Top Banner
1 Architecture and Protocol Design for Cognitive Radio Networks* Microsoft CR Summit, Jun 2008 Rutgers, The State University of New Jersey www.winlab.rutgers.edu Contact: Professor D. Raychaudhuri [email protected] *Collaborative project with Profs. Srini Seshan & Peter Steenkiste, CMU And Prof. Joe Evans, U Kansas
20

1 Architecture and Protocol Design for Cognitive Radio Networks* Microsoft CR Summit, Jun 2008 Rutgers, The State University of New Jersey .

Mar 27, 2015

Download

Documents

Ethan Cobb
Welcome message from author
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
Page 1: 1 Architecture and Protocol Design for Cognitive Radio Networks* Microsoft CR Summit, Jun 2008 Rutgers, The State University of New Jersey .

1

Architecture and Protocol Design for Cognitive Radio Networks*

Microsoft CR Summit, Jun 2008

Rutgers, The State University of New Jerseywww.winlab.rutgers.edu

Contact: Professor D. [email protected] *Collaborative project with

Profs. Srini Seshan & Peter Steenkiste, CMUAnd Prof. Joe Evans, U Kansas

Page 2: 1 Architecture and Protocol Design for Cognitive Radio Networks* Microsoft CR Summit, Jun 2008 Rutgers, The State University of New Jersey .

2

3adate:D:L:dxml_no_ddefault_fro1UTF-8UTF-8default_froSEARCH

Cognitive Radio: Problem ScopeSpectrumAllocation

Rules(static)

INTERNET

BTS

AuctionServer

(dynamic)

SpectrumCoordination

Server(dynamic)

AP

Ad-hocsensor cluster(low-power, high density)

Short-range infrastructuremode network

(e.g. WLAN)

Collaborative ad-hoc networksMAC/PHY adaptation

Wide-area infrastructuremode network (e.g. 802.16)

Dense deployment of wireless devices, both wide-area and short-range

Proliferation of multiple radio technologies, e.g. 802.11a,b,g, UWB, 802.16, 3G femto, 4G, ..

New cognitive radio devices with programmable PHY/MAC

Available options include: Agile radios (interference

avoidance) Dynamic centralized

allocation methods Distributed spectrum

coordination (etiquette) Collaborative ad-hoc

networks

Etiquette policy

SpectrumCoordination

protocols

Spectrum Coordinationprotocols

Dynamic frequencyprovisioning

Scope of CognitiveRadio Protocol Stack

Page 3: 1 Architecture and Protocol Design for Cognitive Radio Networks* Microsoft CR Summit, Jun 2008 Rutgers, The State University of New Jersey .

3

3adate:D:L:dxml_no_ddefault_fro1UTF-8UTF-8default_froSEARCH

Broad range of technology & related policy options for spectrum

Need to determine performance (e.g. bps/Hz or bps/sq-m/Hz) of different technologies taking into account economic factors such as static efficiency, dynamic efficiency & innovation premium

Hardware Complexity

Protocol Complexity(degree of

coordination)

ReactiveRate/Power

Control

ReactiveRate/Power

Control

AgileWideband

Radios

AgileWideband

Radios

Unlicensed Band

with DCA (e.g. 802.11x)

Unlicensed Band

with DCA (e.g. 802.11x)

InternetServer-based

SpectrumEtiquette

InternetServer-based

SpectrumEtiquette

Ad-hoc,Multi-hop

Collaboration

Ad-hoc,Multi-hop

Collaboration

Radio-levelSpectrumEtiquetteProtocol

Radio-levelSpectrumEtiquetteProtocol

StaticAssignment

StaticAssignment

InternetSpectrumLeasing

InternetSpectrumLeasing

“cognitive radio”schemes

UWB,Spread

Spectrum

UWB,Spread

Spectrum

“Open Access”+ smart radios

Unlicensed band +simple coord protocols

Cognitive Radio: Design Space

Needs protocol support unified framework

called “CogNet”

Page 4: 1 Architecture and Protocol Design for Cognitive Radio Networks* Microsoft CR Summit, Jun 2008 Rutgers, The State University of New Jersey .

4

3adate:D:L:dxml_no_ddefault_fro1UTF-8UTF-8default_froSEARCH

CogNet Protocol: Architectural Principles

Decentralized spectrum coordination as an integral part of protocol capabilities

“mutual observability” achieved via explicit exchange of spectrum information

Support for ad hoc network collaboration Beacons that enable network bootstrapping and discovery without infrastructure support

Adaptive selection of PHY, MAC, routing methods Control framework that enables on-the-fly selection of data path protocol components

Cross-layer control exchanged across protocol layers Access to cross-layer information necessary for cross-layer adaptation

Logical separation of control & data for flexible design and low overhead

Minimize contention between control & data (…>>50% overhead in 802.11 networks!)

