Communication Networks Stand und zukünftige attraktive Arbeitsgebiete für den Lehrstuhl für Kommunikationsnetze Prof. Dr.-Ing. Bernhard Walke Kommunikationsnetze,

Communication Networks

Stand und zukünftige attraktive Arbeitsgebiete Stand und zukünftige attraktive Arbeitsgebiete für den für den

Lehrstuhl für KommunikationsnetzeLehrstuhl für Kommunikationsnetze

Prof. Dr.-Ing. Bernhard WalkeProf. Dr.-Ing. Bernhard WalkeKommunikationsnetze, RWTH AachenKommunikationsnetze, RWTH Aachen

[email protected]@comnets.rwth-aachen.de

Sitzung des vorbereitenden Berufungsausschusses Kommunikationsnetze

Mo. 19. Dezember 2005

2Communication Networks, Aachen University (RWTH)

Long Term 3G Evolution

>2008

Fix

ed

Walk

Veh

icle

Mobility / Range

Indoor

Pedestrian

High Speed

VehicularRural

Personal Area

VehicularUrban

Fixed urban

User data rate

10 Mbps0.1

IEEE802.16a,d

1 100

HSDPAIEEE

802.16eNomadic

WLAN(IEEE 802.11x)

GSMGPRS

DECT

bluetooth

3G/WCDMA

EDGE

FlashOFDM (802.20)

Wireless Technology PositioningWireless Technology Positioning


Facts in Communication Networks and ProtocolsFacts in Communication Networks and Protocols

• Digital networks: Fully automated operation; IP Multimedia Sub-System (IMS) is hot issue for future research

• Application layer data transmission rate– Core Network: excessivly high up to Tera bit/s – Wired local loop: ISDN (128 kbit/s) ->xDSL (6-20->1000 Mbit/s)– Wireless (WLAN) 5 Mbit/s -> 25 - 1000 Mbit/s)– Mobile communication: increasing from ISDN to 100 Mbit/s data rate

• Increase in # of air interfaces competing -> multimode operation• Multi-homing: Use of multiple networks/services at same time• Radio resource control for wireless access networks: challenging• Resource Re-use Partitioning (interference avoidance) • Internet Protocol IPv6/8 to be understood/developed • Quality of Service responsive network design: challenge• Security & Privacy in comms. nets needs efficient solutions• Multi operator network co-operation is unsolved • Low cost mobile Internet access will need another decade to come• Next wireless/mobile generation will not be the final one


Status and future research funds in ComNets‘ working Status and future research funds in ComNets‘ working domain: Network Design and Evaluation Researchdomain: Network Design and Evaluation Research

• „Broadband for All“ is a main goal in Europe: The research funding in large scale will continue over the next decade

• Networks and Protocols Research – large amount of unsolved problems– ComNets (& MobNets) don‘t have severe academic competition in EU – Is key for the development & operation of distributed systems like power

plant, automated factory, networked IT centre, process control plant, Airbus, in car/in home infrastructure, etc.

• Current position of ComNets– EU funds expenditures ranking in 2004: RWTH=ComNets is rank 4 for all

broadband disciplines, including PHY– Exceptionally strong BMBF funding: cooperation with many companies– About 1.450 ComNets research publications downloads per month

• 3rd parties‘ funds appear available for at least another decade• ComNets students‘ profile perfectly fits the markets‘ needs• AWRC and UMIC cluster will need ComNets‘ current expertise• ComNets during the last 15 years had an average per year of

– 7 peer reviewed journal articles, 35 peer reviewed conference papers – 43 Diploma theses– 3.4 Ph.D. theses

1 monography, 2 course books, all published by J. Wiley&Sons 2000+


Download Statistics from 1/2001 to 10/2005Download Statistics from 1/2001 to 10/2005

0

2000

4000

6000

8000

10000

12000

14000

Total number of documents downloaded: 84.681 Average of 1435 downloads/month


3072 154 163 185 206 221 224 227 228 253 255 325 340 379 382 384 443 509 664 678 750 792 803

1380

6626 18083US: United States (47,93%)

DE: Germany (17,56%)

CN: China (3,66%)

JP: Japan (2,13%)

GB: United Kingdom (2,1%)

KR: Korea (1,99%)

IN: India (1,8%)

FR: France (1,76%)

IT: Italy (1,35%)

TW: Taiwan (1,17%)

CA: Canada (1,02%)

AU: Australia (1,01%)

ES: Spain (1%)

NL: Netherlands (0,9%)

SE: Sweden (0,86%)

GR: Greece (0,68%)

FI: Finland (0,67%)

AT: Austria (0,6%)

BR: Brazil (0,6%)

ID: Indonesia (0,59%)

CH: Switzerland (0,59%)

PL: Poland (0,55%)

RU: Russian Federation (0,49%)

UA: Ukraine (0,43%)

MY: Malaysia (0,41%)

others (6,66%)

Paper Downloads by Country in 2005Paper Downloads by Country in 2005


Einordnung ComNets/MobNetsEinordnung ComNets/MobNets

Prozessoren, Chips, Bauelemente,Platinen, USB Stick,usw.

PC, Server,Mobile Phone,Funkstrecke,Lokales Netz,Motor/Umformer,Transformator,Elektrofahrzeug

Internet, Mobilfunknetz,Glasfasernetz, Hochregallager,Automatisierte Fertigung,Fabrik, Airbus, Transrapid

ComNets/MobNets Research/Teaching

SMS, MMS,Google/YahooNavigation,Steuerungssoftwarefür komplexe Platt-formen und Systeme,Middleware & embeddedSoftware

Technologien und Plattformen: Produktion/Entwicklung wandern tendenziell in Weltregionen mit geringen Lohnstückkosten aus.

