestion Control vs Flow Control Congestion.
Engineering Internet QoS
Jaringan Komputer Lanjut Traffic Management Aurelio Rahmadian.
Receiver-driven Layered Multicast S. McCanne, V. Jacobsen and M. Vetterli University of Calif, Berkeley and Lawrence Berkeley National Laboratory SIGCOMM.
EECB 473 Data Network Architecture and Electronics Lecture 3 Packet Processing Functions.
Priority Scheduling and Buffer Management for ATM Traffic Shaping Authors: Todd Lizambri, Fernando Duran and Shukri Wakid Present: Hongming Wu.
Presented by Santhi Priya Eda Vinutha Rumale.  Introduction  Approaches  Video Streaming Traffic Model  QOS in WiMAX  Video Traffic Classification.
1 “Multiplexing Live Video Streams & Voice with Data over a High Capacity Packet Switched Wireless Network” Spyros Psychis, Polychronis Koutsakis and Michael.
Receiver-driven Layered Multicast S. McCanne, V. Jacobsen and M. Vetterli SIGCOMM 1996.
Presented By: Pariya Raoufi. Motivations Future applications require: higher bandwidth, generate a heterogeneous mix of network traffic, low latency.
Quality of Service in IN-home digital networks Alina Albu 23 October 2003.
First Generation (1G) Alison Griffiths C203 Ext:3292
Two Media Access Control Protocols (MAC) for Wireless Networks.
1 Short-term Fairness for TCP Flows in b WLANs M. Bottigliengo, C. Casetti, C.-F. Chiasserini, M. Meo INFOCOM 2004.
Long Range Dependent Traffic and Leaky Buckets CS215-winter ’01 Demetrios Laios 3/22/2001.
“On the Integration of MPEG-4 streams Pulled Out of High Performance Mobile Devices and Data Traffic over a Wireless Network” Spyros Psychis, Polychronis.
Traffic Management & QoS. Quality of Service (QoS) J The collective effect of service performances which determine the degree of satisfaction of a user.
Divert: Fine-grained Path Selection for Wireless LANs Allen Miu, Godfrey Tan, Hari Balakrishnan, and John Apostolopoulos MIT Computer Science and Artificial.
Gursharan Singh Tatla Transport Layer 16-May
Channel Allocation Schemes Comparison for Priority Wireless Mobile Networks By Dimas Gutierrez.
1 Computer Communication & Networks Lecture 4 Circuit Switching, Packet Switching, Delays Waleed.
A Distributed Scheduling Algorithm for Real-time (D-SAR) Industrial Wireless Sensor and Actuator Networks By Kiana Karimpour.
Univ. of TehranAdv. topics in Computer Network1 Advanced topics in Computer Networks University of Tehran Dept. of EE and Computer Engineering By: Dr.
Improving QoS Support in Mobile Ad Hoc Networks Agenda Motivations Proposed Framework Packet-level FEC Multipath Routing Simulation Results Conclusions.
1 Why traffic shaping? yIn packet networks that implement resource sharing xadmission control and scheduling alone are insufficient users may attempt to.
Chapter 24. Congestion Control and Quality of Service part 3
William Stallings Data and Computer Communications 7 th Edition Chapter 1 Data Communications and Networks Overview.
CS 447 Network & Data Communication QoS (Quality of Service) & DiffServ Introduction Department of Computer Science Southern Illinois University Edwardsville.
Univ. of TehranAdv. topics in Computer Network1 Advanced topics in Computer Networks University of Tehran Dept. of EE and Computer Engineering By: Dr.
Distribution of Multimedia Data Over a Wireless Network (DMDoWN): An Introduction Presented By: Rafidah Md Noor Faculty of Computer Science & Information.
November 4, 2003APOC 2003 Wuhan, China 1/14 Demand Based Bandwidth Assignment MAC Protocol for Wireless LANs Presented by Ruibiao Qiu Department of Computer.
Chi-Cheng Lin, Winona State University CS 313 Introduction to Computer Networking & Telecommunication Medium Access Control Sublayer.
Multimedia Wireless Networks: Technologies, Standards, and QoS Chapter 3. QoS Mechanisms TTM8100 Slides edited by Steinar Andresen.
Header Compression over Cellular LinksLars-Erik Jonsson, Header Compression for IP-Telephony over Cellular Links Lars-Erik Jonsson (Ericsson.
Slide title In CAPITALS 50 pt Slide subtitle 32 pt Dynamic and Persistent Scheduling for Voice over IP Traffic in the Long-Term Evolution Uplink Master’s.
Vertical Optimization Of Data Transmission For Mobile Wireless Terminals MICHAEL METHFESSEL, KAI F. DOMBROWSKI, PETER LANGENDORFER, HORST FRANKENFELDT,
Uplink Scheduling with Quality of Service in IEEE Networks Juliana Freitag and Nelson L. S. da Fonseca State University of Campinas, Sao Paulo,
TCP-Cognizant Adaptive Forward Error Correction in Wireless Networks
ECS5365 Lecture 6 ATM Traffic and Network Management
Improving TCP Performance over Wireless Networks
A Novel APSD Scheduler for WLAN IEEE e COMMUNICATION SYSTEMS, NETWORKS AND DIGITAL SIGNAL PROCESSING University of Lecce Speaker: Sebastiano Elia.
An Energy Efficient MAC Protocol for Wireless LANs, E.-S. Jung and N.H. Vaidya, INFOCOM 2002, June 2002 吳豐州.
Francine Lalooses David Lancia Arkadiusz Slanda Donald Traboini
Queuing Delay 1. Access Delay Some protocols require a sender to “gain access” to the channel –The channel is shared and some time is used trying to determine.
Quality of Service Schemes for IEEE Wireless LANs-An Evaluation 主講人 : 黃政偉.
Multimedia Synchronization I. Fatimah Alzahrani. Definitions Multimedia System : A system or application that supports the integrated processing of several.
On Exploiting Diversity and Spatial Reuse in Relay-enabled Wireless Networks Karthikeyan Sundaresan, and Sampath Rangarajan Broadband and Mobile Networking,
CHANNEL ALLOCATION FOR SMOOTH VIDEO DELIVERY OVER COGNITIVE RADIO NETWORKS Globecom 2010, FL, USA 1 Sanying Li, Tom H. Luan, Xuemin (Sherman) Shen Department.
Mitigating Congestion in Wireless Sensor Networks Bret Hull, Kyle Jamieson, Hari Balakrishnan MIT Computer Science and Artificial Intelligence Laborartory.
CS Spring 2011 CS 414 – Multimedia Systems Design Lecture 16 – Multimedia Transport (Part 2) Klara Nahrstedt Spring 2011.
Providing QoS in IP Networks
Chapter 10 Congestion Control in Data Networks and Internets 1 Chapter 10 Congestion Control in Data Networks and Internets.
Tel Hai Academic College Department of Computer Science Prof. Reuven Aviv Markov Models for data flow In Computer Networks Resource: Fayez Gebali, Analysis.
Topics discussed in this section:
Wireless ATM & Congestion Control
CONGESTION CONTROL.
Analysis and Evaluation of a New MAC Protocol
Lecture 23 WCDMA (Part III) Dr. Ghalib A. Shah
Presentation transcript:

