ashwini kumar kang shin university of michigan aug-6-2009, icccn 2009, san francisco a case study of...

23
Ashwini Kumar Kang Shin University of Michigan Aug-6-2009, ICCCN 2009, San Francisco A Case Study of QoS Provisioning in TV-band Cognitive Radio Networks Jianfeng Wang (presenter) Kiran Challapali Philips Research

Upload: nathan-ramsey

Post on 11-Jan-2016

217 views

Category:

Documents


2 download

TRANSCRIPT

Page 1: Ashwini Kumar Kang Shin University of Michigan Aug-6-2009, ICCCN 2009, San Francisco A Case Study of QoS Provisioning in TV-band Cognitive Radio Networks

Ashwini KumarKang Shin

University of Michigan

Aug-6-2009, ICCCN 2009, San Francisco

A Case Study of QoS Provisioning in TV-band Cognitive Radio Networks

Jianfeng Wang (presenter)Kiran Challapali

Philips Research

Page 2: Ashwini Kumar Kang Shin University of Michigan Aug-6-2009, ICCCN 2009, San Francisco A Case Study of QoS Provisioning in TV-band Cognitive Radio Networks

2

Outline

• Introduction of TV Whitespace• System & QoS Model

• QoS-Provisioned DSA Protocol (QPDP) Overview• Distributed Reservation & Channel Access• Network & Spectrum Management

• Evaluation• Conclusion

Page 3: Ashwini Kumar Kang Shin University of Michigan Aug-6-2009, ICCCN 2009, San Francisco A Case Study of QoS Provisioning in TV-band Cognitive Radio Networks

What is TV Whitespace

• TV Band Incumbents – TV, WM• TV bands only sparsely used today

(see graphic)• Fewer and fewer US households

rely on over-the-air TV (from FCC report)

– 33 % in 1994, 15 % in 2004 – Among these, on average only a

few channels watched• TV bands have nice propagation

characteristics for various applications

• FCC has taken steps on opening TV bands for unlicensed use

Source: New America Foundation

Source: FCC. Reported by New America Foundation

Page 4: Ashwini Kumar Kang Shin University of Michigan Aug-6-2009, ICCCN 2009, San Francisco A Case Study of QoS Provisioning in TV-band Cognitive Radio Networks

White space regulatory milestones – US

Notice of Inquiry

?. 2009

2005

Oct. 2006

20042003

June 2008

Mar. 2007

Notice of Proposed Rule Making

Initial R & OandFurther NPRM

July 2007

Public Notice

Sep.2006

Report on Interference Rej. Cap. of DTV Rx’s

Field Tests Final rules in Federal Registry

Report on Sensing, Interference to DTVs & Other Radios

Nov. 2008

Final Rule and Order

Feb. 2009

Philips CompleteCR Demo @ FCC

Aug. 2008

Mar. 2008Sensing Proto Testing

Dec. 2008

Page 5: Ashwini Kumar Kang Shin University of Michigan Aug-6-2009, ICCCN 2009, San Francisco A Case Study of QoS Provisioning in TV-band Cognitive Radio Networks

FCC 2nd Report and Order

• Personal/Portable TVBD (unlicensed) Devices

– Up to 100mW; limited to 40mW if operating in adjacent channels.

– Any channel between 21 and 51, except channel 37.

– Mode II device (Master device) must employ geo-location database to determine channel availability.

– Mode I device (Client device) operates under signaling control of Mode II device.

– All devices should also employ sensing mechanism to determine channel availability.

– Incorporate a dynamic frequency selection (DFS) mechanism and transmission power control (TPC) mechanism.

– Sensing only device operates <= 50mW.

Page 6: Ashwini Kumar Kang Shin University of Michigan Aug-6-2009, ICCCN 2009, San Francisco A Case Study of QoS Provisioning in TV-band Cognitive Radio Networks

System & QoS Model

• Personal/portable TV band unlicensed devices equipped with one radio

• Cast study to provide HDTV streaming in home WLANs

• QoS met for multimedia traffic

6

Cable/Internet AP(Residential Gateway)

Page 7: Ashwini Kumar Kang Shin University of Michigan Aug-6-2009, ICCCN 2009, San Francisco A Case Study of QoS Provisioning in TV-band Cognitive Radio Networks

Design Challenge

• Complexity and overhead for coordinating sensing incumbents as low as -114dBm

– in personal/portable mobile environments

• Incumbents’ interference and interruption

• Stringent requirements of real-time multimedia traffic (e.g., HDTV streaming)

• Narrow channel-width (6 MHz) – Not much chance to use multiple contiguous channels