Schaltungstechnik


Multimedia Internet Service PlatformMultimedia Internet Service Platform

Zugriff

Zugriff Zugriff

Servers

Anwender

Vermittlungsnetz

Kommunikation Steuerung

Inhalt Inhalt


Layered structure:

Link level…focussing the radio transmission

System level…focussing the entire network behaviour

Protocol level…focussing radio network protocols

22337

6655

441

readingreadingpacket call

ComNets ComNets Simulation ConceptsSimulation Concepts


Current Work at ComNets I

Link Level Simulator of the OFDM transmission chain

– SystemC based including C++ code– Detailled implementation of

transmitter and receiver, including scrambler, RS/CC codec, interleaving, Modulation etc.

– Channel: AWGN, SUI-1 und SUI-5– IEEE 802.16a conformant

Result: Channel model: Bit error rate = f(C/(N+I))

SourceSource

ScrambleScramble

RS EncodeRS Encode

Conv. EncodeConv. Encode

PuncturePuncture

ByteIntleaveByteIntleave

Bit-IntleaveBit-Intleave

ModulateModulate

SinkSink

DescrambleDescramble

RS DecodeRS Decode

Conv. DecodeConv. Decode

DepunctureDepuncture

ByteDeintleaveByteDeintleave

Bit-DeintleaveBit-Deintleave

DemodulateDemodulate

EqualEqual

Pilot InsertPilot Insert

Preamble Preamble InsertInsert

IFFTIFFT

CP AppendCP Append

ChannelChannel

Pilot ExracttPilot Exractt

FFTFFT

CP RemoveCP Remove

Preamble ExtractPreamble Extract

Ch

an

nel

C

ha

nn

el

Est

ima

teE

stim

ate

*SUI=Standford University Interim (for outdoors morpho)


System Level SimulationsSystem Level Simulations

• Stochastic event driven simulation for traffic performance evaluation of mobile radio networks based on implementation of– Radio network protocols (simplified)– Radio resource management strategies– Multi-cellular radio propagation environment– Multi-network / multi-system coexistence– Time-variant traffic and actual interference

characteristics– Input from link-level simulation

ComNets’ expertise in entire network evaluation ComNets tools are being used to drive standardisation of

current and future wireless/mobile systems


Protocol Level Simulation: ParametersProtocol Level Simulation: Parameters

• Radio access mode– Duplex mode (FDD, TDD)– Carrier frequencies (FDMA)– Bandwidth– Radio frame– Time slot structure (TDMA)– Spreading (CDMA)

• Radio resource management– Thresholds– Timer– Target values

• Scenario description

• Services– Type (voice, web, video)– Characteristics– Switching (circuit,

packet)– Priority– Associated bearer

service

• Evaluation– Value ranges– Resolution

• Station data– Position, mobility– Power range


Leistungsbewertung: SimulationsumgebungLeistungsbewertung: Simulationsumgebung

Um

statistische Auswertung

Momente

LRE PDF/CDF Histogramm

Mobilstation

Protokollmodell

Modellstochastisches

Basisstation

SMTP FTPStatus AVLSprache HTTP MM

Lastgenerator

UDPTCPIP

WTPWAP

Instanz

KanalmodellBU TU RA HT

LLC

MAC

PHY

LLC

MAC

PHY


Netz-ArchitekturNetz-Architektur für GSM und den für GSM und den General Packet Radio Service (GPRS)General Packet Radio Service (GPRS)


GSM/GPRS Protokoll-StapelGSM/GPRS Protokoll-Stapel

PHY

MAC

RLC

LLC

SNDCP

IP

UDP/TCP

Anwendung

PHY

MAC

RLC

L1bis

NetzDienste

BSSGP

L1bis

NetzDienste

BSSGP

L1

L2

IP

LLC

SNDCP

UDP/TCP

GTP

L1

L2

IP

UDP/TCP

GTP

IP IP

L2

L1

IP

L2

L1

UDP/TCP

Anwendung

MS BSS SGSN GGSN HostUm Gb Gn Gi

detailgetreumodelliert

vereinfachtesModell


Wartenetz Modell und Anwendung zur Modellierung Wartenetz Modell und Anwendung zur Modellierung eines Teilnehmer-Rechensystemseines Teilnehmer-Rechensystems

Stationi

Stationj

xni

nj

i

q ij x jK StationenN Aufträge

q q q q0i 0j i0 j0

EIN AUSgeschlossen

Auftragsverkehrslast Abgangsrate X0

Zentrales Teilnetz

K Stationen N Aufträge (O < N < M)

TerminalTeilnetz

M TerminalsZ Denkzeit

x0

EIN AUSEINAUS

qij = Übergangsraten Matrix

- stationsspezifische Bedienstrategie

- Wartepuffer mit Prioritäten

- Ergebnisse: P(Nj = nj); Wartezeitverteilung Stationsauslastung Durchsatz pro Auftragsklasse usw.