Efficient integration with the wired Internet Aggregation of routing and cross-layer control information at boundary/gateway nodes

Page 5: 1 Architecture and Protocol Design for Cognitive Radio Networks* Microsoft CR Summit, Jun 2008 Rutgers, The State University of New Jersey .

5

3adate:D:L:dxml_no_ddefault_fro1UTF-8UTF-8default_froSEARCH

“CogNet” Protocol Stack

Global Control Plane (GCP) Common framework for spectrum allocation, PHY/MAC bootstrap, topology

discovery and cross-layer routing

Data plane Dynamically linked spectrum mgmt, PHY, MAC, Network modules and

parameters as specified by control plane protocol

Control PHY

Control MAC

SpectrumMgmt

- BootstrapDiscovery

PHY

MAC

Network

Transport

Application

Control Plane Data Plane

Global Control Plane

Data Plane

Control API

Data Path

Establishment

Naming&

Addressing

Page 6: 1 Architecture and Protocol Design for Cognitive Radio Networks* Microsoft CR Summit, Jun 2008 Rutgers, The State University of New Jersey .

6

3adate:D:L:dxml_no_ddefault_fro1UTF-8UTF-8default_froSEARCH

CogNet Protocol: Common Spectrum Coordination Channel (CSCC)

CSCC enables mutual observation between heterogeneous nodes to explicitly coordinate spectrum usage

CSCC function is an integral part of the CogNet global control plane (GCP)

• Exchange of CSCC messages by an extra narrow-band (low bit-rate) radio • Periodically broadcast spectrum usage parameters to neighbors• Enables distributed algorithms for spectrum co-existence

Page 7: 1 Architecture and Protocol Design for Cognitive Radio Networks* Microsoft CR Summit, Jun 2008 Rutgers, The State University of New Jersey .

7

3adate:D:L:dxml_no_ddefault_fro1UTF-8UTF-8default_froSEARCH

CogNet Protocol: Packet Format

MessageType

Flags SourceAddress

IE length IE(1) IE(n)

1B 1B 6B 2B variable variable

Generic GCP Packet: Ethernet packet format with control payload (consisting of variable length information elements)

. . . Duration (32b)

Service Time . . .Price_bid(8b)Priority (8b)

Channel(8b)Type (8b). .

. . . Device Name and Description

IE length. . . MAC Address

Source MAC Address (cont). . .

0 8 16 24 31

Tx Pwr (8b) Rx Pwr (8b)

Example CSCC message used in WLAN-Bluetooth prototype at WINLAB

Message type Flags Source MAC Address

Page 8: 1 Architecture and Protocol Design for Cognitive Radio Networks* Microsoft CR Summit, Jun 2008 Rutgers, The State University of New Jersey .

8

3adate:D:L:dxml_no_ddefault_fro1UTF-8UTF-8default_froSEARCH

CogNet Protocol: Validating GCP-based Spectrum Coordination on ORBIT

Multi-radio node 802.11a/b/g ad-hoc WiFi infrastructure mode (AP to

clients) Bluetooth

64kbps voice calls File synchronization between

PDAs, phones and laptops Mouse/keyboard

Zigbee Sensors

Potential WiMax Aggregated web/email traffic

to base stations

ORBIT Radio Grid

GCP Coordination Range

Page 9: 1 Architecture and Protocol Design for Cognitive Radio Networks* Microsoft CR Summit, Jun 2008 Rutgers, The State University of New Jersey .

9

3adate:D:L:dxml_no_ddefault_fro1UTF-8UTF-8default_froSEARCH

CogNet Protocol: Validating GCP-based Spectrum Coordination on ORBIT (cont.)

Data Radio Service

PHY Type IEEE 802.11g(Atheros

AR5212)

Bluetooth(USB Dongle)

Frequency 2427-2447MHz 2402-2480MHz

Modulation OFDM (256 FFT) QAM

FHSS

Transmit Power

18dBm 4dBm (~10m) (class 2)20dBm (~100m) (class 1)

PHY Rate 1M-54Mbps AutoRate

Upto 1Mbps (class 2)Upto 4Mbps (class 1)

Data session Pareto ON/OFF variable rate CBR: 5 sec

random session

Constant audio streaming (64, 128,320,512,

1024kbps)

BT

WiFi

Page 10: 1 Architecture and Protocol Design for Cognitive Radio Networks* Microsoft CR Summit, Jun 2008 Rutgers, The State University of New Jersey .

10

3adate:D:L:dxml_no_ddefault_fro1UTF-8UTF-8default_froSEARCH

CogNet Protocol: Validating GCP-based Spectrum Coordination on ORBIT (cont.)

