9. Multimedia Communication Systems – Reservation Model ...€¦ · 9. Multimedia Communication...

92
9. Multimedia Communication Systems Multimedia Kommunikation SS‘99 1 / 92 Contents 9.1 Overview 9.2 Cooperative Computing Cooperation Awareness Group Communication Architecture 9.3 Application Sharing Approach Architecture 9.4 Audio/Video Conferencing Control Session Management 9.5 Transport Subsystems 9.6 Quality of Service Resource Management QoS Negotiation Translation & Scaling 9.7 Resource Admission Control Reservation Model 9.8 Resource Management Scheduling Traffic Shaping End-to-End Error Control Resource Monitoring and Adaptation 9.9 Architecture

Transcript of 9. Multimedia Communication Systems – Reservation Model ...€¦ · 9. Multimedia Communication...

Page 1: 9. Multimedia Communication Systems – Reservation Model ...€¦ · 9. Multimedia Communication Systems Multimedia Kommunikation SS‘99 11 / 92 9.2 Cooperative Computing / Group

9. M

ultim

edia

Com

mun

icat

ion

Syst

ems

Multimedia Kommunikation SS‘99

1 / 9

2C

onte

nts

9.1

Ove

rvie

w

9.2

Coo

pera

tive

Com

putin

g–

Coo

pera

tion

Aw

aren

ess

–G

roup

Com

mun

icat

ion

Arc

hite

ctur

e

9.3

App

licat

ion

Shar

ing

App

roac

h–

Arc

hite

ctur

e

9.4

Aud

io/V

ideo

Con

fere

ncin

g–

Con

trol

–S

essi

on M

anag

emen

t

9.5

Tra

nspo

rt S

ubsy

stem

s

9.6

Qua

lity

of S

ervi

ce–

Res

ourc

e M

anag

emen

t

–Q

oS N

egot

iatio

n

–T

rans

latio

n &

Sca

ling

9.7

Res

ourc

e A

dmis

sion

Con

trol

–R

eser

vatio

n M

odel

9.8

Res

ourc

e M

anag

emen

t–

Sch

edul

ing

–T

raffi

c S

hapi

ng

–E

nd-t

o-E

nd E

rror

Con

trol

–R

esou

rce

Mon

itorin

g an

d A

dapt

atio

n

9.9

Arc

hite

ctur

e

Page 2: 9. Multimedia Communication Systems – Reservation Model ...€¦ · 9. Multimedia Communication Systems Multimedia Kommunikation SS‘99 11 / 92 9.2 Cooperative Computing / Group

9. M

ultim

edia

Com

mun

icat

ion

Syst

ems

Multimedia Kommunikation SS‘99

2 / 9

29.

1 O

verv

iew

In th

is c

hapt

er w

e di

scus

s is

sues

rel

ated

to m

ultim

edia

com

mun

icat

ion

syst

ems

abov

e th

e da

ta li

nk la

yer.

For

this

, we

subd

ivid

e th

e hi

gher

laye

rs o

f th

e M

ultim

edia

Com

mun

icat

ion

Syst

em in

to tw

o su

bsyt

ems:

appl

icta

ion

subs

yste

m, w

hich

is r

espo

nsib

le f

or m

anag

emen

t

and

serv

ice

issu

es f

or g

roup

coo

pera

tion

and

sess

ion

orch

estr

atio

n (s

uppo

rtin

g a

larg

e sc

ale

of m

ulti

med

ia

appl

icat

ions

)

tran

spor

t sub

syts

ems

(her

e tr

ansp

ort a

nd n

etw

ork

laye

r pr

otoc

ols

for

mul

tim

edia

app

licat

ions

are

pre

sent

ed).

7 A

ppl.

6 Pr

esen

t.5

Sess

ion

4 T

rans

p.

3 N

et

2 L

ink

1 PH

V

App

l.

Tra

nsp.

Page 3: 9. Multimedia Communication Systems – Reservation Model ...€¦ · 9. Multimedia Communication Systems Multimedia Kommunikation SS‘99 11 / 92 9.2 Cooperative Computing / Group

9. M

ultim

edia

Com

mun

icat

ion

Syst

ems

Multimedia Kommunikation SS‘99

3 / 9

29.

1 O

verv

iew

Som

e m

ulti

med

ia a

pplic

atio

ns:

–M

ultim

edia

Mai

l

–V

irtua

l Rea

lity

App

licat

ions

–V

ideo

Con

fere

ncin

g

–C

SC

W (

Com

pute

r S

uppo

rted

Coo

pera

tive

Wor

k)

Com

mun

icat

ion

Req

uire

men

ts:

–H

ighs

peed

net

wor

ks w

ith h

igh

tran

sfer

rat

ese.

g.: F

DD

I, S

witc

hed

Eth

erne

t, F

ast E

ther

net,

Gig

abit

Eth

erne

t, D

QD

B, A

TM

, ...

–H

igh

perf

orm

ance

tran

spor

t pro

toco

lse.

g.: T

CP

, XT

P, A

TM

tran

spor

t pro

toco

ls

App

licat

ion

Sub

syst

ems

Tra

nspo

rt S

ubsy

stem

s

Page 4: 9. Multimedia Communication Systems – Reservation Model ...€¦ · 9. Multimedia Communication Systems Multimedia Kommunikation SS‘99 11 / 92 9.2 Cooperative Computing / Group

9. M

ultim

edia

Com

mun

icat

ion

Syst

ems

Multimedia Kommunikation SS‘99

4 / 9

29.

2 C

oope

rativ

e C

ompu

ting

/ App

licat

ion

Subs

yste

m

Coo

pera

tive

Com

putin

g is

gen

eral

ly k

now

n as

Com

pute

r-Su

ppor

ted-

Coo

pera

tive

-Wor

k (C

SCW

).

Too

ls f

or c

oope

rati

ve c

ompu

ting:

–E

lect

roni

c m

ail

–B

ulle

tin b

oard

s

–S

cree

n sh

arin

g to

ols

–A

pplic

atio

n sh

arin

g

–T

ext-

base

d co

nfer

enci

ng s

yste

ms

–V

ideo

con

fere

nce

syst

ems

(e.g

. MB

one

Too

ls, P

roS

hare

from

Inte

l, P

ictu

reT

el,

Tel

es O

nlin

e, N

etM

eetin

g fr

om M

icro

soft)

Page 5: 9. Multimedia Communication Systems – Reservation Model ...€¦ · 9. Multimedia Communication Systems Multimedia Kommunikation SS‘99 11 / 92 9.2 Cooperative Computing / Group

9. M

ultim

edia

Com

mun

icat

ion

Syst

ems

Multimedia Kommunikation SS‘99

5 / 9

29.

2 C

oope

rativ

e C

ompu

ting

/ Coo

pera

tion

Dim

ensi

on

Com

pute

r-Su

ppor

ted

coop

erat

ion

may

be

cate

gori

zed

acco

rdin

g to

the

foll

owin

g pa

ram

eter

s:T

ime: –

Asy

nchr

onou

s co

oper

ativ

e w

ork

(not

at t

he s

ame

time)

–S

ynch

rono

us c

oope

rativ

e w

ork

(at t

he s

ame

time)

Use

r Sc

ale:

–S

ingl

e us

ers,

two

user

s (“

dial

ogue

, poi

nt to

poi

nt“,

dire

ct c

oope

ratio

n) o

r gr

oups

with

mor

e th

an tw

o us

ers

–S

tatic

or

dyna

mic

gro

ups,

if e

ither

the

mem

bers

are

pre

-det

erm

ined

or

not

Con

trol

:–

Cen

tral

ized

- w

hen

cont

rolle

d by

a “

chai

rman

“–

Dis

trib

uted

- w

hen

cont

rol p

rovi

de c

onsi

sten

t coo

pera

tion

Loc

alit

y:–

Coo

pera

tion

at th

e sa

me

plac

e–

Tel

e-co

oper

atio

n of

use

rs a

t diff

eren

t pla

ces

Page 6: 9. Multimedia Communication Systems – Reservation Model ...€¦ · 9. Multimedia Communication Systems Multimedia Kommunikation SS‘99 11 / 92 9.2 Cooperative Computing / Group

9. M

ultim

edia

Com

mun

icat

ion

Syst

ems

Multimedia Kommunikation SS‘99

6 / 9

29.

2 C

oope

rativ

e C

ompu

ting

/ Coo

pera

tion

Aw

aren

ess

Coo

pera

tion-

tran

spar

ent s

yste

ms

–E

xist

ing

appl

icat

ion

exte

nded

for

coop

erat

ion

(Sin

gle

user

doc

umen

t edi

tors

exp

ande

d fo

r si

mul

tane

ous

editi

ng o

f a s

hare

d

docu

men

t am

ong

seve

ral u

sers

, e.g

. som

e te

xt p

roce

ssor

s)

Coo

pera

tion-

awar

e sy

stem

s

–D

edic

ated

sof

twar

e ap

plic

atio

n fo

r C

SC

W

(e.g

. Lot

us N

otes

, con

fere

ncin

g sy

stem

s)

Page 7: 9. Multimedia Communication Systems – Reservation Model ...€¦ · 9. Multimedia Communication Systems Multimedia Kommunikation SS‘99 11 / 92 9.2 Cooperative Computing / Group

9. M

ultim

edia

Com

mun

icat

ion

Syst

ems

Multimedia Kommunikation SS‘99

7 / 9

29.

2 C

oope

rativ

e C

ompu

ting

/ Gro

up C

omm

unic

atio

n A

rchi

tect

ure

Com

mun

icat

ion

(Tra

nspo

rt)

Supp

ort

Con

fere

ncin

g

Gro

up

Ren

devo

us

App

licat

ion

Shar

ing

Gro

up C

omm

unic

atio

n A

gent

Com

mun

icat

ion

(Tra

nspo

rt)

Supp

ort

Con

fere

ncin

g

Gro

up

Ren

devo

us

App

licat

ion

Shar

ing

Gro

up C

omm

unic

atio

n A

gent

Com

mun

icat

ion

(Tra

nspo

rt)

Supp

ort

Con

fere

ncin

g

Gro

up

Ren

devo

us

App

licat

ion

Shar

ing

Gro

up C

omm

unic

atio

n A

gent

Mul

ticas

tC

omm

unic

atio

nN

etw

ork

Page 8: 9. Multimedia Communication Systems – Reservation Model ...€¦ · 9. Multimedia Communication Systems Multimedia Kommunikation SS‘99 11 / 92 9.2 Cooperative Computing / Group

9. M

ultim

edia

Com

mun

icat

ion

Syst

ems

Multimedia Kommunikation SS‘99

8 / 9

29.

2 C

oope

rativ

e C

ompu

ting

/ Gro

up C

omm

unic

atio

n A

rchi

tect

ure

Gro

up c

omm

unic

atio

n:–

sync

hron

ous

or

–as

ynch

rono

us c

omm

unic

atio

n of

mul

tiple

use

rs

Supp

ort M

odel

:–

With

gro

up c

omm

unic

atio

n ag

ents

(co

oper

atin

g vi

a a

mul

ticas

t net

wor

k)

–G

roup

ren

devo

us (

orga

niza

tion

of m

eetin

gs a

nd d

eliv

erin

g in

form

atio

n)

–Sh

ared

app

licat

ions

(si

mul

tane

ous

repl

icat

ion

and

mod

ifica

tion

of in

form

atio

nto

mul

tiple

use

rs, e

.g. t

elep

oint

ing,

join

t edi

ting)

–C

onfe

renc

ing

Syst

em M

odel

:–

clie

nt-s

erve

r m

odel

Inte

rfac

e M

odel

:–

exch

angi

ng in

form

atio

n w

ithin

the

supp

ort m

odel

(ob

ject

orie

nted

)

Page 9: 9. Multimedia Communication Systems – Reservation Model ...€¦ · 9. Multimedia Communication Systems Multimedia Kommunikation SS‘99 11 / 92 9.2 Cooperative Computing / Group

9. M

ultim

edia

Com

mun

icat

ion

Syst

ems

Multimedia Kommunikation SS‘99

9 / 9

29.

2 C

oope

rativ

e C

ompu

ting

/ Gro

up C

omm

unic

atio

n A

rchi

tect

ure

Clie

nts:

–U

ser

inte

rfac

es s

uppo

rtin

g in

tera

ctio

n be

twee

n gr

oup

mem

ebrs

and

the

syst

em

Serv

ers:

–S

peci

aliz

ed s

erve

rs fo

r di

ffere

nt ta

sks

of th

e gr

oup

com

mun

icat

ion

wor

k, e

.g.

·D

irec

tory

ser

ver

(for

gro

up r

ende

zvou

s se

rvic

es)

·A

pplic

atio

n sh

arin

g se

rver

·C

onfe

renc

e se

rver

·M

ultim

edia

ser

ver

(for

inte

rmed

ia s

ynch

roni

zatio

n)

Page 10: 9. Multimedia Communication Systems – Reservation Model ...€¦ · 9. Multimedia Communication Systems Multimedia Kommunikation SS‘99 11 / 92 9.2 Cooperative Computing / Group

9. M

ultim

edia

Com

mun

icat

ion

Syst

ems

Multimedia Kommunikation SS‘99

10 /

929.

2 C

oope

rativ

e C

ompu

ting

/ Gro

up C

omm

unic

atio

n A

rchi

tect

ure

Use

r Pr

esen

tati

on P

roto

cols

:

Bet

wee

n th

e cl

ient

s, e

.g.

–O

pen,

clo

se, d

ynam

ical

ly jo

in o

r le

ave

a m

eetin

g

–F

loor

pas

sing

Gro

up W

ork

Man

agem

ent P

roto

cols

:

Bet

wee

n th

e cl

ient

s an

d th

e se

rver

s, e

.g.

–R

egis

trat

ion

of m

eetin

gs

–Q

uerie

s fo

r in

form

atio

n ab

out m

eetin

gs

Page 11: 9. Multimedia Communication Systems – Reservation Model ...€¦ · 9. Multimedia Communication Systems Multimedia Kommunikation SS‘99 11 / 92 9.2 Cooperative Computing / Group

9. M

ultim

edia

Com

mun

icat

ion

Syst

ems

Multimedia Kommunikation SS‘99

11 /

929.

2 C

oope

rativ

e C

ompu

ting

/ Gro

up R

ende

zvou

s

Setti

ng u

p of

gro

up a

nd d

eliv

erin

g in

form

atio

n ab

out g

roup

s an

d m

eetin

gs

Syn

chro

nous

Ren

dezv

ous

Met

hods

(e.

g. s

dr -

MB

one´

s se

ssio

n di

rect

ory

tool

):

Exp

lizit

invi

tatio

ns (

poin

t-to

-poi

nt o

r po

int-

to-m

ultip

oint

)–

Initi

ator

of a

mee

ting

has

to k

now

whe

re u

sers

res

ide

Usi

ng d

irec

tory

ser

vice

s (X

.500

), k

now

ledg

e ba

se w

ith in

form

atio

n ab

out t

he

mee

ting

, e.g

.–

Nam

e of

the

mee

ting/

conf

eren

ce

–R

egis

tere

d or

aut

hor

ized

par

ticip

ants

–R

ole

of p

artic

ipan

ts

Asy

nchr

onou

s R

ende

zvou

s M

etho

ds:

–U

sin

g e-

mai

l

–U

sing

bul

letin

boa

rds

(loca

l bul

letin

boa

rds

or W

orld

Wid

e W

eb)

–S

impl

e C

onfe

renc

e A

nnou

ncem

ent P

roto

col (

Dra

ft, 1

996)

Page 12: 9. Multimedia Communication Systems – Reservation Model ...€¦ · 9. Multimedia Communication Systems Multimedia Kommunikation SS‘99 11 / 92 9.2 Cooperative Computing / Group

9. M

ultim

edia

Com

mun

icat

ion

Syst

ems

Multimedia Kommunikation SS‘99

12 /

929.

2 C

oope

rativ

e C

ompu

ting

/ Gro

up R

ende

zvou

s

Use

d in

tele

-con

fere

ncin

g fo

r co

oper

ativ

e do

cum

ent e

ditin

g an

d so

ftw

are

deve

lopm

ent

Shar

ed o

bjec

ts (

docu

men

ts)

disp

laye

d in

sha

red

win

dow

s

Shar

ed a

ppli

cati

on p

rogr

am (

e.g.

text

pro

cess

or)

exec

utes

inpu

t fro

m a

part

icip

ant

Exe

cutio

n re

sult

s on

the

docu

men

t dis

trib

uted

am

ong

all p

artic

ipan

ts

App

licat

ion

shar

ing

arch

itect

ure

–C

entr

aliz

ed

–R

eplic

ated

Con

sist

ency

of

shar

ed o

bjec

ts

Floo

r pa

ssin

g co

ntro

l

Page 13: 9. Multimedia Communication Systems – Reservation Model ...€¦ · 9. Multimedia Communication Systems Multimedia Kommunikation SS‘99 11 / 92 9.2 Cooperative Computing / Group

9. M

ultim

edia

Com

mun

icat

ion

Syst

ems

Multimedia Kommunikation SS‘99

13 /

929.

