Cognitive Radio Research@WINLAB Roy Yates WINLAB Rutgers - - PowerPoint PPT Presentation

1

WINLAB

Cognitive Radio Research@WINLAB

Roy Yates

WINLAB Rutgers University

December 10, 2008

ryates@winlab.rutgers.edu www.winlab.rutgers.edu

2

WINLAB

Cognitive Radio Research A Multidimensional Activity

Spectrum Policy

Economics Regulation Legal Business

Theory and Algorithms

Fundamental Limits Information & Coding Theory Cooperative Communications Game Theory & Microeconomics

Hardware/ Software Platforms &

Prototyping

Programmable agile radios

GNU platforms Cognitive Radio Network Testbed

3

WINLAB

FCC Spectrum Management

False Scarcity

All bands are allocated,

many for multiple purposes.

Most bands are actually

largely unused.

Maximum Amplitudes

Frequency (MHz) Amplidue (dBm)

Heavy Use Sparse Use Heavy Use Medium Use

Less than 6% Occupancy Less than 6% Occupancy

All bands are allocated,

many for multiple purposes.

Most bands are actually

largely unused.

4

WINLAB

Spectrum Policy Debate

Property Rights

Triumph of Economics [ Coase, Hazlett, Faulhaber+ Farber] Owners can buy/ sell/ allocate spectrum A spectrum market will yield an efficient solution

Open Access and Com m ons

Triumph of Technology [ Noam, Benkler, Shepard, Reed] Agile radios to dynamically share common spectrum Open Access: Strict technology needs- sensing, interference Commons: Distributed protocol followed in system

6

WINLAB

The Spectrum Debate & Cognitive Radio

What everyone agrees on now: ☺

Spectrum use is inefficient FCC licensing has yielded false scarcity

Possible middle ground?

Dynamic spectrum access Short-term property rights Spectrum use driven by both technology and market forces

Cognitive Radios with ability to incorporate market forces?

Microeconomics based approaches to spectrum sharing

“Dynamic Spectrum Access Models: Towards an Engineering

Perspective in the Spectrum Debate” by Ileri & Mandayam, IEEE Communications Magazine, Jan 2008.

7

WINLAB

Cognitive Approaches: Outlook

Lots of netw ork ( and channel) state inform ation

needed to enable efficient

Discovery Self-organization Cooperation Techniques

A

A

B

B

D

C D E F

Bootstrapped PHY & control link End-to-end routed path From A to F PHY A PHY B PHY C Control (e.g. CSCC) Multi-mode radio PHY Ad-Hoc Discovery & Routing Capability

Functionality can be quite challenging!

8

WINLAB

Cognitive Radios need information

Reactive schemes (without explicit coordination protocols)

have limitations.

• Interference is a receiver property.

Alternative: Infrastructure-based coordination Examples of coordination mechanisms:

Information aids

“Spectrum Coordination Channel” to enable spectrum sharing

Network architectures

“Spectrum Servers” to advise/ mediate sharing

9

WINLAB

Common Spectrum Coordination Channel (CSCC) [Jing, Raychaudhuri]

• CSCC

CSCC can coordinate radios with incompatible PHY

Employs an out-of-band etiquette channel & protocol

Periodic TX of radio parameters on CSCC TX at higher power to reach hidden nodes

Local contention resolved via protocol-independent etiquette policies Also supports ad-hoc multi-hop routing associations

Ad- hoc net B Ad- hoc net A Ad-hoc Piconet Maste r Node CSCC RX range for X CSCC RX range for Y

Y X

Jing, Raychaudhuri

10

WINLAB

802.11 & 16 Co-Existence: Reactive vs. CSCC-based Power Control []

CSCC frequency adaptation when DSS-AP = 200m and traffic load 2Mbps

AP

1km

100m

802.11b Hotspot 802.16a Cell BS SS DSS-AP

802.16 DL 802.11 link 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8

Average Link Throughput (Mbps) 802.16 DL and 802.11 link No Coordination CSCC frequency adaptation

Throughputs vs. DSS-AP by using CSCC pow er adaptation and traffic load 2Mbps Single 802.11 Hot Spot Case

200 400 600 800 1000 0.0 0.2 0.4 0.6 0.8 1.0 1.2

Average Link Throughput (Mbps) Distance between 802.16 SS and 802.11 hotspot (meters) 802.16a DL 802.16a DL with CSCC 802.11 link 802.11 link with CSCC Average No Coordination Average with CSCC

11

WINLAB

(( ))

AP TV broadcasting

Secondary sensors

Wireless LANs Home networking

• rthogonal & non-orthogonal

multiplexing schemes Multiaccess/variable rate transmission schemes Interference channel / wideband transmissions

Spectrum server

multihopping

Wireless world is not flat!

What can a Spectrum Policy Server do?

(Anything & Everything)

Implicit CSCC

12

WINLAB

Cognitive Radio: Spectrum Policy Server

• Internet-based Spectrum Policy Server (a “Google for spectrum”)
• IP-based CSCC
• Coarse location information
• SPS provides centralized local spectrum coordination

Internet Internet

Access Point (AP2) AP1 WLAN

• perator A

WLAN

• perator B

Ad-hoc Bluetooth Piconet Wide-area Cellular data service Spectrum Policy Server w w w .spectrum.net AP1: type, loc, freq, pwr AP2: type, loc, freq, pwr BT MN: type, loc, freq, pwr Master Node Etiquette Protocol

13

WINLAB

Cognitive Radio Networks

Cross Layer Scheduling via a spectrum server

• Scheduling Physical links -

achievable rates depend on PHY layer TX & RX

• Routing – decisions based on

network/ application layer metrics

• Constraints

Link Rate > ∑ flows on link

• Perform ance bounds via

centralized scheduling

• Low -com plexity scheduling

schem es

• Randomized Distributed scheduling

(RDS)