Throughput Drops by ~3-4x in the case of 802.11g nodes and by ~1.5-2x for bluetooth nodes in dense topologies with 4 wifi and 4 Bt links. Results Averaged over 5 different topologies & load conditions. indicates the need for spectrum coordination

Wifi Performance

0.00

20.00

40.00

60.00

80.00

100.00

120.00

No-Interf With-Interf

Coexistence Effect on Wifi

Perc

en

tag

e T

hro

ug

hp

ut

802.11g Throughput

BT-Performance

0

20

40

60

80

100

120

No-Interf With-Interf

Coexistence Eff ect on BT

BT-Throughput

UDP throughput results with and without interference from other BT/WiFi users

Page 11: 1 Architecture and Protocol Design for Cognitive Radio Networks* Microsoft CR Summit, Jun 2008 Rutgers, The State University of New Jersey .

11

3adate:D:L:dxml_no_ddefault_fro1UTF-8UTF-8default_froSEARCH

Characteristics : Each individiual in the room carries two

radios bluetooth and wifi Node density High

28 radios in ~3000sqft 14 Bluetooth radio 14 Wifi radio

1M 5M 10M 15M

-50

0

50

100

WiFi offered load (bps)BT load 1Mbps

Wifi (BT-Rate) Wifi (BT-BO) BT (BT-Rate) BT (BT-BO) Total (BT-Rate) Total (BT-BO)

Thr

ough

put I

mpr

ovem

ent (

%)

CogNet Protocol: Validating GCP-based Spectrum Coordination on ORBIT (cont.)

Page 12: 1 Architecture and Protocol Design for Cognitive Radio Networks* Microsoft CR Summit, Jun 2008 Rutgers, The State University of New Jersey .

12

3adate:D:L:dxml_no_ddefault_fro1UTF-8UTF-8default_froSEARCH

CogNet Protocol: Beacon Format

Beacon format: (extended form of CSCC) Short message, low-layer function

Link weight/metric calculation: Estimate maximum supported data PHY rate

Direct link weight (proportional to achievable link rate)

M S G Type Flags S equenc e NumberS ourc e ...

...Identifier M ax P HY RateM ax T rans mit P ower B eac on T rans mit P ower

Num o f Reac h M AC Type M AC B us y Indic a to r

1 8 16 24 32

NA CF FD 0 0 0 0 0

8 10 12 14 16

Flags :

}),(min{},min{ maxmaxmaxmax jijmapjiij RSNRfRRR 0

)(

)(max Pr

NPt

PtSNR

Bji

Bjii

ij

},min{max MACjMACiijij RL MAC Idle Ratio

Page 13: 1 Architecture and Protocol Design for Cognitive Radio Networks* Microsoft CR Summit, Jun 2008 Rutgers, The State University of New Jersey .

13

3adate:D:L:dxml_no_ddefault_fro1UTF-8UTF-8default_froSEARCH

CogNet Protocol: Network Discovery

Obtain global awareness by aggregating local link states Discover end-to-end paths with path weight Use only one-hop broadcast for periodical update Trade-off between network setup time and overhead

Link state aggregation message format Flags: PR – Poll (0) / Response (1), UB – Unicast (0) / Broadcast (1) response

required, FD – Forwarded or not, FU – Full or updated

MSG Type Flags Source......Identifier

TTL Valid Time Number of VectorsMessage Hash ID

Link S tate Vector 1

Link S tate Vector 2

. . . . . .

1 8 16 24 32

P R UB FD FU 0 0 0 09 10 11 12 13 14 15 16

F lags :

Destination Node......Identifier E2E P ath Weight

Next Hop Node......Identifier Hop Count

Page 14: 1 Architecture and Protocol Design for Cognitive Radio Networks* Microsoft CR Summit, Jun 2008 Rutgers, The State University of New Jersey .

14

3adate:D:L:dxml_no_ddefault_fro1UTF-8UTF-8default_froSEARCH

CogNet Protocol: Data Path Establishment Hop-by-hop cross-layer parameter setup

Configure data plane and reserve radio resources by joint frequency/power/rate/bandwidth allocation

Unified message format for “up/down” hop setup

Multi-channel Data Path

Link State Aggregation

Control Plane Coverage

Link State Table

SourceDestination

Hop-by-hop Resource Allocation

M S G Type Flags M es s age S ender ...... Identifie r

Flow Des tina tion ...... Identifie r S es s ion Dura tion

C urrent T ime S tampHop R ec e ive r ...

... Identifie r M AC Type as S enderC hanne l Ava ilab ility M ap

M in P WR M ax P WR M in R a te M ax R a teHop S ender ...

... Identifie r M AC Type as R ec e ive rFrequenc y B andwidth

M odula tion C oding TX P ower P HY R a te

1 8 16 24 32

UC R V S D O T 0 0 0 09 10 11 12 13 14 15 16

F lags :

Page 15: 1 Architecture and Protocol Design for Cognitive Radio Networks* Microsoft CR Summit, Jun 2008 Rutgers, The State University of New Jersey .