3 A

pplic

atio

n Sh

arin

g A

ppro

ach

Cen

tral

ized

Sing

le c

opy

of th

e sh

ared

app

licat

ion

runs

at o

ne s

ite

éA

ll in

put f

orw

arde

d to

the

loca

l site

éO

utpu

t (sh

ared

doc

umen

t) d

istr

ibut

ed to

all

site

s

+E

asy

cons

iste

ncy

mai

nten

ance

(on

ly o

ne c

opy

of th

e ap

plic

atio

n)

−H

igh

netw

ork

traf

fic

(for

out

put d

istr

ibut

ion)

−R

elat

ivel

y un

relia

ble

(if

cent

ral f

ails

then

eve

ryth

ing

is d

own)

Rep

licat

ed:

éC

opy

of th

e sh

ared

app

licat

ion

runs

loca

lly a

t eac

h si

te

éIn

put d

istr

ibut

ed to

all

site

s

+L

ow n

etw

ork

traf

fic

and

resp

onse

tim

es (

only

dis

trib

utio

n) a

part

fro

m th

e(e

norm

ous?

) tr

affi

c ar

isin

g fr

om c

onsi

sten

cy m

aint

enan

ce

−D

iffi

cult

cons

iste

ncy

mai

nten

ance

Page 14: 9. Multimedia Communication Systems – Reservation Model ...€¦ · 9. Multimedia Communication Systems Multimedia Kommunikation SS‘99 11 / 92 9.2 Cooperative Computing / Group

9. M

ultim

edia

Com

mun

icat

ion

Syst

ems

Multimedia Kommunikation SS‘99

14 /

929.

3 A

pplic

atio

n Sh

arin

g A

ppro

ach

/ Arc

hite

ctur

e

cent

raliz

ed

Shar

ed A

pplic

atio

n

Net

wor

kSh

ared

Win

dow

Floo

r H

olde

r

Shar

ed W

indo

w

Inpu

tO

utpu

t Ì b

ottle

neck

outp

ut

Shar

ed W

indo

w

Floo

r H

olde

rre

plic

ated

Shar

ed A

pplic

atio

n

Net

wor

kSh

ared

Win

dow

Floo

r H

olde

r

Inpu

tO

utpu

t

Shar

ed A

pplic

atio

n

Inpu

tO

utpu

t

inpu

t cons

iste

ncy

trou

ble

Page 15: 9. Multimedia Communication Systems – Reservation Model ...€¦ · 9. Multimedia Communication Systems Multimedia Kommunikation SS‘99 11 / 92 9.2 Cooperative Computing / Group

9. M

ultim

edia

Com

mun

icat

ion

Syst

ems

Multimedia Kommunikation SS‘99

15 /

929.

3 A

pplic

atio

n Sh

arin

g A

ppro

ach

/ Rep

licat

ed A

rchi

tect

ure

Con

sist

ency

Mai

nten

ance

of

Shar

ed D

ocum

ents

, e.g

.

–C

entr

aliz

ed lo

cks

–D

epen

denc

y de

tect

ions

–F

loor

pas

sing

Floo

r Pa

ssin

g C

ontr

ol:

Floo

r ho

lder

has

rig

ht to

man

ipul

ate

shar

ed d

ocum

ents

in s

hare

d w

indo

ws,

e.g

.

–O

pen

or c

lose

sha

red

win

dow

s

–Lo

ad a

doc

umen

t int

o a

shar

ed w

indo

w

–Is

sue

inpu

t to

shar

ed a

pplic

atio

n

Page 16: 9. Multimedia Communication Systems – Reservation Model ...€¦ · 9. Multimedia Communication Systems Multimedia Kommunikation SS‘99 11 / 92 9.2 Cooperative Computing / Group

9. M

ultim

edia

Com

mun

icat

ion

Syst

ems

Multimedia Kommunikation SS‘99

16 /

929.

3 A

pplic

atio

n Sh

arin

g A

ppro

ach

/ Arc

hite

ctur

e CSC

WC

ontr

ol

Inpu

t Dis

play

Shar

ed D

ata

Shar

edPr

ogra

m

Mer

ged

Inpu

t

CSC

WC

ontr

ol

Inpu

t Dis

play

Shar

ed D

ata

Shar

edPr

ogra

m

Mer

ged

Inpu

t

CSC

WC

ontr

ol

Inpu

t Dis

play

Shar

ed D

ata

Shar

edPr

ogra

m

Mer

ged

Inpu

t

Page 17: 9. Multimedia Communication Systems – Reservation Model ...€¦ · 9. Multimedia Communication Systems Multimedia Kommunikation SS‘99 11 / 92 9.2 Cooperative Computing / Group

9. M

ultim

edia

Com

mun

icat

ion

Syst

ems

Multimedia Kommunikation SS‘99

17 /

929.

3 A

pplic

atio

n Sh

arin

g A

ppro

ach

/ Arc

hite

ctur

e

CSC

W (

com

pute

r su

ppor

ted

coop

erat

ive

wor

k)

cont

rol c

ompo

nent

che

cks

wet

her

the

activ

e si

te is

flo

or h

olde

r

–flo

or h

olde

r ac

cept

s an

d pr

oces

ses

inpu

ts; d

istr

ibut

es in

pute

s to

oth

er s

ites

–no

floo

r ho

lder

s: d

isca

rds

its o

wn

inpu

ts; a

ccep

ts in

put w

hich

com

es fr

om o

ther

site

s

Page 18: 9. Multimedia Communication Systems – Reservation Model ...€¦ · 9. Multimedia Communication Systems Multimedia Kommunikation SS‘99 11 / 92 9.2 Cooperative Computing / Group

9. M

ultim

edia

Com

mun

icat

ion

Syst

ems

Multimedia Kommunikation SS‘99

18 /

929.

4 A

udio

/Vid

eo C

onfe

renc

ing

Man

agem

ent s

ervi

ce w

hich

con

trol

s si

mul

tane

ous

face

-to-

face

com

mun

icta

ion

betw

een

mul

tipl

e us

ers

usin

g m

ultip

le m

edia

(vi

deo,

aud

io e

tc.)

Vid

eo:

–C

onfe

renc

e pa

rtic

ipan

ts (

all,

a fe

w, s

peak

er o

r m

oder

ator

onl

y)

–La

rge

vide

o w

alls

with

mul

tiple

hig

h-re

solu

tion

scre

ens

may

be

used

esp

ecia

lly fo

r

·C

onfe

renc

es w

ith m

ore

than

four

par

ticip

ants

·D

ispl

ay o

f vie

w-g

raph

s, im

ages

, ani

mat

ions

etc

.

Aud

io (

true

ful

l-du

plex

hig

h-qu

ality

com

mun

icta

ion

with

ech

o ca

ncel

latio

n):

–D

iscu

ssio

ns

–Im

port

ant f

or c

larif

ying

vis

ual i

nfor

mat

ion

Page 19: 9. Multimedia Communication Systems – Reservation Model ...€¦ · 9. Multimedia Communication Systems Multimedia Kommunikation SS‘99 11 / 92 9.2 Cooperative Computing / Group

9. M

ultim

edia

Com

mun

icat

ion

Syst

ems

Multimedia Kommunikation SS‘99

19 /

929.

4 A

udio

/Vid

eo C

onfe

renc

ing

Serv

ices

Req

uire

men

ts:

–H

igh

band

wid

th (

for

data

-inte

nsiv

e m

edia

)

–Lo

w la

tenc

y (f

or u

ser

inte

ract

ivity

)

–D

istr

ibut

ed m

essa

ging

of d

ata

and

cont

rol i

nfor

mat

ion

Con

fere

nce

Con

trol

:–

Man

agem

ent o

f con

fere

nces

(E

stab

lishi

ng, C

losi

ng, A

ddin

g/re

mov

ing

user

s)

–P

rovi

ding

info

rmat

ion

abou

t con

fere

nces

(C

onfe

renc

e na

me,

dur

atio

n, e

tc.)

Cen

tral

ized

or

dist

ribu

ted

cont

rol

–C

entr

aliz

ed c

ontr

ol is

eas

y to

impl

emen

t

–D

istr

ibut

ed c

ontr

ol is

muc

h m

ore

com

plic

ated

but

less

sen

sitiv

e to

failu

re(c

entr

al c

ontr

ol d

oes

not w

ork

if ce

nral

is in

oper

ativ

e)

Page 20: 9. Multimedia Communication Systems – Reservation Model ...€¦ · 9. Multimedia Communication Systems Multimedia Kommunikation SS‘99 11 / 92 9.2 Cooperative Computing / Group

9. M

ultim

edia

Com

mun

icat

ion

Syst

ems

Multimedia Kommunikation SS‘99

20 /

929.

4 A

udio

/Vid

eo C

onfe

renc

ing

/ Con

trol

Stat

e In

form

atio

n of

the

Con

fere

nce,

e.g

.:–

Con

fere

nce

nam

e

–S

tart

of t

he c

onfe

renc

e

–P

olic

ies

asso

ciat

ed w

ith th

e co

nfer

ence

–P

artic

ipan

ts

–D

urat

ion

–T

opic

Func

tions

:–

Est

ablis

hing

a c

onfe

renc

e (u

sers

agr

ee u

pon

a co

mm

on c

onfe

renc

e st

ate,

e.g

.en

codi

ng r

ules

, acc

ess

right

s, c

hairm

an o

f the

con

fere

nce)

–A

ddin

g ne

w p

artic

ipan

ts a

nd r

emov

ing

part

icip

ants

dur

ing

a co

nfer

ence

–C

losi

ng a

con

fere

nce

Page 21: 9. Multimedia Communication Systems – Reservation Model ...€¦ · 9. Multimedia Communication Systems Multimedia Kommunikation SS‘99 11 / 92 9.2 Cooperative Computing / Group

9. M

ultim

edia

Com

mun

icat

ion

Syst

ems

Multimedia Kommunikation SS‘99

21 /

929.

4 A

udio

/Vid

eo C

onfe

renc

ing

/ Con

trol

- C

entr

aliz

ed

Cen

tral

Ser

ver:

–S

tore

s th

e co

nfer

ence

sta

te (

i.e. c

onfe

renc

e st

ate

is c

enra

lly o

rgan

ized

)

–M

anag

es c

onfe

renc

e co

ntro

l

–Is

abl

e to

exc

hang

e ex

plic

ity th

e co

nfer

ence

sta

te

Est

ablis

hmen

t of

a C

onfe

renc

e w

ith C

entr

aliz

ed C

ontr

ol:

1. In

itiat

ion

(of t

he c

hairm

an)

of th

e co

nfer

ence

sel

ects

an

initi

al g

roup

of p

artic

ipan

ts

2. H

e in

quire

s ad

dres

ses

of th

e co

nfer

ence

par

ticip

ants

from

a c

enra

l dire

ctor

y se

ver.

3. T

hese

par

ticip

ants

are

exp

licitl

y in

vite

d.

4. E

ach

invi

ted

clie

nt r

espo

nds

to th

e in

vita

tion

Í in

itiat

or is

info

rmed

who

will

par

ticip

ate.

5. N

egot

iatio

n of

con

fere

nce

polic

ies

and

adm

issi

on o

f res

ourc

es a

re p

erfo

rmed

.

T

his

shar

ed c

onfe

rnce

sta

te is

sto

red

on a

cen

tral

ser

ver

and

dist

ribut

ed to

all

p

artic

ipan

ts u

sing

a r

elia

ble

mes

sagi

ng s

ervi

ce.

Page 22: 9. Multimedia Communication Systems – Reservation Model ...€¦ · 9. Multimedia Communication Systems Multimedia Kommunikation SS‘99 11 / 92 9.2 Cooperative Computing / Group

9. M

ultim

edia

Com

mun

icat

ion

Syst

ems

Multimedia Kommunikation SS‘99

22 /

929.

4 A

udio

/Vid

eo C

onfe

renc

ing

/ Con

trol

- D

istr

ibut

ed

Dis

trib

uted

Con

trol

:

–M

anag

es no

glo

bal n

otio

ns o

f gro

up m

embe

rshi

p

–G

ives

no c

onsi

sten

cy g

uara

ntee

s (s

ame

stat

e vi

ew b

y al

l par

ticip

ants

)

–E

xcha

nges

per

iodi

cally

the

conf

eren

ce s

tatu

s be

twee

n th

e pa

rtic

ipan

ts to

min

imiz

e th

e lo

ose

of c

ontr

ol b

y an

unr

elia

ble

mes

sagi

ng s

ervi

ce

Est

ablis

hmen

t of

a C

onfe

renc

e w

ith D

istr

ibut

ed C

ontr

ol:

1. In

itiat

or o

f the

con

fere

nce

esta

blis

hes

a m

ultic

ast s

pace

(m

ultic

ast t

unne

l in

MB

one)

with

mul

ticas

t ent

ries.

2. P

artic

ipan

ts jo

in th

e co

nfer

ence

by

tuni

ng in

to a

par

ticul

ar m

ultic

ast e

ntry

(

mul

ticas

t add

ress

) an

noun

ced

by g

roup

ren

dezv

ous

mea

ns (

sdr

in M

Bon

e)

3. E

ach

part

icip

ant i

nfor

ms

the

othe

rs a

bout

his

ow

n st

atus

.

Page 23: 9. Multimedia Communication Systems – Reservation Model ...€¦ · 9. Multimedia Communication Systems Multimedia Kommunikation SS‘99 11 / 92 9.2 Cooperative Computing / Group

9. M

ultim

edia

Com

mun

icat

ion

Syst

ems

Multimedia Kommunikation SS‘99

23 /

929.

4 A

udio

/Vid

eo C

onfe

renc

ing

/ Ses

sion

Man

agem

ent A

rchi

tect

ure

Cen

tral

ized

Con

fere

nce

Con

trol

:+

Gua

rant

eed

cons

iste

ncy

of th

e co

nfer

ence

sta

te

+E

asy

to im

plem

ent

–A

ddin

g of

a n

ew p

artic

ipan

t cau

ses

larg

e de

lays

(ex

plic

it ex

chan

ge o

f the

con

fere

nce

stat

e am

ong

all p

artic

ipan

ts)

–D

iffic

ult r

eest

ablis

hmen

t of t

he c

onfe

renc

e st

ate

afte

r a

link

failu

re

Dis

trib

uted

Con

fere

nce

Con

trol

:+

Fau

lt to

lera

nce

(eas

y re

esta

blis

hmen

t afte

r ne

twor

k lin

k fa

ilure

s)

+W

ell s

uite

d fo

r sm

all c

onfe

renc

es

–T

he c

onfe

renc

es p

artic

ipan

ts m

ay n

ot h

ave

the

sam

e vi

ew o

f the

sta

te s

pace

–H

ard

to im

plem

ent,

cost

ly, d

iffic

ult t

o un

ders

tand

Rep

licat

ed C

onfe

renc

e C

ontr

ol:

+Lo

w “

requ

est r

espo

nse

time“

–A

ddin

g of

a n

ew p

artic

ipan

ts c

ause

s la

rge

dela

ys (

expl

icit

exch

ange

of t

he c

onfe

renc

est

ate

amon

g al

l par

ticip

ants

)

–H

ighe

r ne

twor

k lo

ad

Page 24: 9. Multimedia Communication Systems – Reservation Model ...€¦ · 9. Multimedia Communication Systems Multimedia Kommunikation SS‘99 11 / 92 9.2 Cooperative Computing / Group

9. M

ultim

edia

Com

mun

icat

ion

Syst

ems

Multimedia Kommunikation SS‘99

24 /

929.

4 A

udio

/Vid

eo C

onfe

renc

ing

/ Ses

sion

Man

agem

ent A

rchi

tect

ure

Sess

ion

Man

agem

ent:

–Im

port

ant p

art o

f the

mul

timed

ia a

rchi

tect

ure

–S

epar

ates

the

cont

rol p

art a

nd th

e da

ta p

art f

rom

the

actu

al tr

ansp

ort

The

ent

ity

whi

ch ta

kes

this

on

is c

alle

d Se

ssio

n M

anag

er:

–In

clu

ding

loca

l and

–R

emot

e fu

ncto

nalit

ies

Med

ia A

gent

s:

–A

re s

epar

ate

from

the

sess

on m

anag

er a

nd

–ha

ve th

e R

espo

nsib

ility

for

deci

sion

s sp

ecifi

c to

eac

h ty

pe o

n m

edia

Shar

ed W

orks

pace

Age

nt:

–T

rans

mits

sha

red

obje

cts

amon

g th

e sh

ared

app

licat

ions

Page 25: 9. Multimedia Communication Systems – Reservation Model ...€¦ · 9. Multimedia Communication Systems Multimedia Kommunikation SS‘99 11 / 92 9.2 Cooperative Computing / Group

9. M

ultim

edia

Com

mun

icat

ion

Syst

ems

Multimedia Kommunikation SS‘99

25 /

929.

