Top Banner
Chinese University, CSE Dept. Distributed Systems / 11 - 1 Distributed Systems Topic 11: Distributed Multimedia Systems Dr. Michael R. Lyu Computer Science & Engineering Department The Chinese University of Hong Kong
37

© Chinese University, CSE Dept. Distributed Systems / 11 - 1 Distributed Systems Topic 11: Distributed Multimedia Systems Dr. Michael R. Lyu Computer.

Jan 05, 2016

Download

Documents

Elizabeth Ross
Welcome message from author
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
Page 1: © Chinese University, CSE Dept. Distributed Systems / 11 - 1 Distributed Systems Topic 11: Distributed Multimedia Systems Dr. Michael R. Lyu Computer.

© Chinese University, CSE Dept. Distributed Systems / 11 - 1

Distributed Systems

Topic 11:

Distributed Multimedia Systems

Dr. Michael R. LyuComputer Science & Engineering Department

The Chinese University of Hong Kong

Page 2: © Chinese University, CSE Dept. Distributed Systems / 11 - 1 Distributed Systems Topic 11: Distributed Multimedia Systems Dr. Michael R. Lyu Computer.

© Chinese University, CSE Dept. Distributed Systems / 11 - 2

1 Outline

1. Introduction

2. Characters of multimedia data

3. Quality of service management

4. Resource management

5. Stream adaptation

6. Summary

Page 3: © Chinese University, CSE Dept. Distributed Systems / 11 - 1 Distributed Systems Topic 11: Distributed Multimedia Systems Dr. Michael R. Lyu Computer.

© Chinese University, CSE Dept. Distributed Systems / 11 - 3

1 Introduction

Media:– The term media refers to the storage,

transmission, interchange, presentation, representation and perception of different information types, such as text, graphics, voice, audio and video.

Multimedia– The term multimedia is to denote the property of

handling a variety of media representation in an integrated manner.

Page 4: © Chinese University, CSE Dept. Distributed Systems / 11 - 1 Distributed Systems Topic 11: Distributed Multimedia Systems Dr. Michael R. Lyu Computer.

© Chinese University, CSE Dept. Distributed Systems / 11 - 4

1 Introduction

Most multimedia is inherently time-based – the arrival time and arrival order of data packets is important

The Internet guarantees neither when transmitting data

We don’t just want interactive multimedia over our networks… we want it to be reliable and high-quality

A distributed multimedia system can come to the rescue

Page 5: © Chinese University, CSE Dept. Distributed Systems / 11 - 1 Distributed Systems Topic 11: Distributed Multimedia Systems Dr. Michael R. Lyu Computer.

© Chinese University, CSE Dept. Distributed Systems / 11 - 5

1 A Distributed Multimedia System

Wide area gateway Videoserver

DigitalTV/radioserver

Video cameraand mike

Local network Local network

Page 6: © Chinese University, CSE Dept. Distributed Systems / 11 - 1 Distributed Systems Topic 11: Distributed Multimedia Systems Dr. Michael R. Lyu Computer.

© Chinese University, CSE Dept. Distributed Systems / 11 - 6

1.1 Multimedia in A Mobile Environment

*

Page 7: © Chinese University, CSE Dept. Distributed Systems / 11 - 1 Distributed Systems Topic 11: Distributed Multimedia Systems Dr. Michael R. Lyu Computer.

© Chinese University, CSE Dept. Distributed Systems / 11 - 7

1.1 History

60s-70s: Distributed computing research with earliest networks

80s: Compact disc, personal computer explosion

80s-90s: Distributed multimedia system research (video conferencing, et al)

90s: Current prevalent paradigm (quality of service management)

Page 8: © Chinese University, CSE Dept. Distributed Systems / 11 - 1 Distributed Systems Topic 11: Distributed Multimedia Systems Dr. Michael R. Lyu Computer.

© Chinese University, CSE Dept. Distributed Systems / 11 - 8

1.1 Multimedia Application Samples

Web-based multimedia: It provides best-effort access to streams of audio and video data published via web.

Network phone and audio conferencing: It has relatively low bandwidth requirements, especially when efficient compression techniques are used

Video-on-demand services: These supply video information in digital form, retrieving the data from large online storage systems and delivering them to the end-user’s display

Page 9: © Chinese University, CSE Dept. Distributed Systems / 11 - 1 Distributed Systems Topic 11: Distributed Multimedia Systems Dr. Michael R. Lyu Computer.

