slides qualcomm
DESCRIPTION
Slides QualcommTRANSCRIPT
1
Qualcom, SanDiego-Sept.15, 2008(Haykin)Cognitive Radio:
Research Challenges
Simon HaykinMcMaster, University
Hamilton, Ontario, Canada
email: [email protected] site: http://soma.mcmaster.ca
2
rms Possible
adio
Qualcom, SanDiego-Sept.15, 2008(Haykin)
Outline of The Lecture
1. Introductory Remarks2. The Essence of Human Cognition in the Simplest Te3. The Motivation behind Cognitive Radio4. Cognitive Radio Networks5. Major Functional Blocks Constituting a Cognitive R6. Spectrum Sensing7. Transmit-Power Control8. Dynamic-Spectrum Management9. Emergent Behaviour of Cognitive Radio Networks10. Concluding Remarks
ReferencesAcknowledgements
3
Qualcom, SanDiego-Sept.15, 2008(Haykin)Growth of Cognitive Radioduring the last
3 to 4 years(Starting with about 6 to 8
Reports and Conference Papers)
IEEE Papers: 1154
Springer Papers: 189
Elsevier Papers: 33
Cognitive Radio Paper (Haykin):cited 631 times
4
ems, what I haveareer, namely,
hines
.
Qualcom, SanDiego-Sept.15, 2008(Haykin)
Personal Perspective
• Under the umbrella of Cognitive Dynamic Systbeen working on for much of my professional c
Signal ProcessingCommunication TheoryControl TheoryRadar SystemsNeural Networks and Learning Mac
which have all come into focus for the first time
5
x”
Qualcom, SanDiego-Sept.15, 2008(Haykin)
Cognitive Radio:
“Thinking Outside the Bo
6
, both in depth
topic so
ource:
, if any,
Qualcom, SanDiego-Sept.15, 2008(Haykin)
1. Introductory Remarks
• Cognitive Radio is growing in leaps and boundsand breadth, all over the world.
• The question is: Why this surge of interest in a relatively new?
• The answer is twofold:
(i) Cognitive Radio solves a pressing need:Underutilization of a precious natural resThe Radio Spectrum.
(ii) Research Challenge:Cognitive radio is challenging in ways fewother wireless technologies are today.
7
the attention of
Qualcom, SanDiego-Sept.15, 2008(Haykin)
Introductory Remarks (continued)
• It is not just cognitive radio that is attractingresearchers all over the world. Rather, it is:
Cognitive Radar
Cognitive Car
Cognitive Information Processing
Cognitive Control
Cognitive Computation (including software)
Cognitive Optimization
• What I am leading up to is the new discipline:
“Cognitive Dynamic Systems”
. . .
8
n
Qualcom, SanDiego-Sept.15, 2008(Haykin)
2. The Essence of Human Cognitio in the Simplest Terms Possible
Figure 1. Human Cognitive Cycle in its most basic form
TheWorld
FeedbackChannel
Control(Action) (Sensing)
PerceptionPerception (Sensing)
9
n to argue that
ATIONAL TERMS.
Qualcom, SanDiego-Sept.15, 2008(Haykin)
Tasks of a Human Mind
An extract taken from the book:
“The Computer and the Mind”by
Johnson-Laird
• to perceive the world;• to learn, to remember, and to control actions;• to think and create new ideas;• to control communication with others;• to create the experience of feelings, intentions, and self-awareness.
Johnson-Laird, a prominent psychologist and linguist, went o
THEORIES OF THE MIND SHOULD BE MODELLED IN COMPUT
10
adio
nderutiliza-ays:
cation.
by secondary
oes not
Qualcom, SanDiego-Sept.15, 2008(Haykin)
3. Motivation Behind Cognitive R
• Significant underutilization of the radiospectrum
• Basically Cognitive Radio solves the spectrum ution problem in a tightly inter-coupled pair of w
(i) Sense the radio environment to detectspectrum holes in terms of both time and lo
(ii) Control employment of the spectrum holesusers efficiently, subject to the constraint:
The total power in each spectrum hole dexceed a prescribed limit.
11
nsceiver).
scenezer
Qualcom, SanDiego-Sept.15, 2008(Haykin)
4. Cognitive Radio Networks
Figure 2. Basic signal-processing cycle, as seen by a single user (tra
}Users
Radio Environment (Wireless World)
Dynamic spectrum managerand transmit-power controller
Radio- analy
Feedback channel
12
eless
d) by empoweringa continuous-time
formance of eachariations in the
users through
competing users;ess.
Qualcom, SanDiego-Sept.15, 2008(Haykin)
Cognitive Radio Defined
The cognitive radio network is a complex multiuser wircommunication system capable of emergent behaviour.
It embodies the following functions:
• to perceive the radio environment (i.e., outside worleach user’s receiver to sense the environment onbasis;
• to learn from the environment and adapt the pertransceiver (transmitter-receiver) to statistical vincoming RF stimuli;
• to facilitate communication between multiplecooperation in a self-organized manner;
• to control the communication processes amongthrough the proper allocation of available resources
• to create the experience of intention and self-awaren
13
trum in a
l users of the
Qualcom, SanDiego-Sept.15, 2008(Haykin)
Primary objectives of Cognitive Radio Networks
1. To facilitate efficient utilization of the radio specfair-minded way.
2. To provide highly reliable communication for alnetwork.
14
ate their
m hole iser.ser subject
um holess
network
Qualcom, SanDiego-Sept.15, 2008(Haykin)
5. Major Functional Blocksof Cognitive Radio
Function Action1. Spectrum sensing
Detect spectrum holes, estimpower contents.