4 A

udio

/Vid

eo C

onfe

renc

ing

/ Ses

sion

Man

agem

ent F

unct

iona

litie

s

Loc

al f

unct

iona

litie

s:

–M

embe

rshi

p co

ntro

l man

agem

ent (

e.g.

: par

ticip

ant a

uthe

ntic

atio

n)

–C

ontr

ol m

anag

emen

t (fo

r sh

ared

wor

kspa

ce, s

uch

as fl

oor

cont

rol)

–M

edia

con

trol

man

agem

ent (

e.g.

: com

mun

icat

ion

amon

g m

edia

age

nts)

–C

onfig

urat

ion

man

agem

ent (

such

as

exch

ange

of i

nter

rela

ted

QoS

par

amet

ers

or

sele

ctio

n of

app

ropr

iate

ser

vice

s ac

cord

ing

to Q

oS)

–C

onfe

renc

e co

ntro

l man

agem

ent (

such

as

esta

blis

hmen

t, m

odifi

catio

n an

d

clos

ing

of a

con

fere

nce)

Page 26: 9. Multimedia Communication Systems – Reservation Model ...€¦ · 9. Multimedia Communication Systems Multimedia Kommunikation SS‘99 11 / 92 9.2 Cooperative Computing / Group

9. M

ultim

edia

Com

mun

icat

ion

Syst

ems

Multimedia Kommunikation SS‘99

26 /

929.

4 A

udio

/Vid

eo C

onfe

renc

ing

/ Ses

sion

Man

agem

ent F

unct

iona

litie

s

Man

agem

ent t

asks

:

–co

nfig

urat

ion

man

agem

ent

–fa

ult

man

agem

ent

–pe

rfor

man

ce m

anag

emen

t

–ac

coun

ting

man

agem

ent

–se

curit

y m

anag

emen

t

Rem

ote

func

tiona

litie

s:

–C

omm

unic

atio

n w

ith o

ther

ses

sion

man

ager

s

–E

xcha

nge

of s

tate

info

rmat

ion

whi

ch m

ay in

clud

e flo

or in

form

atio

n,

conf

igur

atio

n in

form

atio

n, e

tc.

Page 27: 9. Multimedia Communication Systems – Reservation Model ...€¦ · 9. Multimedia Communication Systems Multimedia Kommunikation SS‘99 11 / 92 9.2 Cooperative Computing / Group

9. M

ultim

edia

Com

mun

icat

ion

Syst

ems

Multimedia Kommunikation SS‘99

27 /

929.

4 A

udio

/Vid

eo C

onfe

renc

ing

/ Med

ia-

and

Shar

ed W

orks

pace

Age

nt

Med

ia A

gent

s:–

Are

sep

arat

ed fr

om th

e se

ssio

n m

anag

er

–E

ach

type

of m

edia

ow

ns it

s pe

rson

al m

edia

age

nt

–E

ach

Med

ia A

gent

per

form

s its

ow

n co

ntro

l mec

hani

sm o

ver

the

part

icul

arm

ediu

m, s

uch

as·

Mut

e / u

nmut

e

·C

hang

e vi

deo

qual

ity

·S

tart

/ st

op s

endi

ng

Shar

ed W

orks

pace

Age

nt: T

rans

mits

sha

red

obje

cts

amon

g th

e sh

ared

appl

icat

ions

, suc

h as

–T

elep

oint

er c

oord

inat

es

–G

raph

ical

or

text

ual o

bjec

ts

Page 28: 9. Multimedia Communication Systems – Reservation Model ...€¦ · 9. Multimedia Communication Systems Multimedia Kommunikation SS‘99 11 / 92 9.2 Cooperative Computing / Group

9. M

ultim

edia

Com

mun

icat

ion

Syst

ems

Multimedia Kommunikation SS‘99

28 /

929.

4 A

udio

/Vid

eo C

onfe

renc

ing

/ Ses

sion

Con

trol

Arc

hite

ctur

e

Sess

ion

Man

ager

Mem

bers

hip

Con

trol

Con

figu

ratio

nC

ontr

ol

Con

fere

nce

Con

trol

Med

iaC

ontr

ol

Floo

r C

ontr

olSe

ssio

n C

ontr

ol P

roto

col

Whi

tebo

ard

Age

ntV

ideo

A

gent

Aud

io

Age

ntSe

nsor

yD

ata

Age

nt

Med

ia

Age

nt

Shar

edW

orks

pace

Age

nt

Rel

iabl

e T

rans

port

Pro

toco

l

Rea

l-T

ime

Prot

ocol

Rea

l-T

ime

Prot

ocol

Rea

l-T

ime

Prot

ocol

Page 29: 9. Multimedia Communication Systems – Reservation Model ...€¦ · 9. Multimedia Communication Systems Multimedia Kommunikation SS‘99 11 / 92 9.2 Cooperative Computing / Group

9. M

ultim

edia

Com

mun

icat

ion

Syst

ems

Multimedia Kommunikation SS‘99

29 /

929.

4 A

udio

/Vid

eo C

onfe

renc

ing

/ Ses

sion

Con

trol

Arc

hite

ctur

e

Eac

h se

ssio

n is

des

crib

ed th

roug

h its

ses

sion

sta

te, e

.g.:

–st

art t

ime

of s

essi

on

–po

licie

s as

soci

ated

with

a s

essi

on

–se

ssio

n na

me

The

ses

sion

sta

te in

form

atio

n is

eith

er

–P

rivat

e (e

.g.:

loca

l res

ourc

es)

or

–P

ublic

and

sha

red

amon

g al

l ses

sion

s pa

rtic

ipan

ts

Sess

ion

stat

es a

re c

hang

ed d

urin

g es

tabl

ishm

ent o

r m

odif

icat

ion

of a

ses

sion

:

–S

essi

on m

anag

er n

egot

iate

s, a

gree

s an

d se

ts th

e lo

gica

l sta

te o

f its

ow

n se

ssio

n

–It

nego

tiate

s, a

gree

s an

d se

ts b

illin

g po

licy

with

oth

er s

essi

on m

anag

ers

·F

urth

er, i

t per

mits

“pu

blis

hing

“ a

sess

ion,

allo

win

g ot

hers

to jo

in a

ses

sion

Page 30: 9. Multimedia Communication Systems – Reservation Model ...€¦ · 9. Multimedia Communication Systems Multimedia Kommunikation SS‘99 11 / 92 9.2 Cooperative Computing / Group

9. M

ultim

edia

Com

mun

icat

ion

Syst

ems

Multimedia Kommunikation SS‘99

30 /

929.

4 A

udio

/Vid

eo C

onfe

renc

ing

/ Fur

ther

Con

trol

s

Tas

ks o

f th

e Fl

oor

Con

trol

:–

Pro

vide

s ac

cess

to th

e sh

ared

wor

kspa

ce

–M

aint

ains

dat

a co

nsis

tenc

y

Con

trol

s th

e le

vel o

f si

mul

tani

ty a

nd g

ranu

lari

ty f

or a

cces

s co

ntro

l, fi

negr

anul

arit

y–

for

sim

ulta

neou

s us

e

–fo

r nu

mbe

r lo

cks

Exa

mpl

e fo

r a

sim

ple

floo

r co

ntro

l: fl

oor-

pass

ing

mec

hani

sm–

At a

ny ti

me

only

one

par

ticip

ant h

as th

e flo

or (

toke

n pr

inci

ple)

–T

he fl

oor

is h

ande

d of

f to

anot

her

part

icip

ant w

hen

requ

este

d

–T

o ob

tain

the

floor

; th

e pa

rtic

ipan

t mus

t exp

licity

take

act

ion

to s

igna

l a fl

oor

chan

ge

With

rea

l-ti

me

audi

o th

e fl

oor

cont

rol p

rom

otes

turn

-tak

ings

Floo

r co

ntro

l for

rea

l-tim

e au

dio

is o

ften

use

d to

con

trol

ban

dwid

th u

sage

Page 31: 9. Multimedia Communication Systems – Reservation Model ...€¦ · 9. Multimedia Communication Systems Multimedia Kommunikation SS‘99 11 / 92 9.2 Cooperative Computing / Group

9. M

ultim

edia

Com

mun

icat

ion

Syst

ems

Multimedia Kommunikation SS‘99

31 /

929.

4 A

udio

/Vid

eo C

onfe

renc

ing

/ Fur

ther

Con

trol

s

Con

fere

nce

Con

trol

:

–F

or a

com

paris

on b

etw

een

cent

raliz

ed a

nd d

istr

ibut

ed r

eplic

ated

con

fere

ncin

g

cont

rol p

leas

e se

e fu

rthe

r lit

erat

ure.

Med

ia C

ontr

ol:

–In

clu

des

mai

nly

a fu

nctio

nally

for

the

sync

hron

izat

ion

of m

edia

str

eam

s.

Con

figu

ratio

n C

ontr

ol:

–In

clu

des

a co

ntro

l of m

edia

qua

lity,

QoS

han

dlin

g, r

esou

rce

avai

labi

lity

and

othe

r sy

stem

s co

mpo

nent

s to

pro

vide

a s

essi

on a

ccor

ding

to u

sers

req

uire

men

ts.

Mem

bers

hip

Con

trol

:

–In

clu

des

serv

ices

for

invi

tatio

n to

and

reg

istr

atio

n in

to a

ses

sion

as

wel

l as

mod

ifica

tion

of th

e m

embe

rshi

pshi

p du

ring

a se

ssio

n.

Page 32: 9. Multimedia Communication Systems – Reservation Model ...€¦ · 9. Multimedia Communication Systems Multimedia Kommunikation SS‘99 11 / 92 9.2 Cooperative Computing / Group

9. M

ultim

edia

Com

mun

icat

ion

Syst

ems

Multimedia Kommunikation SS‘99

32 /

929.

5 T

rans

port

Sub

syst

ems

Mul

tim

edia

app

lica

tion

s de

man

d hi

gh r

equi

rem

ents

on

netw

ork

prot

ocol

s:

1. H

igh

data

thro

ughp

ut

–de

liver

as

muc

h da

ta a

s po

ssib

le

2. F

ast d

ata

forw

ardi

ng

–de

liver

dat

a as

fast

as

poss

ible

3. S

ervi

ce g

uara

ntee

s

–de

liver

dat

a w

ith th

e ne

gotia

ted

polic

ies

4. M

ultic

asti

ng

–1:

n an

d n:

m p

oint

com

mun

icat

ion

Page 33: 9. Multimedia Communication Systems – Reservation Model ...€¦ · 9. Multimedia Communication Systems Multimedia Kommunikation SS‘99 11 / 92 9.2 Cooperative Computing / Group

9. M

ultim

edia

Com

mun

icat

ion

Syst

ems

Multimedia Kommunikation SS‘99

33 /

929.

5 T

rans

port

Sub

syst

ems

Mos

t tra

nspo

rt p

roto

cols

are

des

igne

d fo

r un

relia

ble

and

“rel

ativ

e sl

ow“

netw

orks

.

Mod

ern

high

spee

d ne

twor

ks, s

uch

as A

TM

, nee

d ne

w tr

ansp

ort p

roto

cols

that

con

sid

er

the

pro

pe

rtie

s o

f th

is n

ew

ne

two

rk te

chn

olo

gy.

Fibe

r

wir

eles

s

Err

or r

ate

10-1

4

Err

or r

ate

10-1

4

Page 34: 9. Multimedia Communication Systems – Reservation Model ...€¦ · 9. Multimedia Communication Systems Multimedia Kommunikation SS‘99 11 / 92 9.2 Cooperative Computing / Group

9. M

ultim

edia

Com

mun

icat

ion

Syst

ems

Multimedia Kommunikation SS‘99

34 /

929.

5 T

rans

port

Sub

syst

ems

1. D

ata

thro

ughp

ut:

–A

udio

and

vid

eo d

ata

rese

mbl

e a

stre

am-li

ke b

ehav

ior

–E

ven

in c

ompr

esse

d m

ode,

they

dem

and

high

dat

a th

roug

hput

(16

kbi

t/s fo

rco

mpr

esse

d A

udio

, 64

kbit/

s fo

r or

igin

al P

CM

-aud

io, 2

Mbi

t/s fo

r M

PE

G-c

oded

vide

o)

–In

a w

orks

tatio

n or

net

wor

k se

vera

l of t

hose

str

eam

s m

ay e

xist

con

curr

ently

–T

elep

hone

ser

vice

s or

vid

eo c

onfe

renc

ing

dem

and

real

-tim

e co

mpu

ting

of th

eda

ta s

trea

ms

Thi

s re

quir

es h

igh

perf

orm

ance

wor

ksta

tions

whi

ch a

re a

ble

to c

ompu

te s

ever

al

mul

tim

edia

str

eam

s si

mul

tane

ousl

y an

d ca

n tr

ansm

it th

e pa

cket

s in

app

ropr

iate

spee

d to

the

netw

ork

inte

rfac

e.

Page 35: 9. Multimedia Communication Systems – Reservation Model ...€¦ · 9. Multimedia Communication Systems Multimedia Kommunikation SS‘99 11 / 92 9.2 Cooperative Computing / Group

9. M

ultim

edia

Com

mun

icat

ion

Syst

ems

Multimedia Kommunikation SS‘99

35 /

929.

5 T

rans

port

Sub

syst

ems

If a

lloca

ted

capa

city

is to

o lo

w a

t a g

iven

tim

e:

1. M

aybe

buf

feri

ng w

ill h

elp

2. G

race

ful d

egra

datio

n of

qua

lity;

e.g

skip

on

mor

e B

-fra

mes

in a

n M

PE

G s

trea

redu

ce n

umbe

r of

col

ours

·re

duce

num

ber

of fr

ames

peak

rat

e

aver

age

rate

sust

aina

ble

rate

time

data

rat

e

I-fr

ame

Page 36: 9. Multimedia Communication Systems – Reservation Model ...€¦ · 9. Multimedia Communication Systems Multimedia Kommunikation SS‘99 11 / 92 9.2 Cooperative Computing / Group

9. M

ultim

edia

Com

mun

icat

ion

Syst

ems

Multimedia Kommunikation SS‘99

36 /

929.

5 T

rans

port

Sub

syst

ems

2. F

ast d

ata

forw

ardi

ng:

–E

nd-t

o-en

d de

lay

is li

miti

ng in

the

end

by th

e sp

eed

of li

ght b

ut a

lso

by th

ein

term

edia

te n

etw

ork

node

s (r

oute

r)

–D

iffer

ent a

pplic

atio

ns r

equi

re d

ata

mov

emen

t ran

ging

from

nor

mal

err

or fr

eeda

ta tr

ansm

issi

on to

tim

e-co

nstr

ain

traf

fic

–R

eal t

ime

com

mun

icat

ion

dem

ands

low

end

-to-

end

dela

ys -

typi

cally

max

. 200

mse

c -

and

low

jitte

r -

varia

tion

of p

acke

t-in

ter-

arriv

al-t

ime.

–F

ast d

ata

deliv

erin

g fo

rces

inte

llige

nt a

ckno

wle

dgm

ent m

echa

nism

s, b

ecau

sese

nd-a

nd-w

ait s

trat

egie

s ar

e no

t sui

tabl

e.

–E

rror

free

dat

a de

liver

ing

and

real

tim

e co

mm

unic

atio

n re

quire

men

ts a

redi

fficu

lt to

gra

nt a

t the

sam

e tim

e, b

ecau

se e

rror

free

dat

a de

liver

ing

forc

es d

ata

retr

ansm

issi

on w

hich

will

lead

the

high

del

ays

and

larg

e jit

ters

!

Í T

his

requ

ires

fast

del

iver

ing

from

the

netw

ork

and

inte

llige

nt p

roto

cols

whi

chtr

ansm

it on

ly n

eces

sary

traf

fic!

Page 37: 9. Multimedia Communication Systems – Reservation Model ...€¦ · 9. Multimedia Communication Systems Multimedia Kommunikation SS‘99 11 / 92 9.2 Cooperative Computing / Group

9. M

ultim

edia

Com

mun

icat

ion

Syst

ems

Multimedia Kommunikation SS‘99

37 /

929.