• Column generation methods for

choosing good scheduling modes

(( ))

Flow 1 Flow 2

[ Raman, Mandayam, Yates]

14

WINLAB

Two Tier Dynamic Spectrum Access

Spectrum Policy Server/ Regulator/ Clearinghouse Service Providers ( SP) compete End Users: Adapt rate, power, spectrum use for max net utility Level I SPs obtain spectrum from SPS Level II End users obtain spectrum from SPs Examples: 802.22 Service Providers OFDM tone allocation to end users DimsumNet

15

WINLAB

Two Tier DSA Properties

Develop engineering models for

shaping spectrum policy

Features:

Dynam ic Spectrum Access: Short

term allocation of spectrum resources

Tem porary Exclusive Usage: Parties

do not suffer interference

Market Based Allocation: Supply and

demand determines who gets how much bandwidth

16

WINLAB

Two-Tier Spectrum Access Mechanisms

D-Pass (Dynamic Property-

Rights Spectrum Access)

Allocation based charges

SPs pay for spectrum allocation SPs then compete for users via

simultaneous auctions D-CPass (Dynamic-Commons

Property-Rights Spectrum Access)

Usage based charges

Clearinghouse mediates bidding

among users

SPs only pay for spectrum actually

used D-CPass yields better spectrum

utilization

Ileri, Mandayam

19

WINLAB

Two Tier DSA User and Service Provider Heterogeneity

Spectrum Clearinghouse Service Providers ( SP) compete to maximize profits End Users: Adapt rate, power, spectrum use for max net utility Level I SPs buy/ lease spectrum from clearinghouse Level II End users lease spectrum from SPs Level II End users lease spectrum from SPs

Acharya, Yates

20

WINLAB

Cost Model for the SP

Spectrum Cost

C(X) = CX, X: sum of spectrum from all users Constant C set by clearinghouse

Depends on Geographical region, urban/ rural

Power Cost

Transmit power = νX F(ν,X) = TνX Constant T may depend on

Presence of other providers in band ‘X’

21

WINLAB

User j: spectrum & rates

• hj = downlink gain
• TX: fixed power spectral density ν
• Spectral efficiency

Kj = log(1+ νhj / N0 )

• xj = allocated spectrum
• Rate Rj = Kjxj,
• Higher Spectral Efficiency Kj

= Better Radio Technology

user j SP

22

WINLAB

User j: utility functions

Logarithmic

U(Rj ) = log(1+ Rj )

Elastic data application

(large file download)

Exponential

U(Rj ) = Γj [ 1-exp(-Rj / Γj )]

Application with target rate Γj

Rj U(Rj ) Rj U(Rj ) Γj

23

WINLAB

Objective of the SP and Users

Clearinghouse set spectrum price SP maximizes its net revenue

Expense: Spectrum purchase and transmit power Income: Charges the users

Users maximize their utility minus cost

Expense: Charge paid to the SP Gain: Increase in utility due to spectrum

24

WINLAB

Elasticity of Demand

• Ratio of % change in demand to % change in price
• Logarithmic Utilities: Elastic demand (ε> 1) always
• When price is increased, % fall in demand is higher
• SP can’t arbitrarily overprice spectrum
• Exponential Utilities: Inelastic demand (ε< 1) for low μ
• Low μ: Enough spectrum for users to be rate saturated
• Price changes in this regime, % change in demand is less

25

WINLAB

SP Profit vs Spectrum Cost C

Exponential Utilities Target Rate = 1 Mbps Power Cost, T= 10 Total cost Ce = C+ Tν High C Ce = C+ Tν ~ C User Utility ↑ as ν ↑ SP incentive: High ν Low C Ce = C+ Tν ~ Tν SP incentive: Low ν

0.5 1 1.5 2 2.5 3 2 4 6 8 10 spectrum cost (C) $/MHz Profit of SP ν = 50 dBm/MHz ν = 30 dBm/MHz ν = 20 dBm/MHz ν = 10 dBm/MHz

26

WINLAB

User Net Utility

0.5 1 1.5 2 2.5 3 0.4 0.5 0.6 0.7 0.8 0.9 1

spectrum cost (C) $/Hz User Utilities (Mbps)

ν = 10 dBm/MHz, Strongest User ν = 50 dBm/MHz, Strongest User ν = 50 dBm/MHz, Weakest User ν = 10 dBm/MHz, Weakest User

Exponential Utilities Target Rate = 1 Mbps Power Cost, T= 10 Total cost Ce = C+ Tν Low C and high ν Ce = C+ Tν ~ Tν SP costs rise with ν For high ν

• SP buys less spectrum
• User net utility reduced

High C and high ν Ce = C+ Tν ~ C SP costs indifferent to ν User utility ↑ as ν ↑

27

WINLAB

Barter Mechanisms

• Exchange goods that are of

mutual value

• Shared understanding of

worth

• More immediate appreciation
• f benefits
• Barter Mechanisms
• Content Exchange
• I relay for you, you give me access

to some files

• Connectivity Exchange
• I relay for you, you relay for me at

the same time

• Bandw idth Exchange
• Assume each user has an

exclusive spectrum band

• I relay for you, you lend me some

bandwidth

28

WINLAB

Emerging Themes

Hierarchical Network Architecture wins Capacity scaling, energy efficiency, increases lifetimes,

facilitates discovery

Cooperation wins Achievable rates via information theoretic relay and broadcast

channels

“Global” awareness and coordination wins Space, time and frequency awareness and coordination beyond

local measurements

Efficient operation requires radios that can: Cooperate Collaborate Discover Self-Organize into hierarchical networks