© Chinese University, CSE Dept. Distributed Systems / 11 - 9

1.2 The Window of Scarcity

When dealing with large audio and video streams, many systems are constrained in the quantity and quality of streams they can support.

This situation has need depicted as the window of scarcity.

1980 1990

remotelogin

networkfile access

high-qualityaudio

interactivevideo

insufficientresources

scarceresources

abundantresources

2000

The window of scarcity for computing and communication resources

Page 10: © Chinese University, CSE Dept. Distributed Systems / 11 - 1 Distributed Systems Topic 11: Distributed Multimedia Systems Dr. Michael R. Lyu Computer.

© Chinese University, CSE Dept. Distributed Systems / 11 - 10

2 Characteristics of Multimedia Applications

Large quantities of continuous data Timely and smooth delivery is critical Interactive multimedia applications require low round-trip

delays Need to co-exist with other applications Reconfiguration is a common occurrence Resources required:

– Processor cycles in workstations and servers– Network bandwidth (+ latency)– Dedicated memory– Disk bandwidth (for stored media)

Page 11: © Chinese University, CSE Dept. Distributed Systems / 11 - 1 Distributed Systems Topic 11: Distributed Multimedia Systems Dr. Michael R. Lyu Computer.

© Chinese University, CSE Dept. Distributed Systems / 11 - 11

2 Characteristics of Multimedia Streams

Data rate(approximate)

Sample or frame frequency size

Telephone speech 64 kbps 8 bits 8000/secCD-quality sound 1.4 Mbps 16 bits 44,000/secStandard TV video(uncompressed)

120 Mbps up to 640 x 480pixels x 16 bits

24/sec

Standard TV video (MPEG-1 compressed)

1.5 Mbps variable 24/sec

HDTV video(uncompressed)

1000–3000 Mbps up to 1920 x 1080pixels x 24 bits

24–60/sec

HDTV videoMPEG-2 compressed)

10–30 Mbps variable 24–60/sec

Page 12: © Chinese University, CSE Dept. Distributed Systems / 11 - 1 Distributed Systems Topic 11: Distributed Multimedia Systems Dr. Michael R. Lyu Computer.

© Chinese University, CSE Dept. Distributed Systems / 11 - 12

3 Quality of Service (QoS) Management

Simplicity in and of itself: We want and need high quality, reliable, interactive multimedia

The general Internet structure is not sufficient to accomplish this

A distributed multimedia system will add protocols and architectures on top of the Internet (or LAN) to guarantee quality levels, thereby satisfying our need

Page 13: © Chinese University, CSE Dept. Distributed Systems / 11 - 1 Distributed Systems Topic 11: Distributed Multimedia Systems Dr. Michael R. Lyu Computer.

© Chinese University, CSE Dept. Distributed Systems / 11 - 13

3 Infrastructure Components for Multimedia Applications

Microphones

Camera

Screen

Window system

CodecD

BMixer

PC/workstation PC/workstation

CVideostore

Networkconnections

K

L

M

: multimedia stream

CodecA G

Codec

H

Window system

White boxes represent media processing components, many of which are implemented in software, including:codec: coding/decoding filter

mixer: sound-mixing component

Video file system

Page 14: © Chinese University, CSE Dept. Distributed Systems / 11 - 1 Distributed Systems Topic 11: Distributed Multimedia Systems Dr. Michael R. Lyu Computer.

© Chinese University, CSE Dept. Distributed Systems / 11 - 14

3 QoS Specifications for Application Components

Component Bandwidth Latency Loss rate Resources required

Camera Out: 10 frames/sec, raw video640x480x16 bits

Zero

A Codec In:Out:

10 frames/sec, raw videoMPEG-1 stream

Interactive Low 10 ms CPU each 100 ms;10 Mbytes RAM

B Mixer In:Out:

2 44 kbps audio1 44 kbps audio

Interactive Very low 1 ms CPU each 100 ms;1 Mbytes RAM

H Windowsystem

In:Out:

various50 frame/sec framebuffer

Interactive Low 5 ms CPU each 100 ms; 5 Mbytes RAM

K Networkconnection

In/Out: MPEG-1 stream, approx.1.5 Mbps

Interactive Low 1.5 Mbps, low-lossstream protocol

L Networkconnection

In/Out: Audio 44 kbps Interactive Very low 44 kbps, very low-lossstream protocol

Page 15: © Chinese University, CSE Dept. Distributed Systems / 11 - 1 Distributed Systems Topic 11: Distributed Multimedia Systems Dr. Michael R. Lyu Computer.