5 T

rans

port

Sub

syst

ems

3. S

ervi

ce g

uara

ntee

s:–

Ser

vice

gua

rant

ees

are

impo

rtan

t for

the

acce

ptan

ce o

f dis

trib

uted

mul

timed

iaap

plic

atio

ns

–D

istr

ibut

ed m

ultim

edia

app

licat

ions

nee

d gu

aran

tees

, suc

h as

thro

ughp

ut ≥

min

valu

e, d

elay

≤ m

axva

lue

and

jitte

r ≤ m

axjit

ter

–T

o ac

hiev

e se

rvic

e gu

aran

tees

, res

ourc

e m

anag

emen

t mus

t be

used

- w

ithou

tth

is in

end

-sys

tem

s an

d sw

itche

s/ro

uter

s, m

ultim

edia

sys

tem

s ca

nnot

pro

vide

relia

ble

QoS

to th

eir

user

s be

caus

e tr

ansm

issi

on o

ver

unre

serv

ed r

esou

rces

may

lead

to d

ropp

ed o

r de

laye

d pa

cket

s

4. M

ultic

asti

ng:

–M

ultic

astin

g is

impo

rtan

t for

mul

timed

ia-d

istr

ibut

ed a

ppic

atio

ns in

term

s of

shar

ing

reso

urce

s lik

e th

e ne

twor

k ba

ndw

idth

–M

any

mul

timed

ia a

pplic

atio

ns, s

uch

as v

ideo

con

fere

ncin

g, h

ave

mul

ticas

ting

char

acte

ristic

s

Page 38: 9. Multimedia Communication Systems – Reservation Model ...€¦ · 9. Multimedia Communication Systems Multimedia Kommunikation SS‘99 11 / 92 9.2 Cooperative Computing / Group

9. M

ultim

edia

Com

mun

icat

ion

Syst

ems

Multimedia Kommunikation SS‘99

38 /

929.

5 T

rans

port

Sub

syst

ems

Mul

timed

ia a

pplic

atio

ns g

ener

ate

and

cons

ume

huge

am

ount

of

data

and

lay

high

cla

ims

to th

e un

derl

ying

infr

astr

uctu

re a

nd c

omm

unic

atio

n ar

chite

ctur

e:

–D

ata

shou

ld b

e co

pied

dire

ctly

from

ada

pter

to a

dapt

er to

red

uce

copy

ove

rhea

d e.

g.:

from

the

vide

o bo

ard

to th

e ne

twor

k in

terf

ace

by u

sing

dire

ct m

emor

y ac

cess

(D

MA

).W

ith D

MA

tech

niqu

e, th

e ap

plic

atio

n its

elf n

ever

rea

lly to

uche

s th

e da

ta, i

t onl

y se

ts th

eco

rrec

t sw

itche

s fo

r th

e da

ta fl

ow b

y co

nnec

ting

sour

ces

to s

inks

.

–T

he d

ata

mov

emen

t inv

olve

d by

the

laye

red

stru

ctur

e of

the

prot

ocol

s fo

rm a

bot

tlene

ck,

henc

e ot

her

mec

hani

sms

mus

t be

foun

d.

–F

or e

rror

-rec

over

y so

me

prot

ocol

s us

e re

tran

smis

sion

tech

niqu

es w

hich

impo

sere

quire

men

ts o

n bu

ffer

spac

e fo

r qu

eues

at t

he e

xpan

se o

f lar

ger

end-

to-e

nd d

elay

s

–T

he s

ynch

rono

us b

ehav

ior

of m

ost m

ultim

edia

traf

fic s

trea

ms

mus

t be

map

ped

on th

eas

ynch

rono

us tr

ansf

er m

ode

of th

e un

derly

ing

netw

orks

Are

the

“old

“ tr

ansp

ort p

roto

cols

like

TC

P a

nd U

DP

sui

tabl

e fo

r m

ultim

edia

tran

smis

sion

and

whi

ch n

ew p

roto

cols

do

exis

t?

Page 39: 9. Multimedia Communication Systems – Reservation Model ...€¦ · 9. Multimedia Communication Systems Multimedia Kommunikation SS‘99 11 / 92 9.2 Cooperative Computing / Group

9. M

ultim

edia

Com

mun

icat

ion

Syst

ems

Multimedia Kommunikation SS‘99

39 /

929.

6 Q

ualit

y of

Ser

vice

/ R

esou

rce

Man

agem

ent

Res

ourc

eM

anag

emen

t

Net

wor

k L

evel

Switc

h

Resource Management

App

licat

ion

Lev

el

End

syst

em

Syst

emL

evel

Net

wor

kL

evel

MM

Dev

ice

Lev

el

Resource Management

App

licat

ion

Lev

el

End

syst

em

Syst

emL

evel

Net

wor

kL

evel

MM

Dev

ice

Lev

el

Net

wor

kN

etw

ork

Syst

em L

evel

Res

ourc

eM

anag

emen

t Pro

toco

l

App

licat

ion

Lev

el R

esou

rce

Man

agem

ent P

roto

col

Page 40: 9. Multimedia Communication Systems – Reservation Model ...€¦ · 9. Multimedia Communication Systems Multimedia Kommunikation SS‘99 11 / 92 9.2 Cooperative Computing / Group

9. M

ultim

edia

Com

mun

icat

ion

Syst

ems

Multimedia Kommunikation SS‘99

40 /

929.

6 Q

ualit

y of

Ser

vice

/ R

esou

rce

Man

agem

ent

Mai

n go

al o

f R

esou

rce

Man

agem

ent:

–pr

ovid

ing

guar

ante

ed d

eliv

ery

of m

ultim

edia

dat

a

Í th

ree

mai

n ac

tions

:

–re

serv

e an

d al

loca

te r

esou

rces

(en

d-to

-end

) du

ring

esta

blis

hmen

t

–pr

ovid

e re

sour

ce a

ccor

ding

to Q

oS s

peci

ficat

ion

–ad

apt t

o re

sour

ce c

hang

es d

urin

g da

ta p

roce

ssin

g

Rel

atio

ns b

etw

een

QoS

and

res

ourc

es:

–Q

oS p

aram

ente

rs

⇔ r

esou

rce

quan

tity

allo

cate

d to

the

serv

ice

and

reso

urce

sche

dulin

g Í R

esou

rce

Man

agem

ent h

andl

es d

iffer

ent m

appi

ngs

bet

wee

n Q

oS

para

met

ers

and

thei

r co

rres

pond

ing

reso

urce

s

Page 41: 9. Multimedia Communication Systems – Reservation Model ...€¦ · 9. Multimedia Communication Systems Multimedia Kommunikation SS‘99 11 / 92 9.2 Cooperative Computing / Group

9. M

ultim

edia

Com

mun

icat

ion

Syst

ems

Multimedia Kommunikation SS‘99

41 /

929.

6 Q

ualit

y of

Ser

vice

/ Q

oS N

egot

iatio

n

Bila

tera

l Pee

r-to

-Pee

r-N

egot

iatio

n:–

take

s pl

ace

betw

een

two

serv

ice

user

s

–se

rvic

e pr

ovid

er is

not

allo

wed

to m

odify

prop

osed

val

ues

–C

alle

e m

ay m

odify

pro

pose

d Q

oS v

alu

es

Bila

tera

l Lay

er-t

o-L

ayer

-Neg

otia

tion:

–ta

kes

plac

e be

twee

n se

rvic

e us

er a

nd s

ervi

cepr

ovid

er

–C

alle

e m

ay m

odify

pro

pose

d Q

oS v

alu

es

Uni

late

ral N

egot

iaito

n:–

Ser

vice

use

r or

ser

vice

pro

vide

r is

not

allo

wed

to m

odify

pro

pose

d va

lues

–ta

ke it

or

leav

e it

Cal

ler

Cal

lee

Ser

vice

Pro

vide

r

Pee

r-to

-Pee

rN

egot

iatio

n

t 1t 2

t 4t 3

Con

nect

C

onfi

rm

Con

nect

R

eque

stC

onne

ct

Indi

cati

onC

onne

ct

Res

pons

e

Req

uest

ed

QoS

Val

ue

Cha

nged

QoS

Val

ue

Page 42: 9. Multimedia Communication Systems – Reservation Model ...€¦ · 9. Multimedia Communication Systems Multimedia Kommunikation SS‘99 11 / 92 9.2 Cooperative Computing / Group

9. M

ultim

edia

Com

mun

icat

ion

Syst

ems

Multimedia Kommunikation SS‘99

42 /

929.

6 Q

ualit

y of

Ser

vice

/ Q

oS N

egot

iatio

ns

Tri

angu

lar

Neg

otia

tion

for

a B

ound

ed T

arge

t:

QoS

rep

rese

nted

thro

ugh

two

boun

ds:

–ta

rget

(av

erag

e va

lue)

–lo

wes

t qua

lity

avai

labl

e (m

inim

al v

alue

)

Goa

l: ne

gotia

te ta

rget

val

ue

Tri

angu

lar

Neg

otia

tion

for

Con

trac

tual

Val

ue:

QoS

rep

rese

nted

thro

ugh

two

boun

ds:

–m

inim

al r

eque

sted

val

ue

–bo

und

of s

tren

gthe

ning

Goa

l: ag

ree

on a

con

trac

tual

val

ue “T

riang

ular

“:

C

allin

g us

er

Cal

led

Use

r

P

rovi

der

Cal

ler

Cal

lee

Ser

vice

Pro

vide

r

Pee

r-to

-Pee

rN

egot

iatio

n

t 1t 2

t 4t 3

Con

nect

C

onfi

rm

Con

nect

R

eque

stC

onne

ct

Indi

cati

onC

onne

ct

Res

pons

e

Ava

ilabl

e Q

oS V

alue

Sele

cted

QoS

Val

ueSe

lect

edQ

oS V

alue

Tar

get

QoS

Val

ue

Low

est Q

uali

tyA

ccep

tabl

e

Page 43: 9. Multimedia Communication Systems – Reservation Model ...€¦ · 9. Multimedia Communication Systems Multimedia Kommunikation SS‘99 11 / 92 9.2 Cooperative Computing / Group

9. M

ultim

edia

Com

mun

icat

ion

Syst

ems

Multimedia Kommunikation SS‘99

43 /

929.

6 Q

ualit

y of

Ser

vice

/ Q

oS N

egot

iatio

ns

QoS

neg

otia

tion

are

pro

vide

d by

dif

fere

nt R

eser

vatio

n Pr

otoc

ols,

e.g

.

ST-I

I: (

Stre

am P

roto

col I

I)

QoS

par

amet

ers

rela

ted

to th

roug

hput

are

neg

otia

ted

with

tria

ngul

ar n

egot

iatio

nfo

r a

boun

ded

targ

et

QoS

par

amet

ers

rela

ted

to d

elay

are

not

neg

otia

ted

–ca

lling

use

r sp

ecifi

cs m

axim

um tr

ansi

t del

ay

–ea

ch S

T-a

gent

est

imat

e th

e av

erag

e tr

ansi

t del

ay

–pr

ovid

er p

rese

nts

tota

l est

imat

ed tr

ansi

t dea

ly in

the

indi

catio

n

–ca

lled

user

dec

ides

if th

e (e

xpec

ted)

del

ay is

suf

ficie

nt

QoS

par

amet

ers

rela

ted

to e

rror

con

trol

are

not

neg

otia

ted

RSV

P (R

eser

vatio

n P

roto

col)

, RC

AP:

QoS

par

amet

ers

are

nego

tiate

d w

ith tr

iang

ular

neg

otia

tion

Page 44: 9. Multimedia Communication Systems – Reservation Model ...€¦ · 9. Multimedia Communication Systems Multimedia Kommunikation SS‘99 11 / 92 9.2 Cooperative Computing / Group

9. M

ultim

edia

Com

mun

icat

ion

Syst

ems

Multimedia Kommunikation SS‘99

44 /

929.

6 Q

ualit

y of

Ser

vice

/ T

rans

latio

n

Dif

fere

nt c

ompo

nent

s re

quir

e di

ffer

ent Q

oS p

aram

eter

s

Í tr

ansl

atio

n du

ring

con

nect

ion

esta

blis

hmen

t bet

wee

n la

yers

:

Hum

an I

nter

face

- A

pplic

atio

n Q

oS

–tu

nin

g se

rvic

e: p

rovi

des

Gra

phic

al U

ser

Inte

rfac

e (G

UI)

for

inpu

t/out

put

of

appl

icat

ion

QoS

App

licat

ion

QoS

- S

yste

m Q

oS

–e.

g. tr

ansl

ates

vid

eo fr

ame

size

to tr

ansp

ort p

acke

t siz

e

–m

ay b

e co

nnec

ted

with

seg

men

tatio

n/re

asse

mbl

y fu

nctio

ns

Syst

em Q

oS -

Net

wor

k Q

oS

–m

aps

syst

em Q

oS (

e.g.

tran

spor

t pac

ket e

nd-t

o-en

d de

lay)

to th

e ne

twor

k Q

oS

para

met

ers

(e.g

. end

-to-

end

dela

y of

cel

ls in

AT

M)

Page 45: 9. Multimedia Communication Systems – Reservation Model ...€¦ · 9. Multimedia Communication Systems Multimedia Kommunikation SS‘99 11 / 92 9.2 Cooperative Computing / Group

9. M

ultim

edia

Com

mun

icat

ion

Syst

ems

Multimedia Kommunikation SS‘99

45 /

929.

6 Q

ualit

y of

Ser

vice

/ T

rans

latio

n &

Sca

ling

Tra

nsla

tion

is b

idir

ectio

nal Í

cau

ses

prob

lem

s:

Exa

mpl

e: v

ideo

rat

e an

d vi

deo

fram

e si

ze d

eter

min

e th

e th

roug

hput

for

the

netw

ork

- if

the

thro

ughp

ut is

rel

axed

, it m

ay tr

ansl

ate

into

eith

er lo

wer

ing

the

qual

ity o

f th

e im

age

(fra

me

size

) or

the

fram

e ra

te

Scal

ing:

sub

sam

ple

a da

ta s

trea

m to

pre

sent

a f

ract

ion

of it

s or

igin

al c

onte

nts

–tr

ansp

aren

t sca

ling:

can

be

aplie

d in

depe

nden

tly fr

om u

pper

pro

toco

l lay

ers,

e.g

.

drop

pin

g so

me

iden

tifia

ble

port

ions

of d

ata

stre

am

–no

n-tr

ansp

aren

t sca

ling:

req

uire

s in

tera

ctio

ns o

f the

tran

spor

t sys

tem

with

upp

er

laye

rs, e

.g. m

odifi

catio

n of

som

e co

ding

alg

orith

m p

aram

eter

s or

rec

ordi

ng in

a

diffe

rent

form

at

Page 46: 9. Multimedia Communication Systems – Reservation Model ...€¦ · 9. Multimedia Communication Systems Multimedia Kommunikation SS‘99 11 / 92 9.2 Cooperative Computing / Group

9. M

ultim

edia

Com

mun

icat

ion

Syst

ems

Multimedia Kommunikation SS‘99

46 /

929.

6 Q

ualit

y of

Ser

vice

/ Sc

alin

g

Scal

ing

can

be a

ppli

ed to

aud

io a

nd v

ideo

dat

a:

tran

spar

ent s

cali

ng is

dif

ficu

lt fo

r au

dio,

bec

ause

pre

sent

ing

a fr

actio

n is

eas

ilyno

tice

d by

the

hum

an li

sten

er Í

non

-tra

nspa

rent

sca

ling

, e.g

. cha

ngin

g th

esa

mpl

e ra

te

For

vide

o, s

calin

g m

etho

d de

pend

s on

the

unde

rlyi

ng c

ompr

essi

on m

etho

ds:

–T

empo

ral s

calin

g re

duce

s nu

mbe

r of

tran

smitt

ed v

ideo

fram

es, b

est s

uite

d fo

rvi

deo

with

indi

vid

ual s

elf-

cont

aine

d fr

ames

–S

patia

l sca

ling

redu

ces

num

ber

of p

ixel

s of

eac

h im

age

–F

requ

ency

sca

ling

redu

ces

te n

umbe

r of

DC

T c

oeffi

cien

ts in

cod

ing

algo

rithm

–A

mpl

itude

sca

ling

redu

ces

the

colo

r de

pth

by in

trod

ucin

g a

coar

ser

quan

tizat

ion

of th

e D

CT

coe

ffici

ents

–C

olor

spa

ce s

calin

g, e

.g. s

witc

hing

from

col

or to

gre

y-sc

ale

pres

enta

tion

Page 47: 9. Multimedia Communication Systems – Reservation Model ...€¦ · 9. Multimedia Communication Systems Multimedia Kommunikation SS‘99 11 / 92 9.2 Cooperative Computing / Group

9. M

ultim

edia

Com

mun

icat

ion

Syst

ems

Multimedia Kommunikation SS‘99

47 /

929.

7 R

esou

rce

Adm

issi

on C

ontr

ol /

Res

erva

tion

Mod

el

Adm

issi

on C

ontr

ol c

heck

s re

sour

ces

alon

g th

e ne

twor

k pa

th f

or a

vaila

bilit

y

Thr

ee r

eser

vatio

n m

odel

s:

–S

ingl

e S

erve

r/S

ingl

e R

ecei

ver

(e.g

. RC

AP

)

–S

ingl

e S

erve

r/M

ultip

le R

ecei

ver

(e.g

. ST

-II)

–M

ultip

le S

erve

r/M

ultip

le R

ecei

ver

(e.g

. RS

VP

)

Tw

o re

serv

atio

n di

rect

ions

:

–se

nder

-orie

nted

(e.

g. S

T-I

I)

–re

ceiv

er-o

rient

ed (

e.g.