15

3adate:D:L:dxml_no_ddefault_fro1UTF-8UTF-8default_froSEARCH

CogNet Protocol: ns2 Simulation

Evaluation by ns-2 simulations Bootstrap/Discovery: network setup time, overhead, theoretical end-to-end rate DPE: joint F/P/R allocation success ratio, overhead Naming/addressing: uniqueness of IP/Name

Ad hoc network – nodes randomly boot up

Control Interface

(802.11b)

Data Interface

(generic OFDM radio parameters)

Page 16: 1 Architecture and Protocol Design for Cognitive Radio Networks* Microsoft CR Summit, Jun 2008 Rutgers, The State University of New Jersey .

16

3adate:D:L:dxml_no_ddefault_fro1UTF-8UTF-8default_froSEARCH

CogNet Protocol: Discovery & Path Setup Simulation Results

Maximum and average network setup time (BSB interval 2sec, LSA interval 5sec, nodes randomly start [0, 4]sec)

Control overheadTheoretical max end-to-end rate averaged over the network

Page 17: 1 Architecture and Protocol Design for Cognitive Radio Networks* Microsoft CR Summit, Jun 2008 Rutgers, The State University of New Jersey .

17

3adate:D:L:dxml_no_ddefault_fro1UTF-8UTF-8default_froSEARCH

CogNet Protocol: Dynamic MAC Switching Using GCP Control GCP offers control support necessary for MAC switching, for example from

CSMA to TDMA GCP messages carry state information needed by decentralized MAC

switching algorithm at each node GCP control used to set up TDMA schedule involving multiple nodes

Control link

Data path

Sender

Receiver

CH1_CSMA

CH2_CSMA

CH4_CSMA

CH3_CSMA

CH5_CSMA

CH1_CSMA

Delay increase > 20%Request TDMA SwitchA

B

CH10_TDMA Slot = 1

CH10_TDMA Slot = 3

CH10_TDMA Slot = 5

Page 18: 1 Architecture and Protocol Design for Cognitive Radio Networks* Microsoft CR Summit, Jun 2008 Rutgers, The State University of New Jersey .

18

3adate:D:L:dxml_no_ddefault_fro1UTF-8UTF-8default_froSEARCH

CogNet Protocol: Dynamic MAC Switching Using GCP Control (cont.)

Sender Node A Node B Receiver

Preferred Channel List

Match channel

CH3_CSMA

Delay > 20%

Preferred Channel List

Match channel

CH5_CSMA Preferred Channel List

Match channel

CH1_CSMA

TDMA Join (Slot #3)TDMA Join (Slot #1)

TDMA Join (Slot #5)

CH10_TDMA (Slot #3)CH10_TDMA (Slot #1)

CH10_TDMA (Slot #5)

Request TDMA switchRequest TDMA switch

Request TDMA switch

GNU radio implementation currently in progress Sample protocol exchange between nodes shown below

Page 19: 1 Architecture and Protocol Design for Cognitive Radio Networks* Microsoft CR Summit, Jun 2008 Rutgers, The State University of New Jersey .

19

3adate:D:L:dxml_no_ddefault_fro1UTF-8UTF-8default_froSEARCH

CogNet Protocol: Future Work

Complete validation of key components MAC switching, cross-layer routing protocols, adaptation algorithms, …

Complete baseline v1.0 protocol spec Support for dynamic spectrum, bootstrap/discovery, MAC switching and cross-

layer routing

End-to-end wired Internet integration issues CR supernode and aggregation gateway details

Protocol implementation on GNU radio platform GNU/ORBIT release planned for AY08-09 ORBIT upgrade to URSP2

Experiments with adaptive wireless networks Apply to dynamic networking scenarios (tactical, vehicular) and demonstrate value of

coordination, cooperation and adaptation

Page 20: 1 Architecture and Protocol Design for Cognitive Radio Networks* Microsoft CR Summit, Jun 2008 Rutgers, The State University of New Jersey .

20

3adate:D:L:dxml_no_ddefault_fro1UTF-8UTF-8default_froSEARCH

Future work: ORBIT Node Upgrade to CR ORBIT radio grid testbed currently supports ~10 GNU radios and for ~100 low cost

programmable radio boards Plan to upgrade ~64 radio nodes with combination of GNU/USRP2 boards and WINLAB

hardware platforms for higher performance evaluations; will include baseline CogNet stack

Urban

300 meters

500 meters

Suburban

20 meters

ORBIT Radio Grid

Office

30 meters

Radio Mapping Concept for ORBIT Emulator

400-node Radio Grid Facility at WINLAB Tech Center

ProgrammableORBIT radio node

URSP2CR board

Planned upgrade(2007-08)

Current ORBIT sandbox with GNU radio