Modellierung: Quelle für Sprachpakete über GPRSModellierung: Quelle für Sprachpakete über GPRS

Aktive Phase Inaktive Phase

t

Paketgröße

24 byte

SID Paket

4 byte

Übertragungskanal

FCFS

RR

λ1

λi

λN

Modell der Verkehrsquelle

Wartemodell mit stochastischen Ankunfts- und Bedienprozessen und Bedienstrategie


Zustands-Übergangsdiagramm einer Markov Kette Zustand= aktive Sprachquellen N(t)=i, Pufferbelegung

222

21

220

12

11

10

202

01

2inaktivbleibt Quelle00

2

2

aktivbleibtQuelle

inaktivwirdQuelleaktivbleibtQuelle

inaktivwirdQuelle

aktivwirdQuelleaktivbleibtQuelle

aktivwirdQuelleinaktivwirdQuelleinaktivbleibtQuelleaktivbleibtQuelle

inaktivbleibtQuelleinaktivwirdQuelle

aktivwirdQuelle

aktivwirdQuelleinaktivbleibtQuelle

pp

ppp

pp

ppp

ppppp

ppp

pp

ppp

pp

Übergänge aus den Zuständen N(t) = i. Aus N(t) sind (nach je 60 ms Übergänge zu N(t+1) = i+1, N(t+1) = i-1 und N(t+1) = i-3 möglich entsprechend den Übergangswahrscheinlichkeiten:

0,i

1,i

2,i

0,i-1

1,i-1

2,i-1

0,i-2

1,i-2

2,i-2

0,i-3

1,i-3

2,i-3

0,i+1

1,i+1

2,i+1

0,0

1,0

2,0

0,NQ

1,NQ

2,NQ

p00

p01

p02

p10

p11

p12

p22

p21

p21

Zustand: i,j

i = aktive Sprachquellen

j = Pakete im Puffer


Mathematische Verkehrsleistungs-Analyse für GPRS SpracheMathematische Verkehrsleistungs-Analyse für GPRS Sprache

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

0 20 40 60 80 100

Warteschlangenlänge (x)

P(X

>x)

Simulated

Calculated

500ms1 s

1.5 s

2 s

Komplementäre Verteilungsfunktion der Warteschlangenlänge für verschiedene mittlere Sprach-Phasenlängen (mittlere Sprachpausenlänge =1 s)

95-Perzentil der Wartezeit von Sprachpaketen bei 10 Sprachquellen

0

100

200

300

400

500

600

700

800

0.5 0.7 0.9 1.1 1.3 1.5 1.7 1.9

Mittelwert der Sprachpausenlänge [s]

95-P

erze

nti

l d

er W

arte

zeit

Calculated

Simulated

500 ms

750 ms

1000 ms

1500 ms


UMTS (2000): System Throughput & BER UMTS (2000): System Throughput & BER

Maximum System Throughput for WWW traffic reached with 64 kbit/s DTCH

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

0 0.2 0.4 0.6 0.8 1Block Error Probability

No. of Mobile Stations = 10No. of Mobile Stations = 30No. of Mobile Stations = 60No. of Mobile Stations = 100No. of Mobile Stations = 150No. of Mobile Stations = 200No. of Mobile Stations = 250

Block Error Rate at 256 kbit/s

the BLER increases with increased Number of Stations, reducing throuphput accordingly.


Cell Capacity over Distance Cell Capacity over Distance is Inverse to the Needsis Inverse to the Needs

Range limitation of broadband APS by– high attenuation at high frequencies – limited transmission power (EIRP)– Unfavourable radio propagation conditions, e.g.,

in urban areas Increased # of BS needed with increased carrier

frequency to cover a given area High CAPEX and OPEX High cost/bit transmitted

High capacity available close to AP only. Under constant user density: Number of users increases with d Cell capacity offered per area element differs from

capacity requested by users Future trend makes it more worse

New Deployment Concepts required tobring broadband to wider area than possible with one base station in current systems Reduce the cost/bit transmitted by 2 to 3 orders of magnitude

Sources: B. Walke, H. Wijaya, D.C. Schultz: The Application of Relays in Infrastructure-based Future Mobile Radio Network Deployment ConceptsSubmitted: VTC 2006 Spring, Melbourne, AustraliaT. Irnich, D.C. Schultz, R. Pabst, P. Wienert: Capacity of a Relaying Infrastructure for Broadband Radio Coverage of Urban Areas. Proceedings of the 10th WWRF meeting, New York, 10/2003

Actual Available Capacity vs. Requested Capacity

Available

2005

Cell border

Cap

city

/A

rea

Ele

men

t

2010

Location of Base Station

Distance d

Requested by users

Users

at d

istan

ce d


Relay Enhanced Cells (REC)Relay Enhanced Cells (REC)Using Fixed Relay Stations (FRS)Using Fixed Relay Stations (FRS)

Pros:• Relays in REC

– don’t need a wired backbone access (lowers CAPEX and OPEX)

– Full flexibility of relays (re-)positioning • Relays introduced to a cell can

– enlarge the coverage area bbbbb(using antenna gain)

– Increase capacity at cell border– balance the capacity/area element– reduce transmission power

• increasing public acceptance• Reducing co-channel interference

• (Movable) Relays support – fast network rollout, – outdoor to indoor service– Exploitation of macro-diversity

(co-operative relaying)

Cons:• In band relays consume radio resources • Out of band relays need multiple transceivers • Relays introduce extra delay

AP

FRS

Source: Walke, Bernhard; Wijaya, Harianto, Schultz, Daniel C.: The Application of Relays in Infrastructure-based Future Mobile Radio Network Deployment Concepts. Submitted: VTC 2006 Spring, Melbourne, Australia


1. 2. 3.