RS

VP

)

Res

erva

tion

s ar

e m

ade:

–im

med

iate

–in

adv

ance

Page 48: 9. Multimedia Communication Systems – Reservation Model ...€¦ · 9. Multimedia Communication Systems Multimedia Kommunikation SS‘99 11 / 92 9.2 Cooperative Computing / Group

9. M

ultim

edia

Com

mun

icat

ion

Syst

ems

Multimedia Kommunikation SS‘99

48 /

929.

7 R

esou

rce

Adm

issi

on C

ontr

ol /

Exa

mpl

e: R

CA

P in

Ten

et S

uite

RC

AP

is a

Sin

gle

Serv

er/s

ingl

e R

ecei

ver

send

er-o

rien

ted

rese

rvat

ion

prot

ocol

with

imm

edia

te a

lloc

atio

n in

one

rou

nd tr

ip

Send

er

AD

M_R

EQ

Serv

erT

ests

Nod

e 1

Nod

e n

Serv

erT

ests

Rec

eive

r

...A

DM

_RE

Q

AD

M_R

EQ

Des

tina

tion

Tes

tsA

djus

tR

esou

rces

...

AD

M_C

ON

F

AD

M_C

ON

F

Adj

ust

Res

ourc

es

AD

M_C

ON

F

Dat

a

Page 49: 9. Multimedia Communication Systems – Reservation Model ...€¦ · 9. Multimedia Communication Systems Multimedia Kommunikation SS‘99 11 / 92 9.2 Cooperative Computing / Group

9. M

ultim

edia

Com

mun

icat

ion

Syst

ems

Multimedia Kommunikation SS‘99

49 /

929.

7 R

esou

rce

Adm

issi

on C

ontr

ol /

Exa

mpl

e: R

CA

P in

Ten

et S

uite

Serv

er T

ests

all

ocat

e re

sour

ces

by–

pess

imis

tic a

ppro

ach:

wor

st c

ase

allo

catio

n, e

.g. a

loca

ting

band

wid

th a

ccor

ding

to p

eak

rate

–op

timis

tic a

ppro

ach:

allo

cate

s re

sour

ce to

an

aver

age

wor

kloa

d, e

.g. a

lloca

ting

band

wid

th a

ccor

din

g to

ave

rage

rat

e ov

er a

spe

cific

tim

e

Des

tinat

ion

Tes

ts in

rec

eive

r co

mpu

tes

loca

l con

stra

ints

of

the

inte

rmed

iate

node

s –A

djus

t res

ourc

es a

t way

bac

k to

the

send

er

Serv

er T

ests

:–

Det

erm

inis

tic T

est:

dete

rmin

istic

ban

dwid

th te

st fo

r gu

aran

teed

ser

vice

s–

Sta

tistic

al T

est:

stat

istic

al b

andw

idth

test

for

guar

ante

ed s

ervi

ces

–D

elay

Bou

nd C

ompu

tatio

n: c

ompu

tes

min

imal

node

dea

ly, d

epen

ds o

nsc

hedu

ling

stra

tegy

–B

uffe

r T

est:

allo

cate

s bu

ffer

and

com

pute

s m

axim

um n

ode

dela

y ac

cord

ing

tore

serv

ed b

uffe

r

Page 50: 9. Multimedia Communication Systems – Reservation Model ...€¦ · 9. Multimedia Communication Systems Multimedia Kommunikation SS‘99 11 / 92 9.2 Cooperative Computing / Group

9. M

ultim

edia

Com

mun

icat

ion

Syst

ems

Multimedia Kommunikation SS‘99

50 /

929.

7 R

esou

rce

Adm

issi

on C

ontr

ol /

Exa

mpl

e: R

CA

P in

Ten

et S

uite

Des

tinat

ion

Tes

t:

–D

-Tes

t: te

sts

if m

inim

al d

elay

is le

ss th

an m

axim

al s

peci

fied

del

ay

–W

-Com

puta

tion:

com

pute

s al

l loc

al p

roba

bilit

ies

of p

acke

t los

s du

e to

buf

fer

over

flow

–Z

-Tes

t: co

mpu

tes

alllo

cal p

roba

bilit

ies

of p

acke

t los

s du

e to

del

ay v

iola

tion

–D

elay

Com

puta

tion:

com

pute

s lo

cal n

ode

dela

ys

–J-

Tes

t: ne

cess

ary

for

rate

con

trol

lled

sche

dule

rs -

com

pute

s de

adlin

e to

guar

ante

e jit

ter

Ban

dwid

th te

sts

can

be m

ade

pess

imis

tic (

peak

rat

e al

loca

tion)

or

optim

istic

(max

imum

ave

rage

allo

catio

n)

Seve

ral i

mpr

ovem

ents

of

the

Ten

et a

ppro

ach

exis

t (in

clud

ing

pred

ictiv

e

serv

ice)

Page 51: 9. Multimedia Communication Systems – Reservation Model ...€¦ · 9. Multimedia Communication Systems Multimedia Kommunikation SS‘99 11 / 92 9.2 Cooperative Computing / Group

9. M

ultim

edia

Com

mun

icat

ion

Syst

ems

Multimedia Kommunikation SS‘99

51 /

929.

7 R

esou

rce

Adm

issi

on C

ontr

ol

Adm

issi

on c

ontr

ol: C

an th

e us

er r

equi

rem

ent b

e ac

cept

ed?

Tra

ffic

sha

ping

: How

to a

void

that

the

netw

ork

is o

verl

oade

d?

Sche

dulin

g st

rate

gies

in th

e ro

uter

s:

How

to h

ande

l dif

fere

nt c

onne

ctio

ns s

uch

that

dea

dlin

es a

re s

atis

fied

.

Use

r 1

Use

r 2

Rou

ter

Net

wor

k

Use

r 3

Use

r 4

Req

uire

men

ts(B

andw

idth

, Del

ay, .

..)Ji

tter

Page 52: 9. Multimedia Communication Systems – Reservation Model ...€¦ · 9. Multimedia Communication Systems Multimedia Kommunikation SS‘99 11 / 92 9.2 Cooperative Computing / Group

9. M

ultim

edia

Com

mun

icat

ion

Syst

ems

Multimedia Kommunikation SS‘99

52 /

929.

8 R

esou

rce

Man

agem

ent /

Sch

edul

ing

Exi

stin

g sc

hedu

ling

stra

tegi

es (

e.g.

FIF

O)

are

not s

uita

ble

for

QoS

bas

ed s

ched

ulin

gbe

caus

e st

rict

per

form

ance

gua

rant

ees

are

not p

rovi

ded

The

re a

re d

iffe

rent

sch

edul

ers

prov

idin

g th

ese

guar

ante

es:

–W

eigh

ted

Fai

r Q

ueue

ing

–V

irtua

l Clo

ck

–D

elay

Ear

liest

Due

Dat

e (D

-ED

D)

–Ji

tter

Ear

liest

Due

Dat

e (J

-ED

D)

–S

top-

and-

Go

–R

ate

Con

trol

led

Sta

tic P

riorit

y

Dif

fere

nces

to s

impl

e sc

hedu

ling

stra

tegi

es:

–m

ore

com

plex

que

uein

g

–of

ten

time-

stam

ped

–m

ore

com

plex

com

puta

tion

e.g.

of d

eadl

ines

–no

n-w

orkc

onse

rvin

g (w

orkc

onse

rvin

g =

ser

ver

serv

es w

hene

ver

ther

e is

som

ethi

ng to

do)

Que

ues

Buf

fers

Rou

ter

outp

uts

inpu

ts

how

to d

o a

sche

dulin

g?

Page 53: 9. Multimedia Communication Systems – Reservation Model ...€¦ · 9. Multimedia Communication Systems Multimedia Kommunikation SS‘99 11 / 92 9.2 Cooperative Computing / Group

9. M

ultim

edia

Com

mun

icat

ion

Syst

ems

Multimedia Kommunikation SS‘99

53 /

929.

8 R

esou

rce

Man

agem

ent /

Tra

ffic

Sha

ping

Why

traf

fic

Shap

ing?

–de

scrip

tion

of th

e flo

w´s

cha

ract

eris

tics

to th

e ne

twor

k pr

ovid

er (

adm

issi

onco

ntro

l)–

netw

ork

prov

ider

can

det

erm

ine

if th

e flo

w is

allo

wed

to s

end

(pol

icin

g)–

netw

ork

prov

ider

can

mon

itor

the

traf

fic o

f the

flow

(m

onito

ring)

Req

uire

men

ts to

sha

ping

sch

emes

:–

abili

ty to

des

crib

e a

wid

e ra

nge

of b

ehav

iors

–m

akin

g it

easy

to d

escr

ibe

traf

fic p

atte

rns

from

the

shap

ing

rule

s–

shap

ing

rule

s sh

ould

be

easy

to p

olic

e

Tra

ffic

Sha

ping

Sch

emes

may

be

char

acte

rize

d by

e.g

.–

allo

wed

pea

k ra

te,

–m

axim

um b

urst

siz

e or

–av

erag

e ra

te o

ver

a de

term

ined

inte

rval

The

se p

aram

eter

s m

ay b

e co

mbi

ned

Net

wor

kTra

ffic

shap

ing

Page 54: 9. Multimedia Communication Systems – Reservation Model ...€¦ · 9. Multimedia Communication Systems Multimedia Kommunikation SS‘99 11 / 92 9.2 Cooperative Computing / Group

9. M

ultim

edia

Com

mun

icat

ion

Syst

ems

Multimedia Kommunikation SS‘99

54 /

929.

8 R

esou

rce

Man

agem

ent /

Tra

ffic

Sha

ping

- L

eaky

Buc

ket

data

is tr

ansm

itted

uni

form

ly w

ith r

ate

ρ Í

tran

sfer

is s

moo

thed

burs

ts b

igge

r th

an β

are

dis

card

edÍ

boun

ded

dela

y

easy

to im

plem

ent (

FIFO

+ ti

mer

)

Í L

eaky

Buc

ket g

ener

ates

a m

ore

stea

dyst

ream

inst

ead

of b

urst

y tr

affi

c

ρ

Net

wor

k

stea

dy f

lowB

ucke

t with

fini

te c

apac

ity

Dat

a (p

ossi

bly

burs

ty)

Tok

en β

Page 55: 9. Multimedia Communication Systems – Reservation Model ...€¦ · 9. Multimedia Communication Systems Multimedia Kommunikation SS‘99 11 / 92 9.2 Cooperative Computing / Group

9. M

ultim

edia

Com

mun

icat

ion

Syst

ems

Multimedia Kommunikation SS‘99

55 /

929.

8 R

esou

rce

Man

agem

ent /

Tra

ffic

Sha

ping

- T

oken

Buc

ket

data

may

be

sent

if to

kens

are

ava

ilabl

e in

the

buck

et

thre

e po

ssib

le s

cena

rios

: con

side

r tr

ying

tose

nd a

pac

ket o

f si

ze b

toke

ns (

b <

β):

–to

ken

buck

et fu

ll Í p

acke

t is

sent

and

bto

kens

are

rem

oved

–to

ken

buc

ket e

mpt

y Í w

ait f

or b

toke

ns to

drop

into

the

buck

et

–to

ken

buck

et p

artia

lly fu

ll Í if

ther

e ar

eB

≥ b

in th

e bu

cket

, pac

ket i

s se

ntim

med

iatly

, if B

< b

pac

ket h

as to

wai

t for

the

mis

sing

(b-

B)

toke

ns

ÍT

oken

Buc

ket p

erm

its b

urst

ines

s bu

t

bou

nds

it

ρ

Buc

ket

Dat

a

Tok

en β

Page 56: 9. Multimedia Communication Systems – Reservation Model ...€¦ · 9. Multimedia Communication Systems Multimedia Kommunikation SS‘99 11 / 92 9.2 Cooperative Computing / Group

9. M

ultim

edia

Com

mun

icat

ion

Syst

ems

Multimedia Kommunikation SS‘99

56 /

929.

8 R

esou

rce

Man

agem

ent /

Tra

ffic

Sha

ping

- T

oken

Buc

ket w

ith L

eaky

Buc

ket

Tok

en B

ucke

t per

mits

bur

stin

ess

whi

ch is

smoo

thed

by

a su

bseq

uent

leak

y bu

cket

(C >

ρ)

both

buc

kets

hav

e a

size

β

Í b

urst

y tr

affi

c is

per

mitt

ed b

ut th

esc

hem

e re

gula

te it

max

imum

tran

smis

sion

rat

e C

long

term

ave

rage

rat

e bo

unde

d by

ρ (

<C

)

ρ

Tok

en

Buc

ket

Dat

aC

Lea

ky B

ucke

t

Tok

en β

β

Page 57: 9. Multimedia Communication Systems – Reservation Model ...€¦ · 9. Multimedia Communication Systems Multimedia Kommunikation SS‘99 11 / 92 9.2 Cooperative Computing / Group

9. M

ultim

edia

Com

mun

icat

ion

Syst

ems

Multimedia Kommunikation SS‘99

57 /

929.

8 R

esou

rce

Man

agem

ent /

Sch

edul

ing

insi

de th

e N

etw

ork

How

can

we

guar

ante

e th

at th

e sy

stem

kee

ps it

s pr

omis

es f

or th

e co

nnec

tions

even

if tr

affi

c sh

apin

g ru

les

are

viol

ated

?

Fir

st s

tart

egy:

Nag

le´s

alg

orith

m

Rou

ter

in a

net

wor

k ke

eps

a re

cord

and

que

ue f

or e

very

con

nect

ion

and

serv

esth

e co

nnec

tions

in a

pac

ket-

wis

e ro

und-

robi

n st

arte

gy

Adv

anta

ge: C

onne

ctio

ns w

hich

offe

r to

o m

any

pack

ets

will

be

“pun

ishe

d“

Dis

adva

ntag

e:–

a ve

ry lo

ng p

acke

t will

be

“dan

gero

us“

for

othe

rs

–in

conv

enie

nt fo

r so

me

traf

fic p

atte

rns

(if a

pac

ket a

rriv

esju

st a

fter

the

serv

er w

as th

ere

then

the

wai

ting

time

coul

dbe

com

e su

bsta

ntia

l)

–co

stly

and

diff

icul

t to

impl

emen

t

Thu

s so

me

mod

ific

atio

ns w

ere

prop

osed

conn

ectio

n1

conn

ectio

nk

rout

er

Page 58: 9. Multimedia Communication Systems – Reservation Model ...€¦ · 9. Multimedia Communication Systems Multimedia Kommunikation SS‘99 11 / 92 9.2 Cooperative Computing / Group

9. M

ultim

edia

Com

mun

icat

ion

Syst

ems

Multimedia Kommunikation SS‘99

58 /

929.

8 R

esou

rce

Man

agem

ent /

Sch

edul

ing

insi

de th

e N

etw

ork

1. “

Nag

le“:

–a

sepe

rate

inpu

t que

ue (

of p

acke

ts)

per

conn

ectio

n–

roun

d ro

bin

serv

ice

(if n co

nnec

tions

are

act

ive

then

eac

h on

e ca

n se

nd o

nepa

cket

, if a

vaila

ble)

Adv

anta

ge:

–w

ill b

ehav

ing

conn

ectio

ns (

and

only

thes

e!)

will

see

ove

rload

in t

heir

queu

es(is

olat

ion

phen

omen

on)

Lim

itat

ion:

–th

e al

gorit

hm o

f pac

ket l

engt

h, i.

e. c

onne

ctio

n w

ith b

ig p

acke

ts g

et fa

r be

tter

serv

ice

than

con

nect

ions

with

sm

all p

acke

ts–

algo

rithm

is s

ensi

tive

to a

rriv

al p

atte

rns:

If a

pack

et a

rriv

es in

an

empt

y qu

eue

just

afte

r “s

erve

r w

as h

ere“

then

it m

ayha

ve to

wai

t ver

y lo

ng

rout

erco

nnec

tions

conn

ectio

n1

conn

ectio

nk

rout

er

Page 59: 9. Multimedia Communication Systems – Reservation Model ...€¦ · 9. Multimedia Communication Systems Multimedia Kommunikation SS‘99 11 / 92 9.2 Cooperative Computing / Group

9. M

ultim

edia

Com

mun

icat

ion

Syst

ems

Multimedia Kommunikation SS‘99

59 /

929.

8 R

esou

rce

Man

agem

ent /

Sch

edul

ing

insi

de th

e N

etw

ork

2. “

Bit

by

Bit

Ro

un

d R

ob

in“

(fa

ir q

ue

ue

ing

)

–on

e qu

eue

per

conn

ectio

n (a

s be

fore

)

–bu

t: bi

twis

e ro

und

robi

n (in

stea

d as

of p

acke

t wis

e R

ound

Rob

in)

(

this

is o

nly

a th

eore

tical

sch

eme)

3. “

Mod

ified

fair

que

uein

g“ (

by D

emas

, Ker

chav

and

Sch

enke

r)

–pa

cket

wis

e se

rvic

e

–or

der

the

pack

et tr

ansm

issi

ons

acco

rdin

g to

the

time

by w

hich

thei

r la

st b

it

wou

ld b

e tr

asnm

itted

if s

trat

egy

(2.)

wou

ld h

ave

been

app

lied

–a

new

pac

ket w

ill b

e sc

hedu

led

afte

r al

l the

old

pac

kets

Page 60: 9. Multimedia Communication Systems – Reservation Model ...€¦ · 9. Multimedia Communication Systems Multimedia Kommunikation SS‘99 11 / 92 9.2 Cooperative Computing / Group

9. M

ultim

edia

Com

mun

icat

ion

Syst

ems

Multimedia Kommunikation SS‘99

60 /

929.