Dynamic versus Static Traffic Policing: A New Approach for Videoconference Traffic over Wireless Cellular Networks Author: Polychronis Koutsakis Anukrati Gupta / Manoj Maskara CS: Mobile Computing, Fall 2009

2 Outline Introduction Policing Mechanism System Model Model Definition Channel Error Model The Token Bucket – Basic, Dual & Triple Jumping Window Sliding Window Conclusions

3 Introduction…1/2 Video traffic is becoming a major portion of the traffic carried over wired and wireless networks Video traffic calls for: –New sets of control procedures –Strict Qos requirements –Strict packet delay requirements –Strict packet dropping requirements Video traffic is modeled using various statistical models like – lognormal, Gamma, and a hybrid Gamma/lognormal distribution Videoconference traffic needs to be modeled differently – reason?

4 Introduction…2/2 Inherent Characteristics of Videoconference – very high autocorrelation Video coding standards – H.261, H.262, H.263, H.264 Latest version H.264 Widely used H.263 – Compresses moving picture component of audio/video services at low bit rates. “This is the first work in the literature which addresses the problem of modeling single H.263 videoconference traces” – Koutsakis

5 Policing Mechanism…1/2 Static and Dynamic – based on traffic flow Token Bucket and its variations Jumping Window Sliding Window (Moving Window)

6 Policing Mechanism…2/2 Focus on traffic control mechanism at the entrance of the system to prevent congestion Implemented all three traffic policing mechanism and proposed modification to improve performance Used GBAR process (gamma distribution)

7 System Model….1/3 MSC Parking Lot BS MS BS MS The study focuses on one cell of the network, also called picocell Source terminals are spatially dispersed and share a radio channel Radio channel connects the source terminals to the fixed Base Station (BS) The BS allocates channel resources, delivers feedback information, and provides an interface to the Mobile Switching Center (MSC) The MSC provides access to the fixed network infrastructure

8 System Model….2/3 Downlink wireless channel (20Mbps) – BS to MS Divided into time frames of equal length Each frame has a duration of 12ms and 566 information slots Information slot contains 1 fixed length packet with a header & information 3% or 16 slots (control Interval) of the bandwidth is used for uplink

9 System Model….3/3 The information Interval of the frame = Mbps needs efficient traffic policing mechanism Used computer simulations to study the performance of the traffic policing mechanisms Used C programming language to conduct simulations Each simulation point is the result of an average of 10 independent runs

10 Model Definition…1/2 Video conference traffic is modeled using GBAR (1) model as it takes advantage of the properties of Pearson V distribution GBAR (1) model is based on following results shape parameter scale parameter