Line of Sight

AP

Cellular Multi-hop deployment in highly shadowed environmentCellular Multi-hop deployment in highly shadowed environment

Channel Group 1Channel Group 2

Source: ComNets 2003


Capacity at Relay (FRS) with Antenna GainCapacity at Relay (FRS) with Antenna Gain

• All AP capacity “transferred” to one FRS sub-cell• Capacity of FRS rises with antenna gain until highest PHY mode can be applied • Cost of relaying: 6.67 Mbit/s of AP capacity at 30 dBi gain (example: IEEE 802.11a PHY using a WiMax like

MAC protocol)

FRS 1

FRS 2

FRS 3

FRS 4 AP

0 5 10 15 20 25 30 35 400

5

10

15

20

25

FRS receive antenna gain (dBi)

cap

acit

y (

Mb

it/s

)

FRS sub-cell

AP sub-cell

6.67 Mbit/s

P. Gupta and P. R. Kumar: The capacity of wireless networks. IEEE Transactions on Information Theory, 46(2):388 - 404, 2000: Multi-hop reduces capacity.

Pabst, Ralf; Esseling, Norbert; Walke, Bernhard: Fixed Relays for Next Generation Wireless Systems - System Concept and Performance Evaluation. Journal of Communications and Networks, Vol.7, No. 2, p.p. 104-114, Korea, 06/2005: Spectrum capacity can be increased by multi-hop, if mesh hops are narrow beam based.


Source: Walke, Bernhard; Pabst, Ralf; Schultz, Daniel C.: A Mobile Broadband System

based on Fixed Wireless Routers. Proc. ICCT 2003 Intern. Conf. Comm. Techn., 04/2003

ComNets Vision of a Mobile Low Cost Internet Access: ComNets Vision of a Mobile Low Cost Internet Access: Relay-based Cellular Wireless Mobile Broadband SystemRelay-based Cellular Wireless Mobile Broadband System

Access Point

1. Hop Relay

2. Hop Relay

Relay Enhanced Cell


f 1 f 3 f 7 f 9

f 4f 6 f 2

f 7 f 9

f 1

f 1

f 11

f 3

f 2

f 0f 10

f 10

f 8f 6

f 4

f 8

f 0

f 5

f 9

R

R

h h

f 5

f 5

f 5

f 5

f 5

i

i

j

i

j

i

i

j

j

i

j

j D1

D2

D3

D4

D5

D6

Reuse shift parameter for a Reuse shift parameter for a N N = 12 Relay-Cell cluster= 12 Relay-Cell clusterand Cell Radius Rand Cell Radius R

FMTAP Antenna


Single-Hop and Relay Enhanced Cell Throughput Single-Hop and Relay Enhanced Cell Throughput compared (3 FRS)compared (3 FRS)

Entfernung AP und M

T (x) [m

]

En

d-z

u-E

nd

e-D

urc

hsa

tz [

Mb

it/s]

Entfernung AP und MT (y) [m]

600400

2000

-200-400

-600

10

20

30

40

600400

2000

-200-400

-600

End

e-zu

-End

e-D

urch

satz

[Mbi

t/s]

-400

-200

0

200

400

-400

-200

0

200

400

0

5

10

15

20

25

30

AP

FMT

FMT

-400

-200

0

200

400

x

-400-2000200400y

2428

30

20161412

2Mbit/s

4Mbit/s

6Mbit/s

10Mbit/s

12Mbit/s

8Mbit/s

200m central cell

346m single hop cell

Area=

200m

200m

Iso-throughput curvesIso-throughput curves

Esseling, Norbert: Ein Relaiskonzept für das hochbitratige drahtlose lokale Netz HIPERLAN/2, ABMT 42, 1. Auflage Jul/2004, 307 Seiten, ISBN: 3-86130-169-5www.comnets.rwth-aachen.de/Dissertati.178.0.html


End-to-End Throughput DownlinkEnd-to-End Throughput Downlinkalong y-Axisalong y-Axis

y

3er, 3FMTs12er (346m)7er (346m)3er (346m)3er (200m)

0

5

10

15

20

25

30

35

40

12er (346m), kein FMT

45

0 50 100 150 200 250 300 350 400

12er

7er3er

1.Hop

1.Hop

1.Hop

2.Hop2.Hop 2.Hop

End

-to-

End

Thr

ough

put [

Mbi

t/s]

Distance [m]

3er, Relais-Sys.>1-Hop-Sys.



7er (346m), kein FMT3er (346m), kein FMT

12er, 3FMTs7er, 3FMTs


Single-Hop and Relay Enhanced Cell Throughput Single-Hop and Relay Enhanced Cell Throughput compared (3 FRS)compared (3 FRS)

Entfernung AP und M

T (x) [m

]

En

d-z

u-E

nd

e-D

urc

hsa

tz [

Mb

it/s]

Entfernung AP und MT (y) [m]

600400

2000

-200-400

-600

10

20

30

40

600400

2000

-200-400

-600

End

e-zu

-End

e-D

urch

satz

[Mbi

t/s]

-400

-200

0

200

400

-400

-200

0

200

400

0

5

10

15

20

25

30

AP

FMT

FMT

-400

-200

0

200

400

x

-400-2000200400y

2428

30

20161412

2Mbit/s

4Mbit/s

6Mbit/s

10Mbit/s

12Mbit/s

8Mbit/s

200m central cell

346m single hop cell

Area=

200m

200m

Iso-throughput curvesIso-throughput curves

Esseling, Norbert: Ein Relaiskonzept für das hoch bitratige drahtlose lokale Netz HIPERLAN/2, ABMT 42, 1. Auflage Jul/2004, 307 Seiten, ISBN: 3-86130-169-5www.comnets.rwth-aachen.de/Dissertati.178.0.html