8 R

esou

rce

Man

agem

ent /

Sch

edul

ing

insi

de th

e N

etw

ork

3. is

sim

ilar

to 2

. but

not

iden

tical

.

Exa

mpl

e:tim

e

arri

val o

f pa

cket

p 1 =

a1 a 2

... a

20

arri

val o

f pa

cket

p2

= b

1 b 2

b3

(p1

and

p 2 f

rom

dif

fere

nt c

onne

ctio

ns)

“Bit

by B

it“ :

a 1.. a

5 a 6

a 7 ...

.. a 20

b 1

b 2 b 3

p 1 le

aves

p 2 le

aves

p 2 st

arts

“Mod

ified

“ :

a 1.. .

......

......

......

.... a

20p 1

leav

es

p 2 st

arts

b 1 b 2

b 3p 2 le

aves

Star

t tim

e of

p2

is d

elay

ed b

y “m

odifi

ed“

(whe

n co

mpa

red

to “

bit-

by-b

it“).

Max

imum

del

ay w

hich

is p

ossi

ble:

Max

imum

pac

ket s

ize

allo

wed

by

the

netw

ork

Page 61: 9. Multimedia Communication Systems – Reservation Model ...€¦ · 9. Multimedia Communication Systems Multimedia Kommunikation SS‘99 11 / 92 9.2 Cooperative Computing / Group

9. M

ultim

edia

Com

mun

icat

ion

Syst

ems

Multimedia Kommunikation SS‘99

61 /

929.

8 R

esou

rce

Man

agem

ent /

Sch

edul

ing

insi

de th

e N

etw

ork

4. “

Wei

ghte

d fa

ir qu

euei

ng“

Fai

r qu

euei

ng g

ives

equ

al fr

actio

n of

ban

dwid

th p

er c

onne

ctio

n (1

/n o

f tot

al

band

wid

th if

n c

onne

ctio

n ar

e ac

tive)

.

Thi

s m

ay n

ot b

e re

ason

able

if (

for

exam

ple)

:

–co

nnec

tion1

is b

etw

een

supe

rcom

pute

rs

–co

nnec

tion

2 is

bet

wee

n w

orks

tatio

ns

In th

is c

ase

we

wou

ld li

ke to

giv

e co

nnec

tion

1 m

ore

band

wid

th.

4.1

Div

ide

the

band

wid

th in

to m

-bit

cycl

es (

m >

n =

num

ber

of a

ctiv

e co

nnec

tions

)

Allo

cate

the

m-n

“ex

tra

bits

“ to

the

conn

ectio

ns w

ith h

ighe

r ne

ed

4.2

Eac

h co

nnec

tion

i get

s a

wei

ghte

d se

rvic

e ra

te g

i fro

m th

e ro

uter

(gi

mus

t be

grea

ter

than

ρi i

f th

e ou

tsid

e tr

affi

c sh

apin

g is

by

toke

n bu

cket

with

leak

y bu

cket

rate

con

trol

; i.e

. if ρ

i is

the

filli

ng r

ate

of th

e to

ken

buck

et f

or c

onne

ctio

n i)

Page 62: 9. Multimedia Communication Systems – Reservation Model ...€¦ · 9. Multimedia Communication Systems Multimedia Kommunikation SS‘99 11 / 92 9.2 Cooperative Computing / Group

9. M

ultim

edia

Com

mun

icat

ion

Syst

ems

Multimedia Kommunikation SS‘99

62 /

929.

8 R

esou

rce

Man

agem

ent /

Sch

edul

ing

- W

eigh

ted

Fair

Que

uein

g

Que

ue p

er c

onne

ctio

n

Rou

nd R

obin

str

ateg

y Í

fai

r

queu

eing

del

ay b

ound

of

conn

ectio

n i

–i d

enot

es th

e flo

w

–D

i den

otes

the

del

ay f

low

i

–β i d

enot

es th

e to

ken

buck

et s

ize

–g i d

enot

es th

e w

eigh

ted

flow

rat

e (g

i≥ρ i)

–l i d

enot

es th

e m

axim

um p

acke

t siz

e of

flow

i

–l a

deno

tes

the

max

imum

pac

ket s

ize

of th

e ne

twor

k

–h i d

enot

es th

e to

tal n

umbe

r of

hop

s of

flow

i

–r m

den

otes

the

outb

ound

ban

dwid

th a

t hop

m

()

whe

re,

1

1∑ =

+⋅

−+

≤ih m

ma

i

ii

iii

rl

g

lh

gD

β

(b1,

ρ1)

(b2,

ρ2)

(b3,

ρ3)

(b4,

ρ4)

(b5,

ρ5)

(b6,

ρ6)

(b7,

ρ7)

(b8,

ρ8)

r i

⋅ ⋅

Page 63: 9. Multimedia Communication Systems – Reservation Model ...€¦ · 9. Multimedia Communication Systems Multimedia Kommunikation SS‘99 11 / 92 9.2 Cooperative Computing / Group

9. M

ultim

edia

Com

mun

icat

ion

Syst

ems

Multimedia Kommunikation SS‘99

63 /

929.

8 R

esou

rce

Man

agem

ent /

Sch

edul

ing

- W

eigh

ted

Fair

Que

uein

g

Proo

f of

del

ay b

ound

:

(*)

dela

y if

eac

h ot

her

rout

er in

the

netw

ork

is f

ully

load

ed; i

n th

is c

ase

the

toke

n bu

cket

siz

e β i

is s

erve

d w

ith a

rat

e g i

(g i

≥ρi).

Thi

s la

sts

β i/ρ

i sec

onds

.

(**)

som

ewha

t mys

teri

ous

term

“de

lay

due

to p

acke

tisat

ion

effe

cts“

(***

) A

pac

ket m

ay b

e de

laye

d (in

eac

h ho

p) b

y l

a b

it tim

es

(l a =

max

imum

pack

et le

ngth

). S

ince

hop

m s

erve

s ra

te r

m th

is n

eeds

l a/r m

sec

onds

.

()

∑ =

+⋅

−+

≤ih m

ma

i

ii

iii

rl

g

lh

gD

1

1β (*

)(*

*)(*

**)

Page 64: 9. Multimedia Communication Systems – Reservation Model ...€¦ · 9. Multimedia Communication Systems Multimedia Kommunikation SS‘99 11 / 92 9.2 Cooperative Computing / Group

9. M

ultim

edia

Com

mun

icat

ion

Syst

ems

Multimedia Kommunikation SS‘99

64 /

929.

8 R

esou

rce

Man

agem

ent /

Sch

edul

ing

- W

eigh

ted

Fair

Que

uein

g

Exa

mpl

e:

Def

ine:

–th

e lin

k ba

ndw

idth

r 1 = r

2 =

r3 =

155

Mbi

t/s (

AT

M s

peed

)

–th

e m

axim

um p

acke

t siz

e of

eac

h flo

w a

nd o

f the

net

wor

k l

1 =

l 2 =

l a =

424

Bit

(AT

M c

ell)

–th

e w

eig

hted

flow

rat

e g

1 =

ρ1 =

100

Mbi

t/s a

nd g 2

= ρ

2 =

50

Mbi

t/s

–th

e to

ken

buck

et s

ize β 1

= β

2 =

1 M

bit

–flo

w 1

is tr

ansm

itted

ove

r 3

hops

(h

1 =

3),

flow

2 o

ver

2 ho

ps (

h 2 = 2

)

The

del

ay b

ound

s ar

e

r 1co

nn. 1

conn

. 2

r 2co

nn. 1

conn

. 2

r 3co

nn. 1

()

ms

01

66,

103

1

11

11

111

=⋅

+⋅

−+

≤rl

g

lh

gD

()

ms

01

395

,20

21

12

22

222

=⋅

+⋅

−+

≤rl

g

lh

gD

dom

inat

ing

term

(in

this

exa

mpl

e!)

Page 65: 9. Multimedia Communication Systems – Reservation Model ...€¦ · 9. Multimedia Communication Systems Multimedia Kommunikation SS‘99 11 / 92 9.2 Cooperative Computing / Group

9. M

ultim

edia

Com

mun

icat

ion

Syst

ems

Multimedia Kommunikation SS‘99

65 /

929.

8 R

esou

rce

Man

agem

ent /

Sch

edul

ing

- W

eigh

ted

Fair

Que

uein

g

Impl

emen

tati

on I

ssue

s:

Que

uein

g:

–P

er-C

onne

ctio

n-Q

ueue

ing:

nee

ds m

ore

com

plex

buf

fer

man

agem

ent

–G

ranu

larit

y of

Que

uein

g: d

epen

ds o

n ne

twor

k, e

.g. c

ell i

n A

TM

, byt

e in

pack

etsw

itche

d ne

twor

ks, b

its in

our

exa

mpl

e

Sche

dulin

g:

–G

ranu

larit

y or

wei

ghts

: uni

t of t

he w

eig

hted

flow

gi?

Í D

epen

ds o

n gr

anul

arity

of q

ueue

ing!

–N

on-w

orkc

onse

rvin

g: n

eeds

tim

er c

ontr

ol if

the

band

wid

th is

not

com

plet

ely

allo

cate

d

Adv

anta

ge: a

naly

tic d

elay

bou

nds!

Dis

adva

ntag

e: m

ore

com

plex

impl

emen

tatio

n

Page 66: 9. Multimedia Communication Systems – Reservation Model ...€¦ · 9. Multimedia Communication Systems Multimedia Kommunikation SS‘99 11 / 92 9.2 Cooperative Computing / Group

9. M

ultim

edia

Com

mun

icat

ion

Syst

ems

Multimedia Kommunikation SS‘99

66 /

929.

8 R

esou

rce

Man

agem

ent /

Sch

edul

ing

- D

elay

Ear

liest

Due

Dat

e (E

DD

)

Inco

min

g pa

cket

s ar

e tim

esta

mpe

d

dead

line

is c

ompu

ted

acco

rdin

g to

–av

erag

e ra

te o

r

–pe

ak r

ate

pack

et is

inse

rted

in c

alen

dar

(tim

e or

dere

d)qu

eue

Sche

dule

r se

rves

pac

ket w

ith d

eadl

ine

sam

e as

actu

al s

ervi

ng ti

me Í

non

-wor

kcon

serv

ing

Í D

-ED

D p

rovi

des

dete

rmin

istic

del

aybo

und

with

gra

nula

rity

bou

nded

by

the

timer

gran

ular

ity!

⋅ ⋅

FIFO

-ord

ered

time

orde

red

⋅ ⋅

reor

derin

gpa

cket

with

sho

rtes

tde

adlin

e is

firs

t in

th

e qu

eue

Page 67: 9. Multimedia Communication Systems – Reservation Model ...€¦ · 9. Multimedia Communication Systems Multimedia Kommunikation SS‘99 11 / 92 9.2 Cooperative Computing / Group

9. M

ultim

edia

Com

mun

icat

ion

Syst

ems

Multimedia Kommunikation SS‘99

67 /

929.

8 R

esou

rce

Man

agem

ent /

Sch

edul

ing

- R

ate

Con

trol

led

Serv

ers

To

rene

w th

e tr

affi

c ch

arac

teri

stic

s in

eve

ryno

de o

f th

e ne

twor

k ra

te c

ontr

olle

rs a

re a

dded

Rat

e C

ontr

olle

r m

ay r

egul

ate

–de

lay

jitte

r (v

aria

tion

of d

elay

)–

rate

jitte

r (v

aria

tion

of r

ate)

Pack

ets

are

hold

by

a de

fine

d el

igib

ility

tim

ebe

fore

sch

edul

ed b

y th

e bo

unde

d de

lay

serv

erÍ

times

tam

p or

deri

ngÍ

addi

tiona

l com

plex

ityÍ

loca

l del

ays

can

be c

ompu

ted

acco

rdin

g to

th

e tr

affi

c ch

arac

teri

stic

s at

the

entr

ance

of

th

e ne

twor

buff

er s

pace

req

uire

d is

dis

trib

uted

to th

e

nod

es a

long

the

netw

ork

path

boun

ded

dela

y se

rver

rate

co

ntro

llers

Cha

nnel

2

Cha

nnel

1

Cha

nnel

n

Page 68: 9. Multimedia Communication Systems – Reservation Model ...€¦ · 9. Multimedia Communication Systems Multimedia Kommunikation SS‘99 11 / 92 9.2 Cooperative Computing / Group

9. M

ultim

edia

Com

mun

icat

ion

Syst

ems

Multimedia Kommunikation SS‘99

68 /

929.

8 R

esou

rce

Man

agem

ent /

Sch

edul

ing

- St

op a

nd G

o

Con

side

r th

e tr

affi

cs c

onsi

stin

g of

fra

mes

with

T b

its (

the

tran

smis

sion

of

a

fram

e ne

eds

T b

it tim

es).

A fr

ame

cont

ains

sev

eral

pac

kets

(“F

ram

e“ =

"Con

tain

er“)

.

The

rou

ter

man

ages

inpu

t fra

mes

and

out

put f

ram

es.

1. A

pac

ket o

f a fr

ame

cann

ot b

e fo

rwar

ded

until

ist e

ntire

fram

e ha

s be

en

r

ecei

ved.

2. “

Sto

p an

d G

o“: T

he p

acke

t will

be

forw

arde

d in

the

next

out

put f

ram

e w

hich

s

tart

s tr

ansm

issi

on a

fter

the

inpu

t fra

me

has

fully

arr

ived

.

Page 69: 9. Multimedia Communication Systems – Reservation Model ...€¦ · 9. Multimedia Communication Systems Multimedia Kommunikation SS‘99 11 / 92 9.2 Cooperative Computing / Group

9. M

ultim

edia

Com

mun

icat

ion

Syst

ems

Multimedia Kommunikation SS‘99

69 /

929.

8 R

esou

rce

Man

agem

ent /

Sch

edul

ing

- St

op a

nd G

o

Min

imum

del

ay o

f a

rout

er:

T b

it tim

es (

nece

ssar

y fo

r re

cept

ion

of th

e fr

ame)

Max

imum

del

ay o

f a

rout

er:

2T b

it tim

es (

if th

e in

put f

ram

e ar

rive

s ju

st a

fter

an

outp

ut f

ram

e ha

s st

arte

d)

TIn

put f

ram

es

Out

put f

ram

es(m

in. d

elay

)

max

imum

del

ay

h ho

ps Í

del

ay b

ound

s h

⋅T ≤

D ≤

2h

⋅T

Page 70: 9. Multimedia Communication Systems – Reservation Model ...€¦ · 9. Multimedia Communication Systems Multimedia Kommunikation SS‘99 11 / 92 9.2 Cooperative Computing / Group

9. M

ultim

edia

Com

mun

icat

ion

Syst

ems

Multimedia Kommunikation SS‘99

70 /

929.

8 R

esou

rce

Man

agem

ent /

Sch

edul

ing

- St

op a

nd G

o

Jitt

er δ

for

“S

top

an

d G

o“:

The

onl

y po

ssib

le ji

tter

is d

ue to

diff

eren

t pos

ition

of a

pac

ket i

n in

put f

ram

eve

rsus

out

put f

ram

e

The

“di

spla

cem

ent“

will

be

in th

e in

terv

al

(-T

: T

)

If th

e fr

ame

ente

rs th

e ne

twor

k at

the

end

of a

fram

e bu

t lea

ves

it at

the

begi

nnin

g of

a fr

ame

Page 71: 9. Multimedia Communication Systems – Reservation Model ...€¦ · 9. Multimedia Communication Systems Multimedia Kommunikation SS‘99 11 / 92 9.2 Cooperative Computing / Group

9. M

ultim

edia

Com

mun

icat

ion

Syst

ems

Multimedia Kommunikation SS‘99

71 /

929.

8 R

esou

rce

Man

agem

ent /

Sch

edul

ing

- R

ate

Con

trol

led

Stat

ic P

rior

ity

Eac

h ch

anne

l has

ist o

wn

rate

con

trol

ler

dela

yed

pack

et is

inse

rted

in F

CFS

que

ues

FCFS

que

ues

are

serv

ed w

ith p

rior

ities

(no

n-pr

eem

ptiv

e)

n de

lay

boun

ds a

re s

uppo

rted

- c

ompu

tatio

n

depe

nds

on tr

affi

c sp

ecif

icat

ion

- de

lay

boun

ds

for

TE

NE

T tr

affi

c sp

ecif

icat

ion

exis

ts (

Zha

ng)

sche

dule

r is

non

-wor

kcon

serv

ing

firs

tim

plem

enta

tion

in h

ardw

are

exis

ts

Cha

nnel

2

Cha

nnel

1

Cha

nnel

m

. . .