11 Model Definition…2/2 If X n-1 = Ga(β, λ), and A n = Be(α, β- α) and B n = Ga(β- α, λ), and all are mutually independent, then GBAR(1) process is defined by {X n } where X n = A n X n-1 + B n Since the current value is determined by only one previous value, this is an autoregressive process of order 1

12 Pearson V Distribution The parameters (α, β) of the pearson V is computed using GBAR(1) mean and variance Mean = β/(α – 1) Variance = β 2 /[(α – 1) 2 (α – 2)] Pearson V distribution (Z n ) is also known as the inverted gamma distribution which is defined as Z n = 1/ X n

13 Traces Used A video stream from Office Cam A video stream from Lecture Cam A video stream from Parking Cam A video stream from N3 Talk A video stream from ARD Talk

14 Comparison Burstiness = peak/mean

15 Channel Error Model…1/2 15 States Markov Model Diagram

16 Channel Error Model…2/2 State S 0 represents “good state” and all other states represents “bad states” When channel is in S 0 it can either remain in S 0 or can transition to state S 1 When channel is in bad state it can transition to either the next higher state or back to state S 0 based on the received packet p good is due to very strict QoS requirements

17 Concepts to help understand Token approach Mean rate—Also called the committed information rate (CIR), it specifies how much data can be sent or forwarded per unit time on average. Burst size—Also called the committed burst (Bc) size, it specifies for each burst how much data can be sent within a given time without creating scheduling concerns. Time interval—Also called the measurement interval, it specifies the amount of time in seconds per burst. mean rate = burst size / time interval

18 Token Bucket Basic Approach Static Approach Conforming –> Size of Token = Packet Size Non Conforming –> Size of Token < Packet Size Packets either wait or discarded Max no. of Tokens in bucket = Size of Bucket Overflow – New incoming tokens are discarded Token

19 The Dual Token Bucket Approach Proposed method of generating tokens with the use of Pearson V-based variant of the GBAR model Token

20 Packet Loss Comparison Static versus Dual Token Bucket bucket size equal to the declared peak frame size of the source Increase in bucket size results in decrease in video packet dropping and the difference between the static approach and dynamic method decreases.

21 High packet dropping - Solutions Three solutions to deal with high packet dropping rate in dual token bucket as seen in previous slide. Sol 1 - Infinite buffer size Third bucket is introduced for other two solutions Sol 2 - Use Triple Token Bucket where third bucket is of fixed size Sol 3 - Use Triple Token Bucket where third bucket is of dynamic nature whose size varies depending on source transmission rate – DYNAMIC

22 The Triple Token Bucket Approach Behavior of Bucket 1 & Bucket 2 is similar to Dual Bucket Scheme. When Bucket 2 is full, it start marking packets with “non- conform” All 3 buckets are connected in parallel

23 Properties of 3 rd Bucket Responsible for source packet drop Token rate = average size of “token frame” Token frame is equal to the average no. of token needed in each video frame for the source to lose only 0.01% of packets In case of dynamic bucket size varies: –If source transmit at higher rate than its mean, size of bucket is reduced in order to prevent long overuse of network resources by malicious users –If source transmit at lower rate, size of bucket is increased to help accommodate anticipated bursts

24 Function of Buckets in Different Approaches Dual Token Bucket Approach Triple Token Bucket Approach Bucket 1Same as Token Bucket basic approach Bucket 2Once full, packets are “discarded” Once full, packets are marked as “non-conform” Bucket It handles packets that are marked “non-conform” Static Bucket Size Dynamic Bucket Size

25 Jumping Window Size of window is of fixed length T, placed side by side through time During a window only K bytes (or packets) are submitted by the source to the network Source transmits > K bytes = Packets are dropped/marked non conformed Mechanism is implemented using Token Counter In each new window, packet counter starts with initial value of zero

26 Modification in Jumping Window Token Bucket Jumping Window Reason: Significantly larger window size (40 frames, 3.2 seconds) If less than K bytes are transmitted within a window then token counter is not set to 0 It starts with an initial value equal to the remaining tokens

27 Sliding Window Similar to Jumping Window Difference – Each video frame size is remembered for the width of exactly one window, starting with specific video frame and ending T frames later This mechanism can be interpreted as a window which steadily moves along the time axis, with the requirement that the frame size of T frames are stored for the duration of one window Complexity is directly related to the window size. Strictest bandwidth enforcement. > Jumping Window

28 Modification of Sliding Window To implement more dynamic policy: in case of less than K bytes transmitted by the source within one window W, the tokens left in the bucket are not discarded, but they are added to the token bucket of the next window (W+1)

29 Conclusion Dynamic approach provides significantly better results in policing the burstiness of video traffic sources than static traffic policing mechanisms Proposed triple token bucket scheme, which showed the best performance Showed results using various example and studied the Token Bucket, Jumping Window and Sliding Window schemes Low paying users – Triple Token Bucket or Triple Sliding Window High paying users – Triple Jumping Window Token Window less complex to implement rather than Sliding Window but is more sticker. Videoconference applications are very greedy in terms of bandwidth requirements

30 Thanks and wish you all Good Luck for exam next week !!!