At 11,8 dbi


Multi-hop Relay TechnologiesMulti-hop Relay TechnologiesR.R. Pabst, B. Walke, D.C. Schultz:Pabst, B. Walke, D.C. Schultz:Relay-Based Deployment Concepts for Wireless and Mobile Broadband Relay-Based Deployment Concepts for Wireless and Mobile Broadband

RadioRadio. In . In IEEE Communications MagazineIEEE Communications Magazine, p.p. 80-89, New York, US, 09/2004, p.p. 80-89, New York, US, 09/2004

Time domain relay(FWR)

Frequency domain relay

Frequency domain relaywith pure forwarding


Forwarding Concept: Case 2Forwarding Concept: Case 2

MTs served by APMTs served by AP

MTs served by FRS#4MTs served by FRS#4


FRS#4served by AP

FRS#4served by AP

FRS#3served by AP

FRS#3served by AP



Time

TMP-MTTAP-FRS

FRS#2served by AP

FRS#2served by AP

FRS#1served by AP

FRS#1served by AP

FRS 1

FRS 2

FRS 3

FRS 4 AP

TMP-MT

• One carrier frequency

• Exploitation of environment

2 Groups of FRSs that can serve their MTs in parallel

• One carrier frequency

• Exploitation of environment

2 Groups of FRSs that can serve their MTs in parallel


Coordination Across BSCoordination Across BSResource PartitioningResource Partitioning

MTs served by AP CMTs served by AP C

MTs served by FRS#C4MTs served by FRS#C4


FRS#C4served by AP

FRS#C4served by AP

FRS#C3served by AP

FRS#C3served by AP



Time

TMP-MTTAP-FRS

FRS#C2served by AP

FRS#C2served by AP

FRS#C1served by AP

FRS#C1served by AP

TMP-MT

MTs served by AP BMTs served by AP B

MTs served by FRS#B4MTs served by FRS#B4


FRS#B4served by AP

FRS#B4served by AP

FRS#B3served by AP

FRS#B3served by AP



FRS#B2served by AP

FRS#B2served by AP

FRS#B1served by AP

FRS#B1served by AP

MTs served by AP AMTs served by AP A

MTs served by FRS#A4MTs served by FRS#A4


FRS#A4served by AP

FRS#A4served by AP

FRS#A3served by AP

FRS#A3served by AP



FRS#A2served by AP

FRS#A2served by AP

FRS#A1served by AP

FRS#A1served by APC

ell T

ype

AC

ell T

ype B

Cel

l Type

C

Time Slot to Feed FRSs Time Slot X Time Slot Y Time Slot Z


Coordination Across BSsCoordination Across BSs

• Only one Carrier Freq. Required to cover the scenario

• Distance between “co-channel” sub-cells: 460 m

• Only one Carrier Freq. Required to cover the scenario

• Distance between “co-channel” sub-cells: 460 m

FRS 1

FRS 2

FRS 3

AP

FRS 1

FRS 2

FRS 3

FRS 4 AP

FRS 1

FRS 2

FRS 3

FRS 4 AP

FRS 2

FRS 3

FRS 4 AP

FRS 1

FRS 2

FRS 3

FRS 4 AP

FRS 1

FRS 3

FRS 4 AP

FRS 4

FRS 3

FRS 1

FRS 2FRS 4 AP

FRS 2

FRS 1

Cell Type A

Cell Type B

Cell Type C


Mesh Network applied to IEEE 802.11 WLAN:Mesh Network applied to IEEE 802.11 WLAN:ComNets Proposal ComNets Proposal

• Works under IEEE 802.11 PCF mode• MPs operate as PC (point coordinator)• Beacons with the format of IEEE 802.11’s from the PC inform nodes of the

CFP (contention free period) and CP (contention period)• MN works during CFP, IEEE 802.11 on CP

MNIEEE802.11eMN

CFP CP

IEEE802.11eMN

CFP CP

• Coexistence of MN with IEEE 802.11e

Beacon Guard timeNote:

The guard times are fixedSource: Zhao, Rui; Walke, Bernhard; Hiertz, Guido: W-CHAMB (Wireless CHannel Oriented Ad-hoc Multi-hop Broadband): A new MAC for better support of Mesh networks with QoS, Contribution to IEEE 802.11 WLAN Working Group Session, September 2004, p. 5, Berlin, Federal Republic of Germany, 09/2004 ComNets 2004And: Wijaya, Harianto: Broadband Multi-Hop Communication in Homogeneous and Heterogeneous Wireless Lan NetworksABMT 46, 1. Auflage Feb/2005, 237 Seiten, ISBN: 3-86130-175-X, available at: www.comnets.rwth-aachen.de/Dissertati.178.0.html


Mesh Network (MN) and IEEE 802.16 combined.Mesh Network (MN) and IEEE 802.16 combined.ComNets proposes dedicated mesh network protocolComNets proposes dedicated mesh network protocol

• Provides meshing of APs and Relays and MS access in the same channel within a Relay Enhanced Cell (REC)

• Base Station/Relay Node are called MeshPoint (MP) • MN connects MPs in RECs and MPs of adjacent RECs using MAC-frame periodic slots • IEEE 802.16 MAC frame serves MSs on first hop to MP

MNIEEE802.16MN

Periodic Frame specific

IEEE802.16MN

Periodic Frame specific

Coexistence of MN with IEEE 802.16

Beacon Guard timeNote:

The guard times are fixedSource: Mangold, S.; Habetha, J.; Choi, S.; Ngo, C.: Coexistence and interworking of IEEE 802.11a and ETSI BRAN

HiperLAN/2 in multi-hop scenarios. In 3rd IEEE Workshop Wireless Local Area Networks, Boston, 09/2001


Possible IEEE 802.16 WiMAX Mesh SolutionPossible IEEE 802.16 WiMAX Mesh Solution

• BSs connected by MN on separate frequency channel• IEEE 802.16 between BS and SSs or RNs (one-hop forwarding possible)


Coexisting WLANs: The Game ModelCoexisting WLANs: The Game Model

player

application

WLAN

supports

required QoSparameters

1..*

has a

QoS parameters

Delta (delay)Theta (throughput) Xi (jitter)

demanded QoSparameters

action

defines

takes into accountstrategy

behavior

determinesmutual

interferencewithin SSG

dependson

observed QoSparameters

leads to

payoff(outcome)

define

defineutility

determines

discountingfactor

long-term maximization

• Overlapping WLANs are represented by a player

• Each player has a strategy to determine what action to select

• An action specifies a behavior

• The players optimize the payoff (i.e. outcome) of the game


WLAN Spectrum Coexistence Scenario: WLAN Spectrum Coexistence Scenario: Two 802.11e QBSSs sharing one ChannelTwo 802.11e QBSSs sharing one Channel

• Basic Service Sets are modeled as players that attempt to optimize their outcomes

• The coexistence problem is modeled as a repeated, stage-based game

HCF 1Player 1

HCF 2Player 2

QSTAEDCF

Player 3represents offeredEDCF traffic within

the OQBSS

signals communicationand control

QSTA

QSTA

QSTA

QSTA

detection rangesof HCFs

QSTA: Quality Station

HCF: Hybrid Coordinator Function


Nash EquilibriumNash Equilibrium

payoff of player i: Vi(ai,a-i)

payoff of player -i: Vi(a-i,ai)

0.5

1

0.5

(C|C)

(D|C)

(D|D)

(C|D)

gain through deviation

punishment through opponent defection

1

player -i cooperatesplayer -i defects

payoff under coop-

eration

1.

2.Nash

Equilibrium

3.

0

player i’s classifications of opponent’s behavior

1.

Definition:

“No player can gain a higher

payoff in deviating from Nash

Equilibrium”

stable and thus predictable point of

interaction

player i defecting

player -i defecting

stable point of interaction


Strategies in Multi Stage Games (I)Strategies in Multi Stage Games (I)

• Strategies describe the alternatives a player has for an action within a Multi Stage Game

• Consideration of interaction with decisions of influenced players

• Strategies modeled as state machines

all outcomes except (C)

C

(C)

D

9090 9090

(*)

n=1

(2) GRIM

C

(C)

(D)D

90

9090

90

(D)9090

(C)

n=1

• Example: Dynamic trigger strategy TitForTat (TFT) – the player cooperates if the opponent cooperates and vice versa

C

(*)9090

n=1

(1) COOP (3) RANDOM

C

50%

50%D

90

9090

90

50%9090

50%

n=1

(4) TFT


Strategies in Multi Stage Games (II)Strategies in Multi Stage Games (II)

TFT versus various strategies

• Multi Stage Games of multiple strategies, evaluated in terms of observed throughput (Θ) and (TXOP) delay

• TFT: Player’s behavior follows the opponent’s leading to predictable MSG outcomes QoS guarantee

• RANDOM: frequent fluctuation in behavior implies instable game course unsatisfying QoS degradation

RANDOM versus various strategies


ComNets Concept for a Flexible Protocol StackComNets Concept for a Flexible Protocol Stack

• Protocols share a lot of commonalities, that can be exploited in an efficient multi-mode capable wireless system

Generic Protocol Stack as “toolbox of parameterizable protocol functions”

• Generic part: Tradeoff of general usability vs. implementation effort

mode 1 mode x

system specific protocol stack

same property

(classic) single protocol stack

generic protocol stack

specific parts

common functionalityspecific functionality

mode 2protocol functions

protocol architecture

data structuresdifferent

protocol framework

protocol management

reconfiguration management/functions

multi-mode (composite) protocol stack

reconfigurable protocol stack

modes convergence management/functions


WINNER Multi-Mode Protocol WINNER Multi-Mode Protocol ArchitectureArchitecture

(2) management that is specifically optimized for the mode1 and mode2 in use probably more efficient

RRC-r2

ge

ne

ric

sta

ck

ma

na

ge

me

nt

MAC-r2

PHY-s2 PHY-r2PHY-s2

PHY-gPHY-g

PHY-r1

PHY-gMAC-g

MAC-r1

RLC-g

control-plane

RRC-r1

RRC-g

join

t m

od

e 1

/2 st

ack

m

an

age

me

nt

alternatives

control usermanagement

modes-switching

and coexistence

configuration and information transfer

reco

nfig

ura

tion

ma

na

ge

me

nt

Stack Management alternatively: (1) generic management more flexible


Reference Structure of Layer or SublayerReference Structure of Layer or Sublayer

L(N) - mode 1, 2 or 3

L(N) - SAP-s1.1

L(N-1) -SAP-s1.1

L(N) - SAP-s1.2

services of layer (N) / sublayer (N.n)