7

Page 8: Ashwini Kumar Kang Shin University of Michigan Aug-6-2009, ICCCN 2009, San Francisco A Case Study of QoS Provisioning in TV-band Cognitive Radio Networks

Self coexistence issue

• Resource sharing and QP synchronization across neighboring networks

8

Page 9: Ashwini Kumar Kang Shin University of Michigan Aug-6-2009, ICCCN 2009, San Francisco A Case Study of QoS Provisioning in TV-band Cognitive Radio Networks

QPDP Overview

• QPDP logically consists of Lower & Upper MAC

• Upper MAC• Spectrum management and

network management • Based on overlay master-slave

architecture

• Lower MAC• Slot reservation• Self-coexistence

9

Distributed Reservation Access Based on WiMedia MAC

Upper MAC

Lower MAC

Overlay Master-Slave Operation

Spectrum Management

Function

Network Management

Function

QPDP MAC architecture

Page 10: Ashwini Kumar Kang Shin University of Michigan Aug-6-2009, ICCCN 2009, San Francisco A Case Study of QoS Provisioning in TV-band Cognitive Radio Networks

QPDP Lower MAC functions

• Channel access follows time-recurring superframe structure– Each superframe consists of 256 MASs– MASs divided between BP, DSSP, SW

• Distributed beaconing and channel reservation – MASs reservations negotiated through beacons– BP merge for multiple network coexistence 10

Data/Sense/Sleep Period (DSSP)

…...

…...

mMASLength

Medium Access Slots (MASs)

Superframe m Superframe m+1Superframe m-1…... …...

...

Beacon Period (BP)

SignallingWindow

(SW)

…...

...

Adjustable

mBeaconSlotLength

Beacon Period (BP)

0 1 N 0 1 N

QP

Page 11: Ashwini Kumar Kang Shin University of Michigan Aug-6-2009, ICCCN 2009, San Francisco A Case Study of QoS Provisioning in TV-band Cognitive Radio Networks

QPDP Upper MAC Functions

• Overlay master manages channel, sensing and device association

• Channel management made intelligent to reduce disruptions– Prioritized channel list– Backup channels– Channel-imaging

• Multi-level spectrum sensing to minimize overheads– Multiple short QPs within CDT– Long QPs scheduled on-demand

• Network entry & device discovery automated through boot-up scan and beacons

11

Page 12: Ashwini Kumar Kang Shin University of Michigan Aug-6-2009, ICCCN 2009, San Francisco A Case Study of QoS Provisioning in TV-band Cognitive Radio Networks

Evaluation Setup

• To analyze QPDP performance w.r.t. QoS provisioning– Efficiency in supporting high data-rate, low error-rate & delay– Robustness in response to incumbent disruptions

• Simulations using OPNET Modeler• Home network setting, with HDTV streaming as multimedia

application

• Simulation parameters:– Sender-receiver pair, distance=30m– Exponential rayleigh multipath fading– Transmit power=30dbm, path loss factor=3– PHY based on OFDM: 128 FFT

12

Page 13: Ashwini Kumar Kang Shin University of Michigan Aug-6-2009, ICCCN 2009, San Francisco A Case Study of QoS Provisioning in TV-band Cognitive Radio Networks

Results• Requirements of HDTV streaming achieved (~19.3 Mbps, <100ms

delay, BER<0.05) with proper setting • Impact of sensing schedule:

– FS-1 to FS-6: same long-term overhead , differ in short-term– Recovery is quick in both low & high-power incumbent case

130 10 20 30 40 500

0.5

1

1.5

2

2.5x 10

7

Time (s)

Thr

ough

put

(bit/

s)

FS-1

FS-2

FS-3

FS-4

FS-5

FS-6

Low Power Incumbent (iRxPr = -100.25dBm)

0 10 20 30 40 500

0.5

1

1.5

2

2.5x 10

7

Time (s)

T

hrou

ghpu

t (b

it/s)

High Power Incumbent (iRxPr = -40.25dBm)

FS-1

FS-2

FS-3

FS-4

FS-5

FS-6

Page 14: Ashwini Kumar Kang Shin University of Michigan Aug-6-2009, ICCCN 2009, San Francisco A Case Study of QoS Provisioning in TV-band Cognitive Radio Networks

Results (contd.)

