Cognitive Radios for Open Access to Spectrum

Narayan MandayamChristopher Rose

Predrag SpasojevicRoy Yates

(in partnership with Lucent Technologies)

Rutgers University www.winlab.rutgers.edu

Email: narayan@winlab.rutgers.edu

Project MotivationTriumph of Technology vs. Triumph of Economics

• What everyone agrees on:– Spectrum use is inefficient– FCC licensing has yielded false scarcity

• The Spectrum Debate:– Open Access (Commons)

• [Noam, Benkler, Shepard, Reed …]• Agile wideband radios will dynamically share a commons

– Spectrum Property Rights• [Coase, Hazlett, Faulhaber+Farber] • Owners can buy/sell/trade spectrum• Flexible use, flexible technology, flexible divisibility,

transferability• A spectrum market will (by the force of economics) yield

an efficient solution

The Open Access Conundrum• Systems of end-user devices• Technology Panacea

– Spread spectrum, UWB, MIMO– Short range communications– Ad hoc multi-hop mesh networks

• Evidence: success of 802.11 vs. 3G• Minor technical rules for transceivers

– power, spreading

• Technical arguments against Open Access– Partially developed theory

• ad hoc network capacity, with/without mobility• IT relay and interference channel

– Infant technology• UWB, antenna arrays• Transmitter agility

– Technology not separable from user assumptions• Capabilities of technology vary with cooperation

Technology for Enabling Open Access?

Research themes that have emerged from adhoc/sensor networks research:

• Hierarchical Architecture wins– Capacity scaling, energy efficiency, increases

lifetimes, facilitates discovery• Cooperation wins

– Fundamental performance limits via information theoretic relay and broadcast channels

• “Global” awareness and coordination wins– Space, time and frequency awareness and

coordination beyond local measurements

Technology for Enabling Open Access?

Require radios that can :– Discover– Cooperate– Collaborate– Self-Organize into hierarchical networks

• Agility is necessary at every layer of the “protocol stack”

• But cannot yet predict environmentsThe Answer? “Cognitive Radios”

Cognitive Radios for Enabling Open AccessPOOF! Let there be cognitive radios!

(let Bryan worry about details)Then What???

• Discovery strategies for available spectrum– Algorithms and protocols for frequency selection, coordination and

cooperation• Coding strategies for efficient sharing

– Cooperative diversity and coding for increased spectrum utilization• Information strategies for efficient cooperation

– “Spectrum Servers” to advise/mediate sharing• Negotiation strategies for situations of conflict

– Pricing and game theoretic strategies to promote cooperation• Domination strategies for situations of conflict

– Spectrum warfare with agile waveforms

Internet-based Spectrum Policy ServerInternet-based Spectrum Policy Server can help to coordinate wireless networks - needs connection to Internet even under congested conditions (...low bit-rate OK)- some level of position determination needed (..coarse location OK)- spectrum coordination achieved via etiquette protocol centralized at server

Internet

Access Point(AP2)

AP1WLANoperator A

WLANoperator B

Ad-hocBluetoothPiconet

Wide-areaCellular data

service

SpectrumPolicy Server

www.spectrum.net

AP1: type, loc, freq, pwrAP2: type, loc, freq, pwrBT MN: type, loc, freq, pwr

MasterNode

EtiquetteProtocol

Spectrum Policy Server (SPS)

SPS Methodology• New users get an SPS

address– Analogous to DHCP

• Users send activity traces to SPS server

• SPS maintains database of activity traces

SPS

SPS Issues• What to measure?• Coarseness of

measurements?• Trace update frequency?• SPS database organization

to facilitate fast searching?

SPS Actions• Share wireless node

descriptors?• Coordinate comm. between

nodes?• Mediate spectrum sharing

among nodes?

• Forwarding reduces interference, saves battery life, increases coverage area

• There are two kinds of costs associated with packet forwarding:

– Power Consumption (REAL COST)

– Delay in Own Data (OPPORTUNITY COST)

• Why (and how) will autonomous nodes forward ?

Incentives for Forwarding?

Fostering Cooperation Through “Bribery”

A Microeconomic Framework for Enabling Forwarding: Pricing & Reimbursement

• Nodes control their:– Transmit powers– Forwarding Preferences– Destination

Preferencesto maximize bits/Joule

• The network controls:– Pricing Coefficient– Reimbursement

Coefficient: to maximize its

revenue

NETWORK ( ACCESS POINT )

Runs an optimization algorithm to determine the

Price for channel use

Reimbursement for forwarding

to maximize the revenue

Flow of parameters between adjacent nodes

THE USERS PLAY A NON-COOPERATIVE GAME TO MAXIMIZE THEIR UTILITIES

USER USER USER text

text

text

The network declares thepricing parameters

The network inspects therevenue as an output of

declared pricingparameters

Horizontal Trajectory

Radio tower

5 m 10 m

-5 m

-

USER 2(POTENTIAL

FORWARDER)

USER 1(NON-FORWARDER)

ACCESS POINT

5 m POSITIONS OFUSER 1 IN

HORIZONTALGEOMETRY

• Incentives for forwarding works when nodes tend to “cluster” • The aggregate bits/Joule in network is higher

• Network revenue is also higher

0 1 2 3 4 5 6 710

14

1015

1016

1017

1018

1019

Aggregate bits/JouleAggregate bits/Joule, no reimbursement.

Two User-One Access Point Example

user 2 distance

What You Learned in KindergartenThe World Is A Rough Place

• Everybody wants what they want– Greed is the norm

• The Golden Rule– Be Polite– Play Nice

• The REAL Golden Rule– Be aware of your surroundings– Develop a good left hook– Bruises are bad– Forgiveness is divine (when you have a good left

hook)

A Toy Problem

2 Users with Unit Power2 Orthogonal ChanNels

Symmetric -10dB InterferenceHigh (30dB) SNR

Signal Space Game

1.0 1.0 5.7 2.2 5.0 5.0

2.2 5.7 2.0 2.0 2.2 5.7

5.0 5.0 5.7 2.2 1.0 1.0

Preliminary ResultsIf Better Performers Survive in

Marketplace…• Adaptive Policy

– tit for tat• Even Better

– tit for tat with random start– forgiveness is divine

• Use Genetic Algorithms– evolve against chosen policies– evolve against evolving policies

Evaluation of Cognitive Approaches

HardwareComplexity

Efficiency

ProtocolComplexity

Pricing collaboration

Spectrumserver

Agileradio

Coded cooperation

Cognitive protocols

• Capacity performance– Information rates, spectral

efficiency, BER, FER, outage • Cost performance

– Hardware complexity– Software complexity

• Emulation of large-scale networks of cognitive radios – ORBIT Testbed– Emergent behavior