provided jointly through generic and

specific parts

(N)-MCM

(re)configures

composes L(N) – mode 1 – specific part

L(N) – mode 2 – specific part

L(N) – generic partfor mode 1-2-3convergence


L(N) - SAP-s2.1

L(N) - SAP-s2.2

L(N) - SAP-s3.1

L(N-1) - SAP-s2.1

L(N-1) -SAP-s3.1

stack management

L(N) - SAP-g

L(N) - mode 1, 2 or 3

L(N) - SAP-s1.1

L(N-1) -SAP-s1.1

L(N) - SAP-s1.2

services of layer (N) / sublayer (N.n)

provided jointly through generic and

specific parts

(N)-MCM

(re)configures

composes L(N) – mode 1 – specific part


L(N) – generic partfor mode 1-2-3convergence


L(N) - SAP-s2.1

L(N) - SAP-s2.2

L(N) - SAP-s3.1

L(N-1) - SAP-s2.1

L(N-1) -SAP-s3.1

stack management

L(N) - SAP-g

(N) Layer Modes Convergence Manager ((N)-MCM):• Facilitates the structuring of an arbitrary layer into generic and

specific parts• Responsible for composition and (re-)configuration• Controlled by the stack management Optimization potential is marked up in questioning the necessity of

indicated differences


Realization of the Flexible Protocol StackRealization of the Flexible Protocol Stack

• Functionality of the Layers is composed from a toolbox of functional units• Mode-specificness can either be specific modules or specific configuration /

parameterization of the stack, individual layers or even functional units• Reference Implementation for WINNER Layer 2 currently performed at

ComNets

PHY

PHY

RLC

MAC

TCP/UDP / IP

PHY

Channel (Modem)

Stack Config.

data

data

Functional Modul

Functional Modul

Functional Unit

Functional Unit

Functional Unit

Mode Specific

(Sub-)layer Interface

(N)-Layer Configuration

Service Access Point

Service Access Point

Data

Composition, Parameterization and Data Query

modespecific

generic

Interface

RRC


Spectrum Requirement Estimation at a GlanceSpectrum Requirement Estimation at a Glance

Market info Calculation algorithm Radio technology info

Future services

Offered traffic

Required Quality of Service (QoS)

Scenarios definition

Traffic distributionto Radio Access

Techniques (RAT) & Radio Environments

Capacity dimensioning

Adjustments & weighting

Capabilities

Availability/ Coverage

Technical spectrum requirements


General Approach for Capacity CalculationGeneral Approach for Capacity Calculation

• In packet based systems QoS constraints require certain amount of free capacity

System Load

System Throughput Mean Delay

100%

PhysicalLayerThroughput

Delay Target

Usable fraction of system capacity

MACLayerThroughput

RLCLayerThroughput

overloadunderload

Tmax = Crlc System Load


βN, βN(2)

β1, β1(2)

Packet-switched Capacity CalculationPacket-switched Capacity Calculation

• Required system capacity calculated from M/G/1/FCFS/NONPRE queue (“head of the line priority queue“)

• Throughput requirements per SC derived under steady state operation • To meet the delay requirement of a Traffic Class needs proper

dimensioning of capacity C

λ1

λ2

λN

C

Priority 1

Priority 2

Priority N

Server

Parameters of the model:

• λi : arrival rate of packets with priority i

• βi(i) : i-th moment of service

duration of packets with priority I

• C: capacity searched for

Highest priority

Lowest priority

β2, β2(2)


Aggregate Spectrum RequirementAggregate Spectrum Requirement

• Results shown above do not include last step of new methodology (i.e., accounting for multiple operators, guard bands, FSU, etc.)

• Some parameters for PS capacity calculation have been reasonably chosen, other choice would have led to different results

• Small difference resulting is more or less coincidence, since a number of effects partly compensate each other

• The scenario considered is not a likely scenario to be looked at in spectrum requirement calculation in preparation for WRC-07

• Comparison shows that results are in line with earlier results• New methodology’s concepts and algorithms represent state of the art

Required spectrum [MHz] Relative change [%]

Service environment

UL DL UL+DL UL DL UL+DL

SE1 18.332 21.742 40.074 40.91 -21.37 -1.44

SE2 130.754 257.224 387.978 12.85 -0.37 3.72

SE3 9.332 9.772 19.304 73.79 -23.55 4.81

Sum 154.418 288.738 447.356 18.00 -3.33 3.28


Communication Networks (Walke): Communication Networks (Walke): “We do mostly layers 2..4”“We do mostly layers 2..4”

Transport Services & Protocols4

Radio Resource & Mobility 3 Control

Location Based Services3

Medium Access & Link Control 2 Protocols

Smart Antenna Protocol 1-2 Support B

road

ban

d W

irel

ess

Tra

nsp

ort

Pla

tfo

rms

Mesh

Netw

ks.

& R

ela

yin

g f

or

cellu

lar

Fix

ed

an

d M

ob

ile N

etw

ork

s

Con

verg

en

ce

IEEE 8

02.1

1/1

5/1

6/2

1

Sta

nd

ard

izati

on

Wir

ele

ss N

etw

ork

s &

In

terw

ork

ing

Sp

ectr

um

Co

-exi

sten

ce

Res

earc

h

Tra

ffic

Per

form

ance

E

valu

atio

n

(Th

eory

of

Lar

ge

Sys

tem

s)

Ad

apti

ve P

roto

col

Sta

ck

So

ftw

are


ComNets ProfileComNets Profilehttp://http://www.comnets.rwth-aachen.dewww.comnets.rwth-aachen.de// (Oct. 2005)(Oct. 2005)

ComNets research focuses on OSI-layers 2, 3 and 4. We work also in radio spectrum co-existence & design of adaptive protocol stack solutions for multi-radios. Some of our people are in domains of the-spectrum & regulation, - cognitive radios, - SW-defined re-configurable radios.