⋅ ⋅

m r

ate

cont

rolle

rsn

FCFS

serv

ers

. . .

Page 72: 9. Multimedia Communication Systems – Reservation Model ...€¦ · 9. Multimedia Communication Systems Multimedia Kommunikation SS‘99 11 / 92 9.2 Cooperative Computing / Group

9. M

ultim

edia

Com

mun

icat

ion

Syst

ems

Multimedia Kommunikation SS‘99

72 /

929.

8 R

esou

rce

Man

agem

ent /

Sch

edul

ing

- Ji

tter

Ear

liest

Due

Dat

e

each

cha

nnel

has

its

own

rate

con

trol

ler

diff

eren

ce b

etw

een

dead

line

and

actu

al f

inis

hing

time

at c

urre

nt s

erve

r is

car

ried

in th

e pa

cket

abov

e de

scri

bed

diff

eren

ce is

the

time

the

pack

etis

hol

d ba

ck in

the

regu

lato

r of

the

follo

win

gno

de (

i.e. n

ext h

op c

ompe

nsat

es f

itter

of

prev

ious

hop

)

dead

line

in ti

me-

orde

red

queu

e is

com

pute

dsi

mila

r to

D-E

DD

ÍJ-

ED

D p

rovi

de b

ound

s fo

r lo

cal d

elay

a

nd lo

cal d

elay

jitte

need

s ad

ditio

nal t

imes

tam

p fi

eld

in

pac

ket h

eade

r

Cha

nnel

2

Cha

nnel

1

Cha

nnel

m

. . .

⋅ ⋅

m r

ate

cont

rolle

rs

⋅ ⋅

A h

op h

as :

a qu

euei

ng b

udge

t b

a

jitte

r bo

und

j

Tra

nsm

issi

on o

f a

pack

et m

ay b

e st

arte

d at

at t

ime

in th

e in

terv

al [

b-j :

b]

. . .

Page 73: 9. Multimedia Communication Systems – Reservation Model ...€¦ · 9. Multimedia Communication Systems Multimedia Kommunikation SS‘99 11 / 92 9.2 Cooperative Computing / Group

9. M

ultim

edia

Com

mun

icat

ion

Syst

ems

Multimedia Kommunikation SS‘99

73 /

929.

8 R

esou

rce

Man

agem

ent /

Sch

edul

ing

- Ji

tter

Ear

liest

Due

Dat

e

If th

e pa

cket

is s

tart

ed a

t tim

e b-

τ (i

.e. τ

sec

onds

ear

ly)

then

in th

e ne

xt h

op

τ w

ill b

e ad

ded

to th

e bu

dget

of

the

next

hop

, i.e

. the

tran

smis

sion

in th

e ne

xt

hop

wil

l be

sche

dule

d in

the

inte

rval

[b´-

j : b

´] w

here

b´=

τ +

b.

Í G

loba

l jitt

er is

lim

ited

to x

∈ [o

:j].

Tec

hnic

al p

robl

em

:

The

act

ual v

alue

τ ha

s to

be

fille

d in

into

the

pack

et´s

hea

der.

Page 74: 9. Multimedia Communication Systems – Reservation Model ...€¦ · 9. Multimedia Communication Systems Multimedia Kommunikation SS‘99 11 / 92 9.2 Cooperative Computing / Group

9. M

ultim

edia

Com

mun

icat

ion

Syst

ems

Multimedia Kommunikation SS‘99

74 /

929.

8 R

esou

rce

Man

agem

ent /

End

-to-

End

Err

or C

ontr

ol

Rel

iabi

lity

of

mul

tim

edia

str

eam

s is

nec

essa

ry b

ecau

se–

Dec

ompr

essi

on: m

ost a

udio

and

vid

eo c

ompr

essi

on s

chem

es c

anno

t tol

erat

e lo

ssof

dat

a

–H

uman

per

cept

ion:

Los

s of

e.g

. dig

ital a

udio

is d

etec

ted

by a

hum

an e

ar v

ery

quic

kly Í

low

er a

ccep

tanc

e of

mul

timed

ia s

yste

ms

–D

ata

inte

grity

: E.g

. in

a re

cord

ing

appl

icat

ion,

one

can

not r

ecov

er fr

om a

n er

ror

in th

e fir

st r

ecor

ding

, for

tuna

tely

, rec

ordi

ng a

pplic

atio

ns h

ave

ofte

n le

ssst

ringe

nt r

eal-t

ime

requ

irem

ents

for

the

rece

iver

End

-to-

End

err

or c

ontr

ol c

onsi

sts

of–

erro

r de

tect

ion

–er

ror

corr

ectio

n

Page 75: 9. Multimedia Communication Systems – Reservation Model ...€¦ · 9. Multimedia Communication Systems Multimedia Kommunikation SS‘99 11 / 92 9.2 Cooperative Computing / Group

9. M

ultim

edia

Com

mun

icat

ion

Syst

ems

Multimedia Kommunikation SS‘99

75 /

929.

8 R

esou

rce

Man

agem

ent /

End

-to-

End

Err

or C

ontr

ol -

Err

or D

etec

tion

Prob

lem

: iso

late

the

erro

rs, e

.g. s

ome

wro

ng c

olor

in a

fra

me

may

not

mat

ter

(har

dly

visi

ble

to h

uman

use

r) b

ut w

rong

fra

me

boun

dari

es c

anno

t be

reco

vere

d

Exa

mpl

e: w

ithi

n an

MPE

G-2

vid

eo it

is im

port

ant n

ot to

loos

e th

e I-

Fram

es(c

onta

ins

stru

ctur

al in

form

atio

ns),

P-

and

B-F

ram

es lo

sses

are

tole

rate

d

Í s

truc

tura

l inf

orm

atio

n ne

eds

to b

e pr

otec

ted,

not

alw

ays

cont

ent!

Íex

istin

g er

ror

dete

ctio

n m

echa

nism

s (c

heck

sum

min

g, s

eque

ncin

g) m

ust b

eex

tend

ed

Exa

mpl

e: la

tene

ss c

once

pt. P

DU

s th

at a

rriv

e to

o la

te a

re u

sele

ss f

or th

eap

plic

atio

n (e

.g. v

ideo

fra

me)

ÍPD

Us

are

peri

odic

ally

tim

e-st

ampe

d

Íer

ror

dete

ctio

n ca

n st

art a

t the

tim

e-st

ampe

d PD

Us

Íre

quir

es s

ynch

roni

zed

syst

em c

lock

at s

ende

r an

d re

ceiv

er (

e.g.

Net

wor

k T

ime

Prot

ocol

)

Page 76: 9. Multimedia Communication Systems – Reservation Model ...€¦ · 9. Multimedia Communication Systems Multimedia Kommunikation SS‘99 11 / 92 9.2 Cooperative Computing / Group

9. M

ultim

edia

Com

mun

icat

ion

Syst

ems

Multimedia Kommunikation SS‘99

76 /

929.

8 R

esou

rce

Man

agem

ent /

End

-to-

End

Err

or C

ontr

ol -

Err

or C

orre

ctio

n

Tra

diti

onal

met

hod

for

reli

abil

ity

prov

isio

n:

–R

etra

nsm

issi

on

Not

sui

tabl

e fo

r m

ultim

edia

com

mun

icat

ion

beca

use:

–ex

plic

it ac

know

ledg

emen

t inc

reas

es th

e am

ount

of d

ata

to b

e st

ored

in th

ese

nder

, e.g

. vid

eo d

ata

–w

indo

w-b

ased

flow

con

trol

may

forc

e th

e se

nder

to s

uspe

nd th

e (c

ontin

uous

)da

ta s

trea

m li

ke v

ideo

–re

tran

smitt

ed d

ata

mig

ht b

e re

ceiv

ed to

o la

te to

be

cons

umed

in ti

me

–tr

aditi

onal

mec

hani

sms

do n

ot s

cale

to m

ultip

le-t

arge

t com

mun

icat

ion

like

mul

ticas

t sce

nario

s -

they

are

typi

cally

des

igne

d fo

r po

int-

to-p

oint

com

mun

icat

ion

Page 77: 9. Multimedia Communication Systems – Reservation Model ...€¦ · 9. Multimedia Communication Systems Multimedia Kommunikation SS‘99 11 / 92 9.2 Cooperative Computing / Group

9. M

ultim

edia

Com

mun

icat

ion

Syst

ems

Multimedia Kommunikation SS‘99

77 /

929.

8 R

esou

rce

Man

agem

ent /

End

-to-

End

Err

or C

ontr

ol -

Err

or C

orre

ctio

n

Cur

rent

ly d

iscu

ssed

sch

emes

for

mul

timed

ia c

omm

unic

atio

n: G

o-B

ack-

n R

etra

nsm

issi

on

SR

1 2 3 4 5 6 7 8 4 5 6 7 8 9

t

SR

1 2 3 4 5 6 7 4 5 6 7 8 9 10

t

Tim

er c

ontr

olle

dm

ay le

ad to

gap

intr

oduc

tion

and

viol

atio

n of

thro

ughp

ut p

aram

eter

s

Usa

ge o

f sp

ecia

lne

twor

k pr

oper

ties

like

sequ

ence

ord

erin

g in

AT

M

Tim

er e

nds

and

the

rece

iver

no

tice

the

mis

sing

of

pack

et 4

Rec

eive

r st

arts

the

Tim

er

The

se p

acke

ts a

re lo

st

Aft

er r

ecei

ving

the

pack

et 5

, the

rec

eive

rno

tice

the

mis

sing

of

pack

et 4

.

All

lost

pac

kets

mus

t be

ret

rans

mit

ted

Page 78: 9. Multimedia Communication Systems – Reservation Model ...€¦ · 9. Multimedia Communication Systems Multimedia Kommunikation SS‘99 11 / 92 9.2 Cooperative Computing / Group

9. M

ultim

edia

Com

mun

icat

ion

Syst

ems

Multimedia Kommunikation SS‘99

78 /

929.

8 R

esou

rce

Man

agem

ent /

End

-to-

End

Err

or C

ontr

ol -

Err

or C

orre

ctio

n

Cur

rent

ly d

iscu

ssed

sch

emes

for

mul

timed

ia c

omm

unic

atio

n: S

elec

tive

Ret

rans

mis

sion

SR

1 2 3 4 5 6 7 4 8 9 10 11 12 13

t

SR

1 2 3 4 5 6 7 8 9 4 10 11 12 13

t

Tim

er c

ontr

olle

dU

sage

of

spec

ial n

etw

ork

prop

erti

es li

kese

quen

ce o

rder

ing

in A

TM

impl

enta

tion

mor

e co

mpl

icat

ed m

ore

buff

er n

eeds

Tim

er e

nds

and

the

rece

iver

no

tice

the

mis

sing

of

pack

et 4

Rec

eive

r st

arts

the

Tim

er

The

pac

kets

are

not

lost

Aft

er r

ecei

ving

the

pack

et 5

, the

rec

eive

rno

tice

the

mis

sing

of

pack

et 4

.

Onl

y th

e pa

cket

4 h

as to

be

ret

rans

mit

Page 79: 9. Multimedia Communication Systems – Reservation Model ...€¦ · 9. Multimedia Communication Systems Multimedia Kommunikation SS‘99 11 / 92 9.2 Cooperative Computing / Group

9. M

ultim

edia

Com

mun

icat

ion

Syst

ems

Multimedia Kommunikation SS‘99

79 /

929.

8 R

esou

rce

Man

agem

ent /

End

-to-

End

Err

or C

ontr

ol -

Err

or C

orre

ctio

n

Part

ially

Rel

iabl

e St

ream

s:–

num

ber

of r

etra

nsm

itted

pac

kets

are

lim

ited

and

calc

ulat

ed fr

om th

e tim

ing

cons

trai

nts

of th

e ca

ll se

tup

Forw

ard

Err

or C

orre

ctio

n (F

EC

):–

addi

tiona

l inf

orm

atio

n is

add

ed to

the

data

Í

low

end

-to-

end

dela

y an

d no

excl

usiv

e bu

fferin

g. F

EC

wor

ks o

nly

for

erro

r de

tect

ion/

corr

ectio

n w

ithin

pack

ets

but n

ot fo

r pa

cket

loss

Prio

rity

Cha

nnel

Cod

ing:

–m

ultip

le d

ata

stre

ams

are

dist

ingu

ishe

d by

diff

eren

t prio

ritie

s, d

urin

g co

nges

tion

perio

ds th

e ne

twor

k is

mor

e lik

ely

to d

isca

rd lo

w-p

riorit

y pa

cket

s. F

or e

xam

ple,

I-F

ram

es in

MP

EG

-2 a

re tr

ansm

itted

with

hig

h, B

- an

d P

-Fra

mes

with

low

prio

rity

Page 80: 9. Multimedia Communication Systems – Reservation Model ...€¦ · 9. Multimedia Communication Systems Multimedia Kommunikation SS‘99 11 / 92 9.2 Cooperative Computing / Group

9. M

ultim

edia

Com

mun

icat

ion

Syst

ems

Multimedia Kommunikation SS‘99

80 /

92

part

ial r

etra

nsm

issi

on s

chem

espr

even

tive

mec

hani

sms

Go-

Bac

k-N

Sele

ctiv

e R

etra

nsm

issi

onPa

rtia

lly R

elia

ble

Stre

ams

Forw

ard

Err

or C

orre

ctio

n (F

EC

)Pr

iori

ty C

hann

el C

odin

g

9.8

Res

ourc

e M

anag

emen

t / E

nd-t

o-E

nd E

rror

Con

trol

- E

rror

Cor

rect

ion

Err

or c

ontr

ol/c

orre

ctio

n sc

hem

es a

re d

ivid

ed in

Part

ial r

etra

nsm

issi

on s

chem

es la

ck th

e po

ssib

ility

of

intr

oduc

ing

adi

scon

tinui

ty o

r la

rge

end-

to-e

nd d

elay

s w

ith la

rge

buff

ers Í

pre

vent

ive

sche

mes

sho

uld

be u

sed

Page 81: 9. Multimedia Communication Systems – Reservation Model ...€¦ · 9. Multimedia Communication Systems Multimedia Kommunikation SS‘99 11 / 92 9.2 Cooperative Computing / Group

9. M

ultim

edia

Com

mun

icat

ion

Syst

ems

Multimedia Kommunikation SS‘99

81 /

929.

8 R

esou

rce

Man

agem

ent /

For

war

d E

rror

Cor

rect

ion

- C

RC

Cyc

lic

Red

unda

ncy

Che

cks

(CR

C)

are

used

to p

rote

ct p

acke

ts a

gain

st b

it er

rors

.

Giv

en m

dat

a bi

ts (

w1,

...,

wm

).

Def

ine

and

the

gene

rato

r po

lyno

mia

l G(x

) of

deg

ree

k ac

cord

ing

to

Usi

ng th

e m

odul

o-2

arith

met

ic c

ompu

te

∑ =

−⋅

=m i

ii

xw

xU

1

1)

(

{}

∑ =

==

∀∈

⋅=

k ik

ii

ig

gi

gx

gx

G0

01

;1,0

)

(

()

)(

)(

)(

)(

)(

)(

Re

)(

xR

xG

xP

xU

xx

G

xU

xm

xR

kk

+⋅

=⋅

⋅=

Page 82: 9. Multimedia Communication Systems – Reservation Model ...€¦ · 9. Multimedia Communication Systems Multimedia Kommunikation SS‘99 11 / 92 9.2 Cooperative Computing / Group

9. M

ultim

edia

Com

mun

icat

ion

Syst

ems

Multimedia Kommunikation SS‘99

82 /

92

=

+⋅

=

+−

⋅=

)(

)(

Re

)(

)(

)(

)(

)(

Re

)(

)(

)(

)(

Re

)(

xG

xe

mx

G

xe

xG

xG

xP

mx

G

xe

xR

xU

xm

xV

k

9.8

Res

ourc

e M

anag

emen

t / F

orw

ard

Err

or C

orre

ctio

n -

CR

C

The

res

t pol

ynom

ial R

(x)

is o

f de

gree

< k

and

is a

ppen

ded

at th

e m

dat

a bi

ts a

ttr

ansf

er Í

xk

⋅ U(x

) +

R(x

) is

tras

nmitt

ed (

xk sh

ifts

the

data

bits

to p

rovi

de b

itsfo

r th

e re

mai

nder

)

The

rec

eive

r re

ceiv

es x

k ⋅ U

(x)

+ R

(x)

+ e

(x)

(e(x

) is

the

erro

r po

lyno

mia

l) a

ndco

mpu

tes

If V

(x)

≠0 Í

1. e

i > 0

in th

e da

ta p

art Í

ser

ious

tras

nmis

sion

err

or

2

. ei >

0 in

the

rem

aind

er p

art Í

can

´t b

e di

stin

guis

hed

from

seri

ous

erro

rs

If V

(x)

= 0

an

unde

tect

able

err

or m

ay b

e oc

cure

d if

e(x)

=a(

x)G

(x)

(rat

her

unlik

ely

even

t)

Page 83: 9. Multimedia Communication Systems – Reservation Model ...€¦ · 9. Multimedia Communication Systems Multimedia Kommunikation SS‘99 11 / 92 9.2 Cooperative Computing / Group

9. M

ultim

edia

Com

mun

icat

ion

Syst

ems

Multimedia Kommunikation SS‘99

83 /

929.