© Chinese University, CSE Dept. Distributed Systems / 11 - 15

3 The QoS Manager’s Task

Application components specify their QoS requirements to QoS manager

Yes No

Yes No

Flow spec.

Resource contract

Admission control QoS negotiation

QoS manager evaluates new requirementsagainst the available resources.

Sufficient?

Reserve the requested resources

Allow application to proceed

Application runs with resources as per resource contract

Negotiate reduced resource provision with application.Agreement?

Do not allow application to proceed

Application notifies QoS manager of increased resource requirements

Page 16: © Chinese University, CSE Dept. Distributed Systems / 11 - 1 Distributed Systems Topic 11: Distributed Multimedia Systems Dr. Michael R. Lyu Computer.

© Chinese University, CSE Dept. Distributed Systems / 11 - 16

3.1 Quality of Services Negotiation

Bandwidth: data rate through a component Latency: time needed for a packet to travel

end to end Jitter: the rate of change of latency Loss rate: acceptable drop-frame ratio Quality of service management: negotiation

and allocation of computing resources

Page 17: © Chinese University, CSE Dept. Distributed Systems / 11 - 1 Distributed Systems Topic 11: Distributed Multimedia Systems Dr. Michael R. Lyu Computer.

© Chinese University, CSE Dept. Distributed Systems / 11 - 17

3.1.1 Specifying QoS Parameters

The values of QoS parameters can be stated explicitly or implicitly

Bandwidth: Most video compression techniques produce a stream of frames of different sizes.

Latency: Some timing requirements in multimedia result from the stream itself.

Loss rate: Loss rate is the most difficult QoS parameter to specify.

Page 18: © Chinese University, CSE Dept. Distributed Systems / 11 - 1 Distributed Systems Topic 11: Distributed Multimedia Systems Dr. Michael R. Lyu Computer.

© Chinese University, CSE Dept. Distributed Systems / 11 - 18

3.1.2 Traffic Shaping Algorithms

Traffic shaping: using buffers at source and destination to smooth data flow

Token generator

(a) Leaky bucket (b) Token bucket

Page 19: © Chinese University, CSE Dept. Distributed Systems / 11 - 1 Distributed Systems Topic 11: Distributed Multimedia Systems Dr. Michael R. Lyu Computer.

© Chinese University, CSE Dept. Distributed Systems / 11 - 19

3.1.3 Flow Specification

Flow specification: explicit representation of required resources

Protocol version

Maximum transmission unit

Token bucket rate

Token bucket size

Maximum transmission rate

Minimum delay noticed

Maximum delay variation

Loss sensitivity

Burst loss sensitivity

Loss interval

Quality of guarantee

Bandwidth:

Delay:

Loss:

The RFC 1363 Flow Spec

Page 20: © Chinese University, CSE Dept. Distributed Systems / 11 - 1 Distributed Systems Topic 11: Distributed Multimedia Systems Dr. Michael R. Lyu Computer.

© Chinese University, CSE Dept. Distributed Systems / 11 - 20

3.2 Admission Control

Admission control: allowing or denying client requests based on available resources– Bandwidth reservation

» A common way to ensure QoS level for multimedia stream is to reserve some portion of resource bandwidth for its exclusive use.

– Statistical multiplexing» It is based on the hypothesis that for large number of

streams the required aggregate bandwidth remains nearly constant regardless of the bandwidth of individual streams.

» Multimedia traffic may not obey this hypothesis.

Page 21: © Chinese University, CSE Dept. Distributed Systems / 11 - 1 Distributed Systems Topic 11: Distributed Multimedia Systems Dr. Michael R. Lyu Computer.

© Chinese University, CSE Dept. Distributed Systems / 11 - 21

3.3 Overall Structure

1: Resources provide flow spec to main QoS manager through local QoS managers

2: Main QoS ready to reserve resources

3: Client send request to main QoS

4: Main QoS decides if client can be served based on available resources

5: If so, main QoS tells local QoS to allocate resources (if not, client is rejected)

6: Service begins

7: Main QoS and local QoS monitor resource usage / quality, adjust allocated resources if necessary

8: Return to step 4 if new client connects

9: Service ends, resources are freed

Controller Client

Resource Resource Resource

QoS QoS QoS

Main QoS

Network Transmission LineClient

Page 22: © Chinese University, CSE Dept. Distributed Systems / 11 - 1 Distributed Systems Topic 11: Distributed Multimedia Systems Dr. Michael R. Lyu Computer.