Research is both strongly theoretical and experimental. Experimentation capabilities at ComNets cover the „down to bit level“ prototype like implementation of radio access networks based on software based tools and include the possibility to implement protocol stacks and new algorithms for test. Design and Optimisation of Disruptive Deployment Concepts for Future Cellular Radio is our strongest key research area, this extends to wireless mesh networks. We have contributed to standards like GSM/GPRS, ETSI HiperLAN2, IEEE 802.11e,k,s, IEEE 802.15.3, IEEE 802.16 CEN TC 278 DSRC, ITU-R WP8F spectrum estimation methodology. Our theoretical basic research, especially in game theory applied to radio systems‘ co-existience in frequency spectrum will hopefully permit better exploitation of spectrum. We are leading in mesh networking protocols for wireless systems.Our wireless broadband multi-hop ad-hoc communication network design does not have any severe competition. We are able to evaluate really large communication systems based on the unique tools that we have developed.


Comparison MobNets (left) and ComNets (right)Comparison MobNets (left) and ComNets (right)

Transport Protocobls4

Network Optimization & Theory3

Service Discovery3

Link Layer Protocols2

Cross Layer Issues &Low-Power1/2

Sen

sor

Net

wo

rks

and

A

pp

lica

tio

ns

Per

son

al A

rea

Net

wo

rks

Sel

f-C

on

fig

ura

tio

n (

ad

ho

c)W

irel

ess

LA

Ns

Ad

van

ced

Cel

lula

r N

etw

ork

sC

og

nit

ive

Rad

ios

and

N

etw

ork

s

Transport Services & Protocols4

Radio Resource & Mobility 3 Control

Location Based Services3

Medium Access & Link Control 2 Protocols

Smart Antenna Protocol 1-2 Support

Bro

adb

and

Wir

eles

s T

ran

spo

rt

Pla

tfo

rms

Mes

h N

etw

ork

s &

Rel

ayin

g f

or

Cel

lula

rIE

EE

802

.11/

15/1

6/21

S

tan

dar

diz

atio

nS

pec

tru

m C

o-e

xist

ence

R

esea

rch

Tra

ffic

Per

form

ance

E

valu

atio

n

Ad

apti

ve P

roto

col

Sta

ck

So

ftw

are


Courses, Labs and SeminarsCourses, Labs and Seminars

• Networking & Protocols Expertise is a must for Information Technology Engineers

• Both Curricula– Information & Communications (ET & IT)– Technical Computer Science (TI)Contain mandatory courses and courses to be selected

from catalogues on Networking & Protocols

• The load from Course Lecturing, Labs and Seminars is by far to big to be shouldered by one chair

• It is agreed that MobNets is not lecturing courses in basic studies


Curriculum lecture, Labs, Seminars Type Credits ResponsiblesET&IT u. TI Grundgebiete der Informatik 3 Mandat V2 Ü1 Walke u. wiss. Mitarb.ET & IT u. TI Praktikum Grundgeb. Informatik 1 MandatP4 Walke/Kraiss und wiMi

TI und ET & IT Komm.Netze u. Verkehrstheorie MandatV4 Ü2 Walke u. wiss. Mitarb. TI und ET & IT Praktikum Kommunikationsnetze Elective P4 Walke u. wiss. Mitarb.

ET & IT Praktikum Mobilfunknetze Elective P3/4 Walke u. wiss. Mitarb. ET & IT Seminar Kommunikationsnetze Elective V3 Walke u. wiss. Mitarb.

TI und ET & IT Stochastische Simulationstechnik CatalogV4 Ü2 Walke/jetzt LehrauftragTI und ET & IT Praktikum Stochast.Simulationstechnik Elective P3/4 Walke u. wiss. Mitarb. Techn. Inform. Einführung: Objektorient. Programmierg. Mandat P3 Walke/Gebhardt u.a.Techn. Inform. Projekt CORBA Elective V4 Walke u. wiss. Mitarb.

Courses, Labs and Seminars by ComNets Current (stationary) Status


Proposal for Call for ApplicationsProposal for Call for Applications

Kommunikationsnetze mit den Anwendungsgebieten: 1. Modellierungstechnik, Verkehrstheorie, Bedientheorie, stochastische

Simulationstechnik mit Anwendungen auf• Mobile Breitbandnetze (Mesh und Relay-Netze) • Netzoptimierung und Kooperation von Drahtlos- und Mobilfunknetzen

2. Spektrums-Koexistenz Forschung 3. Software Entwicklungsmethoden (UML) für Netze und multi-mode

Terminals


Wesentliche Ergebnisse in 2004Wesentliche Ergebnisse in 2004

• 38 Conference Papers, 5 Journal Papers, 10 IEEE Standardisation Contributions

• 6 Awards• Packet Relays accepted world-wide as disruptive technology:

- capacity enhancement for 2 & 3G systems- Range extension for wireless broadband systems

• Co-existence of radio research established • Mobile Web services demonstrated • Air Interface Multi-Mode operation through Modes

Convergence Manager • A number of contributions to IEEE 802 Project, namely:

- .11s Mesh- .15s Mesh- .11 and .15 multi-hop support- .16 spatial multiplexing


Ende

Documents

Communication Networks Stand und zukünftige attraktive Arbeitsgebiete für den Lehrstuhl für Kommunikationsnetze Prof. Dr.-Ing. Bernhard Walke Kommunikationsnetze,