8 R

esou

rce

Man

agem

ent /

For

war

d E

rror

Cor

rect

ion

- C

RC

Exa

mpl

e:

(w1,

...,

w8)

= (

1, 1

, 0, 1

, 0, 1

, 1, 1

) Í

U(x

) =

x7

+ x

6 +

x4

+ x

2 +

x1

+ 1

Def

ine

G(x

) =

x5

+ x

3 +

1

Í R

(x)

= x

4 +

x3

+ x

Í (

1101

0111

110

10)

is tr

ansm

itted

Rec

eive

r ge

ts (

1100

0111

110

10)

Thu

s e(

x) =

x9

ÍV

(x)

≠0 Í

seri

ous

tran

smis

sion

err

or!

Err

or d

etec

tion

is o

ften

rea

lized

in h

ardw

are

(e.g

. AT

M).

Err

or c

orre

ctio

n w

ith

CR

C is

mor

e co

mpl

icat

ed (

Max

imum

-Lik

elih

ood

met

hod)

.

Exa

mpl

e: A

TM

hea

ders

are

pro

tect

ed w

ith G

(x)

= x

8 +

x2

+ x

+ 1

(ca

n be

use

d to

corr

ect a

sin

gle

erro

r!)

data

CR

C

erro

r!

Page 84: 9. Multimedia Communication Systems – Reservation Model ...€¦ · 9. Multimedia Communication Systems Multimedia Kommunikation SS‘99 11 / 92 9.2 Cooperative Computing / Group

9. M

ultim

edia

Com

mun

icat

ion

Syst

ems

Multimedia Kommunikation SS‘99

84 /

929.

8 R

esou

rce

Man

agem

ent /

Res

ourc

e M

onito

ring

Res

ourc

e M

onit

orin

g is

an

impo

rtan

t par

t of

reso

urce

man

agem

ent i

n ne

twor

ks

and

et e

nd-p

oint

s!

–M

onito

ring

in n

etw

orks

add

s ov

erhe

ad

Ím

ost o

f the

mon

itorin

g va

riabl

es

shou

ld b

e op

tiona

l and

mon

itorin

g sh

ould

be

able

to tu

rn o

n/of

f

–M

onito

ring

in e

nd-s

yste

ms

incl

udes

sup

ervi

sor

func

tion

to c

ontin

uous

ly m

onito

r

the

proc

esse

d Q

oS p

aram

eter

s (e

.g. c

ontr

ollin

g a

com

pres

sion

com

pone

nt)

Íde

sign

and

impl

emen

tatio

n is

non

-triv

ial

Page 85: 9. Multimedia Communication Systems – Reservation Model ...€¦ · 9. Multimedia Communication Systems Multimedia Kommunikation SS‘99 11 / 92 9.2 Cooperative Computing / Group

9. M

ultim

edia

Com

mun

icat

ion

Syst

ems

Multimedia Kommunikation SS‘99

85 /

929.

8 R

esou

rce

Man

agem

ent /

Res

ourc

e M

onito

ring

Res

ourc

e A

dmin

istr

atio

n Pr

otoc

ols

prov

ide

com

mun

icat

ion

abou

t res

ourc

es

betw

een

reso

urce

man

ager

s at

inte

rmed

iate

nod

es a

nd e

nd-s

yste

ms

and

can

be

impl

emen

ted

as

–pa

rt o

f net

wor

k m

anag

emen

t pro

toco

ls li

ke C

MIS

/CM

IP (

Com

mon

Man

agem

ent I

nfor

mat

ion

Ser

vice

and

Pro

toco

ls)

and

SN

MP

(S

impl

e N

etw

ork

Man

agem

ent P

roto

col)

–se

para

te r

esou

rce

man

agem

ent p

roto

cols

like

RC

AP

in th

e T

enet

pro

toco

l sta

ck

Adm

inis

trat

ion

Prot

ocol

may

be

part

of

the

tran

smis

sion

pro

toco

l lik

e ST

-II

or

like

RC

AP

whi

ch is

dec

oupl

ed f

rom

RT

IP (

Rea

l Tim

e In

tern

et P

roto

col)

Page 86: 9. Multimedia Communication Systems – Reservation Model ...€¦ · 9. Multimedia Communication Systems Multimedia Kommunikation SS‘99 11 / 92 9.2 Cooperative Computing / Group

9. M

ultim

edia

Com

mun

icat

ion

Syst

ems

Multimedia Kommunikation SS‘99

86 /

929.

8 R

esou

rce

Man

agem

ent /

Res

ourc

e A

dapt

ion

Dyn

amic

ally

cha

ngin

g th

e ne

twor

k ca

paci

ty o

f ea

ch s

essi

on n

eeds

sup

port

of

dyna

mic

chan

ge o

f Q

oS p

aram

eter

s

Res

ourc

e ad

aptio

n re

quir

es–

notif

icat

ion

and

rene

gotia

tion

of Q

oS p

aram

eter

s

–a

dapt

ive

reso

urce

sch

emes

Ren

egot

iatio

n re

ques

t can

com

e fr

om e

ither

–th

e us

er (

e.g.

cha

nge

QoS

like

vid

eo r

ate)

who

invo

kes

the

reso

urce

adm

inis

trat

ion

prot

ocol

to c

heck

the

reso

urce

if n

etw

ork

adap

tion

is r

equi

red,

–th

e ho

st s

yste

m (

e.g.

due

to w

orks

tatio

n ov

erlo

ad)

whi

ch in

voke

s e.

g. a

not

ifica

tion

ofQ

oS d

egra

datio

n to

a m

isbe

have

d us

er/a

pplic

atio

n. N

etw

ork

QoS

ren

egot

iatio

n m

ay a

lso

be in

voke

d by

the

reso

urce

adm

inis

trat

ion

prot

ocol

–th

e ne

twor

k (d

ue to

ove

rload

or

cong

estio

n) w

hich

invo

kes

a no

tific

atio

n to

the

host

repo

rtin

g th

at th

e al

loca

tion

of r

esou

rces

mus

t cha

nge

·n

etw

ork

ca

n ad

apt

to o

verl

oad Í

ne

two

rk n

otifi

es

the

ho

st b

eca

use

som

e d

eg

rad

atio

n m

ay

occ

ur

du

rin

g m

odifi

catio

n o

f re

sou

rce

s Í

th

is m

ay

inte

rru

pt t

he m

ed

ia f

low

Í

hos

t m

ust

reac

t

·n

etw

ork

ca

nnot

ada

pt t

o o

verl

oad

Í

hos

t m

ust

ad

ap

t to

ove

rlo

ad

Page 87: 9. Multimedia Communication Systems – Reservation Model ...€¦ · 9. Multimedia Communication Systems Multimedia Kommunikation SS‘99 11 / 92 9.2 Cooperative Computing / Group

9. M

ultim

edia

Com

mun

icat

ion

Syst

ems

Multimedia Kommunikation SS‘99

87 /

929.

8 R

esou

rce

Man

agem

ent /

Res

ourc

e A

dapt

ion

- N

etw

ork

Ada

ptio

n

Prop

er b

alan

cing

of

netw

ork

load

is d

esir

able

and

nec

essa

ry to

–in

crea

se n

etw

ork

avai

labi

lity

–re

clai

m r

esou

rces

by

netw

ork

adm

inis

trat

ors

–re

duce

the

impa

ct o

f uns

ched

uled

, run

-tim

e m

aint

enan

ce o

n cl

ient

s w

ithgu

aran

teed

ser

vice

s

Impl

emen

ting

load

bal

anci

ng p

olic

y re

quir

es

–ro

utin

g, p

erfo

rman

ce m

onito

ring

(det

ect l

oad

chan

ges)

–dy

nam

ic r

e-ro

utin

g

–lo

ad b

alan

cin

g co

ntro

l to

mak

e a

deci

sion

to r

e-ro

ute

a ch

anne

l

Dyn

amic

re-

rout

ing

mec

hani

sm is

don

e by

est

ablis

hing

a n

ew c

hann

el (

shad

owch

anne

l) v

ia a

new

rou

te, s

endi

ng d

ata

on th

is n

ew c

hann

el a

nd te

arin

g do

wn

the

old

chan

nel a

fter

the

max

imum

end

-to-

end

dela

y.

Page 88: 9. Multimedia Communication Systems – Reservation Model ...€¦ · 9. Multimedia Communication Systems Multimedia Kommunikation SS‘99 11 / 92 9.2 Cooperative Computing / Group

9. M

ultim

edia

Com

mun

icat

ion

Syst

ems

Multimedia Kommunikation SS‘99

88 /

929.

8 R

esou

rce

Man

agem

ent /

Res

ourc

e A

dapt

ion

- So

urce

Ada

ptio

n

Ada

ptio

n of

the

sour

ce r

ate

acco

rdin

g to

the

curr

ent n

etw

ork

reso

urce

s by

–ev

olu

tion

of th

e sy

stem

sta

te o

ver

the

time:

sta

te o

f the

com

plet

e ro

ute

is u

sed

toco

mpu

te th

e se

ndin

g ra

te o

f th

e so

urce

·st

ate

info

rmat

ion

may

be

appe

nded

per

iodi

cally

to d

ata

pack

ets

and

is s

ent b

ack

toth

e so

urce

. Int

erm

edia

te n

odes

upd

ate

the

info

rmat

ion

if th

e sw

itch

serv

ice

rate

islo

wer

than

the

prop

osed

rat

e

·fe

edba

ck is

sen

t with

sep

arat

e co

ntro

l mes

sage

s to

the

rece

iver

and

bac

k to

the

sour

ce

–R

ate

cont

rol u

sing

net

wor

k fe

edba

ck: c

hang

es in

the

traf

fic c

ondi

tions

are

dete

cted

by

feed

back

from

the

netw

ork

·e

xplic

it: in

form

atio

n ab

out t

he tr

affic

load

or

buffe

r sp

ace

from

the

netw

ork

·im

plic

it: in

form

atio

n ab

out p

acke

t los

ses

and

dela

y fr

om a

ckno

wle

dgem

ents

–T

raffi

c sh

apin

g at

sou

rce:

sm

ooth

ing

traf

fic a

t the

sou

rce

Ítr

affic

is le

ss b

urst

y

–H

iera

rchi

cal c

odin

g: a

lgor

ithm

s pr

oduc

ing

two

or m

ore

type

s of

cel

ls d

escr

ibin

gsa

me

bloc

k of

pix

els

with

diff

eren

t deg

ree

of d

etai

l Í

code

cs a

re m

ore

com

plex

,us

e gr

eate

r am

ount

of b

andw

idth

Page 89: 9. Multimedia Communication Systems – Reservation Model ...€¦ · 9. Multimedia Communication Systems Multimedia Kommunikation SS‘99 11 / 92 9.2 Cooperative Computing / Group

9. M

ultim

edia

Com

mun

icat

ion

Syst

ems

Multimedia Kommunikation SS‘99

89 /

929.

9 A

rchi

tect

ure

/ Iss

ues

End

poin

t arc

hite

ctur

es n

eed

to in

corp

orat

e co

mpo

nent

s lik

e–

reso

urce

man

ager

s w

hich

incl

ude

·tr

ansl

atio

n se

rvic

es (

tran

slat

e Q

oS)

·a

dmis

sion

con

trol

(ch

ecks

for

avai

labi

lity

of r

esou

rces

reso

urce

res

erva

tion

·re

sour

ce m

anag

emen

t

–se

rvic

e ag

ents

–M

anag

emen

t Inf

orm

atio

n B

ases

(M

IBs)

with

QoS

spe

cific

atio

n

Rou

ter/

Switc

h ar

chit

ectu

re n

eed

to e

mpl

oy–

reso

urce

man

ager

s w

hich

incl

ude

·pa

cke

t cla

ssifi

er

·pa

cket

sch

edul

er·

adm

issi

on c

ontr

ol

–ne

twor

k m

anag

emen

t

–M

anag

emen

t Inf

orm

atio

n B

ases

Page 90: 9. Multimedia Communication Systems – Reservation Model ...€¦ · 9. Multimedia Communication Systems Multimedia Kommunikation SS‘99 11 / 92 9.2 Cooperative Computing / Group

9. M

ultim

edia

Com

mun

icat

ion

Syst

ems

Multimedia Kommunikation SS‘99

90 /

929.

9 A

rchi

tect

ure

/ Exa

mpl

es

Lan

cast

er´s

QoS

-Arc

hite

ctur

e in

corp

orat

es n

otio

ns o

f flo

w, s

ervi

ce c

ontr

act a

ndflo

w m

anag

emen

t ove

r hi

gh-p

erfo

rman

ce A

TM

-bas

ed

netw

orks

Hei

delb

erg

Tra

nspo

rt S

yste

m (

Hei

TS

) pr

ovid

es c

ontin

uous

med

ia e

xcha

nge

with

QoS

gua

rant

ees

(del

ay, j

itter

, thr

ough

put,

relia

bilit

y) b

ased

on

ST

-II,

reso

urce

man

agem

ent a

nd r

eal-t

ime

mec

hani

sms

via

mul

ticas

t

Ten

et P

roto

col S

uite

with

RC

AP

(re

al-t

ime

Cha

nnel

Adm

inis

trat

ion

Pro

toco

l),

RT

IP (

real

-tim

e IP

), R

MT

P (

Rea

l-tim

e M

essa

ge T

rans

port

Pro

toco

l) an

dC

MT

P (

Con

tinuo

us M

edi

a T

rans

port

Pro

toco

l) pr

ovid

es Q

oS n

egot

iatio

n,re

sour

ce a

dmin

istr

atio

n an

d re

al-t

ime

med

ia tr

ansp

ort

Page 91: 9. Multimedia Communication Systems – Reservation Model ...€¦ · 9. Multimedia Communication Systems Multimedia Kommunikation SS‘99 11 / 92 9.2 Cooperative Computing / Group

9. M

ultim

edia

Com

mun

icat

ion

Syst

ems

Multimedia Kommunikation SS‘99

91 /

929.

9 A

rchi

tect

ure

/ Exa

mpl

es

OSI

arc

hite

ctur

e pr

ovid

es Q

oS in

net

wor

k la

yer

and

enha

ncem

ents

in tr

ansp

ort

laye

r. O

SI95

con

side

rs Q

oS s

peci

fica

tion

and

nego

tiatio

n in

tran

spor

t pro

toco

ls

RSV

P, b

ased

on

IP, p

rovi

des

reso

urce

res

erva

tion

UP

enn´

s en

d-po

int a

rchi

tect

ure

(OM

EG

A A

rchi

tect

ure)

pro

vide

s Q

oS h

andl

ing

and

man

agem

ent a

t app

licat

ions

and

tran

spor

t sub

syst

ems

Na

tive

Mod

e A

TM

Pro

toco

l pro

vide

s A

TM

net

wor

k Q

oS g

uara

ntee

s

Page 92: 9. Multimedia Communication Systems – Reservation Model ...€¦ · 9. Multimedia Communication Systems Multimedia Kommunikation SS‘99 11 / 92 9.2 Cooperative Computing / Group

9. M

ultim

edia

Com

mun

icat

ion

Syst

ems

Multimedia Kommunikation SS‘99

92 /

929.

9 T

rend

s in

Tra

nspo

rt S

yste

ms

Tw

o di

rect

ions

:

–sp

ecia

l-pur

pose

pro

toco

ls li

ke T

CP

(re

liabl

e da

ta tr

ansf

er),

UD

P (

unre

liabl

e da

ta

tran

sfer

), R

TP

(au

dio

/vid

eo tr

ansf

er)

on to

p of

IP (

inte

rnet

par

adig

m)

–ge

nera

l-pur

pose

pro

toco

ls li

ke X

TP

to p

rovi

de a

gen

eral

set

of s

ervi

ces

Í a

pplic

atio

n-ta

ilore

d pr

otoc

ols:

cus

tom

ized

for

spe

cifi

c ty

pes

of s

ervi

ces

(

tran

sfer

ring

voi

ce, v

ideo

, tex

t, im

age)

Í tr

ansp

ort s

ubsy

tem

s m

ove

tow

ard

a pr

ovis

ion

of s

ervi

ce c

lass

es li

ke

g

uara

ntee

d se

rvic

es, b

est e

ffor

t

Í w

ithin

thes

e cl

asse

s, s

ervi

ces

are

cust

omiz

ed th

roug

h Q

oS s

peci

fica

tion

Í s

ervi

ces

are

sele

cted

and

cus

tom

ized

tow

ard

appl

icat

ions