• Combining fast sensing with fine sensing performs better• Delay is sensitive to short-term sensing schedule

14

0 10 20 30 40 500

0.5

1

1.5

2

2.5x 10

7

Time (s)

Thr

ough

put

(bit/

s)

ED + FS-1, iRxPr = -40.25dBm

ED + FS-2, iRxPr = -50.25dBm

ED + FS-3, iRxPr = -60.25dBm

ED + FS-4, iRxPr = -70.25dBm

ED + FS-5, iRxPr = -80.25dBm

ED + FS-6, iRxPr = -90.25dBm

FS-1, iRxPr = -40.25dBm

0 10 20 30 400.75

0.8

0.85

0.9

0.95

1

Delay (ms)

Cum

ulat

ive

Dis

trib

utio

n F

unct

ion

(CD

F)

FS-1

FS-2

FS-3

FS-4

FS-5

FS-6

High Power Incumbent (iRxPr = -40.25dBm)

Page 15: Ashwini Kumar Kang Shin University of Michigan Aug-6-2009, ICCCN 2009, San Francisco A Case Study of QoS Provisioning in TV-band Cognitive Radio Networks

Results (contd.)• Fast incumbent detection and optimized channel-switch

minimizes traffic loss and sustains QoS• Packet aggregation very useful in sustaining QoS

155 10 15 20 25 301

1.5

2

2.5x 10

7

Time (s)

Th

rou

gh

pu

t (b

it/s)

MPDU size = 200Bytes

MPDU size = 400Bytes

MPDU size = 600Bytes

MPDU size = 800Bytes

MPDU size = 1000Bytes

MPDU size = 2000Bytes

Page 16: Ashwini Kumar Kang Shin University of Michigan Aug-6-2009, ICCCN 2009, San Francisco A Case Study of QoS Provisioning in TV-band Cognitive Radio Networks

Conclusion

• Presented a system study of HDTV streaming over single TV channel

• Proposed QPDP incorporates both fine-grained and coarse-grained QoS mechanisms, including:

– Distributed beaconing and channel reservation– Overlay based Master-Slave based spectrum

management

• Results and discussions reveal the impact of key design parameters on QoS

16

Page 17: Ashwini Kumar Kang Shin University of Michigan Aug-6-2009, ICCCN 2009, San Francisco A Case Study of QoS Provisioning in TV-band Cognitive Radio Networks

Questions?

17

[email protected]

Page 18: Ashwini Kumar Kang Shin University of Michigan Aug-6-2009, ICCCN 2009, San Francisco A Case Study of QoS Provisioning in TV-band Cognitive Radio Networks

Backup Slides

18

Page 19: Ashwini Kumar Kang Shin University of Michigan Aug-6-2009, ICCCN 2009, San Francisco A Case Study of QoS Provisioning in TV-band Cognitive Radio Networks

19

System Parameters

Parameter Value

Data Traffic Transmission Power (dBm)

20

Noise Power Spectrum Density (dBm)

-174

Noise figure (dB) 6

Implementation loss (dB) 6

Communication Distance (m) 30

Path loss exponent 3

HDTV traffic load (Mbps) 20

SDTV traffic load (Mbps) 6

Page 20: Ashwini Kumar Kang Shin University of Michigan Aug-6-2009, ICCCN 2009, San Francisco A Case Study of QoS Provisioning in TV-band Cognitive Radio Networks

20

PHY-OFDM parameters

Parameter Value

Number of data subcarriers, ND 104

Number of pilot subcarriers, NP 4

Total number of subcarriers, NFFT 128

Inner coding rate 5/6

RS outer coding, t 5

Modulation 64-QAM

Preamble 4 sym

PHY+MAC header 1 sym

Symbol duration (µs) 21.25

Page 21: Ashwini Kumar Kang Shin University of Michigan Aug-6-2009, ICCCN 2009, San Francisco A Case Study of QoS Provisioning in TV-band Cognitive Radio Networks

21

MAC Parameters

Parameter Value

Superframe length (µs) 110,592

mNumberMAS 256

mMASLength (µs) 432

mMaxBPLength (MAS) 5

Regular Quiet Period 1

mBeaconSlotLength (µs) 432

Page 22: Ashwini Kumar Kang Shin University of Michigan Aug-6-2009, ICCCN 2009, San Francisco A Case Study of QoS Provisioning in TV-band Cognitive Radio Networks

UHF Band After Digital Switch Over in UK

Source: Ofcom ConsultationFeb. 16 2009

Page 23: Ashwini Kumar Kang Shin University of Michigan Aug-6-2009, ICCCN 2009, San Francisco A Case Study of QoS Provisioning in TV-band Cognitive Radio Networks

Ofcom on TV White Space

• Released consultation on White Spaces on Feb. 16 2009, with comments due by May 01 2009. Awaiting next statement.

• Proposed parameters:

Source: Ofcom ConsultationFeb. 16 2009