© Chinese University, CSE Dept. Distributed Systems / 11 - 22

3.4 QoS Summary

Serving multimedia requires strict resource control to maintain quality

Resources consist of bandwidth, latency, and loss rate, among others

Resource components declare the resources they need in flow specifications

Quality of service managers negotiate and reserve resources to guarantee quality

Resource + flow spec + QoS manager + transmission lines = distributed multimedia system

Page 23: © Chinese University, CSE Dept. Distributed Systems / 11 - 1 Distributed Systems Topic 11: Distributed Multimedia Systems Dr. Michael R. Lyu Computer.

© Chinese University, CSE Dept. Distributed Systems / 11 - 23

4 Resource Management System

Provide the means to offer QoS to multimedia applications

Addressing issues– QoS Calculation

» To check whether the QoS demands of an application can be satisfied

– Resource Reservation» To reserve an amount of resources according to the given QoS

guarantee

– Resource Scheduling» To enforce that the given QoS guarantees are satisfied by

appropriate scheduling of resource access

Page 24: © Chinese University, CSE Dept. Distributed Systems / 11 - 1 Distributed Systems Topic 11: Distributed Multimedia Systems Dr. Michael R. Lyu Computer.

© Chinese University, CSE Dept. Distributed Systems / 11 - 24

4.1 Resources

Resources– All the entities which participate in the overall task of the

application

– Classification (active vs. passive; exclusive vs. shared; single vs. multiple)

– Scheduled, Assigned for QoS

Resource Capacity– Availability for application when needed

– Be at least as large as the requirements for QoS

– Depending on the mechanisms for QoS calculation, resource reservation and scheduling

Page 25: © Chinese University, CSE Dept. Distributed Systems / 11 - 1 Distributed Systems Topic 11: Distributed Multimedia Systems Dr. Michael R. Lyu Computer.

© Chinese University, CSE Dept. Distributed Systems / 11 - 25

4.2 Reservation Policies

Pessimistic approach– The resource capacities are reserved for the worst case– Advantage: avoid conflicts, offer deterministic guarantees– Disadvantage: high cost, underutilization of resources

Optimistic approach– Resources are reserved on average workload– Advantage : cheaper– Disadvantage : temporal resource conflicts

Resource Reservation Protocol (RSVP)– To exchange and negotiate QoS requirements– Receiver-oriented approach. Receivers are responsible for

initiating and keeping the reservation active.

Page 26: © Chinese University, CSE Dept. Distributed Systems / 11 - 1 Distributed Systems Topic 11: Distributed Multimedia Systems Dr. Michael R. Lyu Computer.

© Chinese University, CSE Dept. Distributed Systems / 11 - 26

4.3 Resource Scheduling

Scheduling of resources to meet QoS

requirements

Fair scheduling: allow all processes some

portion of the resources based on fairness

Real-time scheduling: all to meet real-time

requirements (with deadlines)

– Regular continuous multimedia streams

– Bursty real-time traffic

Page 27: © Chinese University, CSE Dept. Distributed Systems / 11 - 1 Distributed Systems Topic 11: Distributed Multimedia Systems Dr. Michael R. Lyu Computer.

© Chinese University, CSE Dept. Distributed Systems / 11 - 27

4.4 Resource Management Phase

– Phase 1: the set-up or QoS negotiation phase» Applications specify their QoS requirements to be used for

the admission test and the QoS calculation

– Phase 2: the transmission or QoS enforcement phase» Resources are scheduled with respect to the given QoS

guarantees.» Schedulers handle time-critical multimedia streams prior to

time-independent data Resource Monitoring Adaptation

– Phase 3» After the transmission has finished, the allocated resources

must be released.

Page 28: © Chinese University, CSE Dept. Distributed Systems / 11 - 1 Distributed Systems Topic 11: Distributed Multimedia Systems Dr. Michael R. Lyu Computer.

© Chinese University, CSE Dept. Distributed Systems / 11 - 28

4.5 Resource Management System Structure

Contain components used in the enforcement phase

Consist of System Resource Manager and Resource Managers– System Resource Manager

» Control the single ‘Resource Managers’

– Resource Manager» Contain algorithms for admission control and policy control» Keep information about the characteristics of the resource and

its reservations» CPU resource manager, Memory resource manager and so on

Resource management schemes: static vs. dynamic

Page 29: © Chinese University, CSE Dept. Distributed Systems / 11 - 1 Distributed Systems Topic 11: Distributed Multimedia Systems Dr. Michael R. Lyu Computer.