2. Predictive modeling
Predict availability of spectruemployment by secondary us
3. Transmit- power control
Maximize data rate of each uto power constraints
4. Dynamic spectrum management
Control distribution of spectrfairly among secondary user
5. Packet routing Route the packets across the efficiently
15
lihood sense
incorporating
Qualcom, SanDiego-Sept.15, 2008(Haykin)
6. Spectrum Sensing
Potential Candidates
(i) Energy detection:Parametric (Model-dependent)
(ii) Cyclostationarity method:(Nonparametric)
(iii) Multitaper method:• Nonparametric• Close to optimality in the maximum like• Expandable to include
(a) Spatio-temporal processing(b) Temporal-frequency processing by
the Loève transform
16
ideration of both
waterfilling algorithmture.
system using a varia-cted dynamic systems
ensitivity and stabilityarying delays.
Qualcom, SanDiego-Sept.15, 2008(Haykin)
7. Transmit-power Control
• A cognitive radio network is a hybrid dynamic systemo Continuous dynamicso Discrete events
• Theoretical analysis of the resource allocation problem with consequilibrium and transient behaviours.
• Formulating the transmit power control problem within iterative (IWFA) framework based on the concept ofinterference tempera
o Robust non-cooperative gameo Max-min optimizationo Worst-case analysis regarding a specified uncertainty-set
• Modelling the network as a constrained piecewise affine (PWA)tional inequality (VI) reformulation of IWFA and theory of proje(PDS).
• Providing tools from control theory to facilitate the analysis of sof the whole network considering uncertainty andmultiple time-v
17
ust IWFA, wheninterference gains
Qualcom, SanDiego-Sept.15, 2008(Haykin)
Figure 3. Resource allocation results of simultaneous IWFA and rob2 new users join a network of 5 users, a subcarrier disappears, and are changed randomly to address the mobility of the users.
18
ent:t
tion on all users
Qualcom, SanDiego-Sept.15, 2008(Haykin)
8. Dynamic-Spectrum ManagemTime- and Location-dependenOptimization Problem
8.1 Centralized Approach
(i) Centre for collecting radio-scene informa
(ii) Globally optimal solution for the problem
(iii) Impractical for two main reasons:• High complexity• Non-scalability
19
elf-organiza-
a local-
ity, and
lexity.
Qualcom, SanDiego-Sept.15, 2008(Haykin)
Dynamic Spectrum Management (continued)
8.2 Decentralized Approach:
• Utilization of neurobiological principles of stion, with emphasis on learning.
• Emphasis on cognitive radio information onneighbourhood basis.
• Complexity is proportional to the user-denstherefore scalable to any size.
• Provision of a stable solution with less comp
• Suboptimal but satisfactory solution.
20
ive
ion of the radio spec-
i.e. co-operation with-
ffic jams, chaos, and
Qualcom, SanDiego-Sept.15, 2008(Haykin)
9. Emergent Behaviour of CognitRadio Networks
• Seemingly irreducible phenomena.
• Phenomena not explicitly programmed.
• Positive emergent behaviour: a harmonious and efficient utilizat
trum by all primary and secondary users of the cognitive radio (
out or with minimal coordination).
• Negative emergent behaviour: characterized by disorder (i.e. tra
unused radio spectrum).
• Emergency networks: swarm intelligence.
21
(encompassingone of the mostury:
g Force
s the candidatetions.
Qualcom, SanDiego-Sept.15, 2008(Haykin)
10. Concluding Remarks
• The Study of Cognitive Dynamic Systemscognitive radio, Cognitive radio, etc.) will beinfluential scientific endeavours in the 21st cent
Computer Thinking will be the Drivin
• Cognitive Radio is already being considered afor the 5th Generation of Wireless Communica
22
Qualcom, SanDiego-Sept.15, 2008(Haykin)Two New Books to watch out for:
1. Neural Networks and Learning MachinesSimon HaykinPrentice-Hall, 3rd editionNovember 2008
2. Foundations of Cognitive Dynamic SystemsSimon HaykinCambridge University Press(In preparation)
23
d wireless com-in Communica-23, pp. 201-220,
ive Radio”, In, edited by Vijay
l for Cognitiveblication subject
Qualcom, SanDiego-Sept.15, 2008(Haykin)
References
A. Papers
S. Haykin, “Cognitive Radio: Brain-empoweremunications, IEEE Journal on Selected Areastions, Special Issue on Cognitive Networks, vol.February. invited
S. Haykin, “Fundamental Issues in CognitCognitive Wireless Communications NetworksBhargava and Ekram Hossain, 2007, Springer.
P. Setoodeh and S. Haykin, “Robust ControRadio, Proc. IEEE, Feb. 2009. Accepted for puto revision.
24
rum Sensing for
Techniques forPower Control”,
Techniques fore Analysis”, US
ognitive Radio”,09.
Qualcom, SanDiego-Sept.15, 2008(Haykin)
S. Haykin, J. Reed, and D.J. Thomson, “SpectCognitive Radio”, ibid, under preparation.
B. Patents
S. Haykin, “Operating Environment AnalysisWireless Communication Systems - TransmitUS Provisional 60/617,639.
S. Haykin, “Operating Environment AnalysisWireless Communication Systems - Radio ScenProvisional 60/617,638.
C. Special Issue
S. Haykin, J. Reed, G. Li, and M. Shafi. “CTwo-volume Special Issue, Proc. IEEE, Feb. 20
25
eering Research
ancial support I
anding group of
eir commitment
uly challenging
Qualcom, SanDiego-Sept.15, 2008(Haykin)
Acknowledgements
I am grateful to the Natural Sciences and Engin
Council (NSERC) of Canada for the sustained fin
have had for so many years.
Last, but by no means least, I thank my outst
graduate students and Post-doctoral Fellow for th
to excellence and willingness to undertake tr
research projects.