© Chinese University, CSE Dept. Distributed Systems / 11 - 29

4.6 Static Resource Management Scheme

Perform QoS calculation and resource reservation during the setup time

Schedule the resource in such a way that processing deadlines are met

Advantage – Offer strong guarantees for the application’s performance– Provide reliable QoS

Drawback– Difficult to determine the amount of resource needed in advance– Not easily cope with a change in the set of running applications

during the run-time of an application– Rely on total knowledge of the set of available resources

Page 30: © Chinese University, CSE Dept. Distributed Systems / 11 - 1 Distributed Systems Topic 11: Distributed Multimedia Systems Dr. Michael R. Lyu Computer.

© Chinese University, CSE Dept. Distributed Systems / 11 - 30

4.7 Dynamic Resource Management Scheme

Renegotiate the resource as changes of requirements at run-time

The goal– Support of variable-bit rate streams – Adaptation to changes in the set of applications to be served– Allowing for a dynamic change in the relative priority of applications– Serving more applications concurrently – Handling of changes in resource availability

Consisting components– Resource monitor– System resource manager: responsible for the negotiations

Page 31: © Chinese University, CSE Dept. Distributed Systems / 11 - 1 Distributed Systems Topic 11: Distributed Multimedia Systems Dr. Michael R. Lyu Computer.

© Chinese University, CSE Dept. Distributed Systems / 11 - 31

4.7 Dynamic Resource Management Scheme

Applications– Receive adaptation notifications

– Decide how to change their behavior to adapt their resource demands

– Monitor the QoS and start a QoS renegotiation

Drawback– Not able to provide guaranteed, constant QoS

Page 32: © Chinese University, CSE Dept. Distributed Systems / 11 - 1 Distributed Systems Topic 11: Distributed Multimedia Systems Dr. Michael R. Lyu Computer.

© Chinese University, CSE Dept. Distributed Systems / 11 - 32

5 Stream Adaptation: Scaling

Scaling reduces flow rate at source– Temporal scaling: skip frames or audio samples– Spatial scaling: reduce frame size or audio

sample quality– Frequency scaling: modify image compression

algorithm without much loss of quality.– Color-space scaling: reduce the number of entries

in color space.

Page 33: © Chinese University, CSE Dept. Distributed Systems / 11 - 1 Distributed Systems Topic 11: Distributed Multimedia Systems Dr. Michael R. Lyu Computer.

© Chinese University, CSE Dept. Distributed Systems / 11 - 33

5 Stream Adaptation: Filtering

Filtering reduces flow at intermediate points– Filtering provides best possible QoS to each

target by applying scaling at each relevant node on the path from the source to the target.

– RSVP is a QoS negotiation protocol that negotiates the rate at each intermediate node, working from targets to the source.

Page 34: © Chinese University, CSE Dept. Distributed Systems / 11 - 1 Distributed Systems Topic 11: Distributed Multimedia Systems Dr. Michael R. Lyu Computer.

© Chinese University, CSE Dept. Distributed Systems / 11 - 34

5 Stream Adaptation: Filtering

SourceTargets

High bandwidth

Medium bandwidth

Low bandwidth

Page 35: © Chinese University, CSE Dept. Distributed Systems / 11 - 1 Distributed Systems Topic 11: Distributed Multimedia Systems Dr. Michael R. Lyu Computer.

© Chinese University, CSE Dept. Distributed Systems / 11 - 35

5.1 A Sample Wavelet Video Filtering

Basic Concept:

Page 36: © Chinese University, CSE Dept. Distributed Systems / 11 - 1 Distributed Systems Topic 11: Distributed Multimedia Systems Dr. Michael R. Lyu Computer.

© Chinese University, CSE Dept. Distributed Systems / 11 - 36

5.1 Wavelet Video Filtering

Video frames are encoded into packets with “priority label”

QoS Filter drops the least important packets in case of insufficient bandwidth

This maximizes visual quality with resource constraints

Page 37: © Chinese University, CSE Dept. Distributed Systems / 11 - 1 Distributed Systems Topic 11: Distributed Multimedia Systems Dr. Michael R. Lyu Computer.

© Chinese University, CSE Dept. Distributed Systems / 11 - 37

6 Summary

Multimedia applications and systems require new system mechanisms to handle large volumes of time-dependent data in real time (media streams).

The most important mechanism is QoS management, which includes resource negotiation, admission control, resource reservation and resource management.

Negotiation and admission control ensure that resources are not over-allocated, resource management ensures that admitted tasks receive the resources they were allocated.

Read textbook Chapter 20.