Cognitive Radio
Communications for Dynamic
Spectrum Access
Slides by: Alexander M. Wyglinski
Research Assistant Professor
ITTC
The University of Kansas
alexw@ittc.ku.edu
This work was generously supported by the National Science Foundation (NSF), via grants
ANI-0230786 and ANI-0335272, and both the Defense Advanced Research Projects Agency
(DARPA) and the Department of the Interior National Business Center, via grant
NBCHC050166
#16 1
Outline
•
•
•
•
•
•
Motivation
What are Cognitive Radios?
How are they “cognitive”?
Agile Transmission
Kansas University Agile Radio (KUAR)
Conclusion
#16 2
Presentation Overview
•
•
•
•
•
•
Motivation
What are Cognitive Radios?
How are they “cognitive”?
Agile Transmission
Kansas University Agile Radio (KUAR)
Conclusion
#16 3
Current Spectrum Allocation
FCC frequency allocations for US radio spectrum
#16 4
Increasing Demand
• Rapid growth in the wireless communications
sector, requiring more spectral bandwidth
– Increasing number of users
– Plethora of new wireless services being offered
• Some applications are bandwidth-intensive
• As a result of this demand, available spectrum
under the legacy command-and-control regime is
becoming increasingly scarce
– Number of licensed transmissions are increasing within a
finite allocated bandwidth
– Unlicensed users constrained to a few overloaded bands
#16 5
Increasing Demand
200 Million
Subscribers!
Source:
CTIA
#16 6
Increasing Demand
1.4 Trillion
Minutes!
Source:
CTIA
#16 7
Apparent Scarcity
• Measurement studies have shown that in both the
time and frequency domains that spectrum is
underutilized
Spectrum
Holes
Spectrum measurement across the 900 kHz –1 GHz band (Lawrence, KS, USA)
#16 8
Potential Solution
• Dynamic Spectrum Access (DSA)
Fill with
secondary
users
Spectrum measurement across the 900 kHz –1 GHz band (Lawrence, KS, USA)
#16 9
But not in my spectrum!
• Incumbent license holders are very concerned
about co-existing transmissions from unlicensed
users
– Large-scale investments in developing
communication infrastructure around spectrum
• Maintain quality-of-service to its paying customers
– Unlicensed users providing competing services
(e.g., VoIP) but without the large-scale
investment
– Transmissions are a time-varying phenomena … a
signal not interfering at one point in time may
do so at another
#16 10
Example
• Conclusion: Wireless equipment designed for DSA
communications must be rapidly reconfigurable and
spectrum-aware
#16 11
Presentation Overview
•
•
•
•
•
•
Motivation
What are Cognitive Radios?
How are they “cognitive”?
Agile Transmission
Kansas University Agile Radio (KUAR)
Conclusion
#16 12
Software-Defined Radios
• Rapid evolution of microelectronics over
the past several decades
• Wireless transceivers are becoming more
versatile, powerful, and portable
• These advancements have given rise to
Software-Defined Radio (SDR) technology
– Baseband radio functions can be entirely
implemented in digital logic and software
#16 13
Software-Defined Radios
Radio functions performed in
the software domain
#16 14
What is a Cognitive Radio?
“Cognitive radio is an intelligent wireless communication system that is
aware of its surrounding environment (i.e., outside world), and uses the
methodology of understanding-by-building to learn from the environment
and adapt its internal states to statistical variations in the incoming RF
stimuli by making corresponding changes in certain operating parameters
(e.g., transmit-power, carrier-frequency, and modulation strategy) in realtime, with two primary objectives in mind:
• highly reliable communications whenever and wherever needed;
• efficient utilization of the radio spectrum.”
S. Haykin, “Cognitive Radio: Brain-Empowered Wireless
Communications”, IEEE J-SAC, Feb. 2005.
#16 15
What is a Cognitive Radio?
• An intelligent wireless communications system
• Based on SDR technology
– Reconfigurable
– Agile Functionality
• Aware of its environment
– RF spectrum occupancy
– Network traffic
– Transmission quality
• Learns from its environment and adapts to new
scenarios based on previous experiences
#16 16
Presentation Overview
•
•
•
•
•
•
Motivation
What are Cognitive Radios?
How are they “cognitive”?
Agile Transmission
Kansas University Agile Radio (KUAR)
Conclusion
#16 17
Cognition Framework
• Distinction between reconfigurability and
adaptability
• Reconfigurability
– Involves choosing radio building blocks
– Choice of blocks lasts for relatively long period of time
– Requires “flashing” of programmable logic
• Adaptability
– Fine-tunes radio operating parameters
– Parameter choices last for a short period of time
– Does not require “flashing” of programmable logic
#16 18
Cognition Framework
Basic schematic of the cognition component of a cognitive radio
#16 19
Reconfigurability
• Given several desired radio requirements, determine bestpossible choices for radio components
#16 20
Adaptation in Cognitive Radios
Cognitive adaptation module possessing several knobs and dials
#16 21
AI-Based Adaptation
• Genetic Algorithms (GA)
– Biologically-inspired technique used typically for
problems with large parameter spaces
– Execution time becomes larger as number of operational
and environmental parameters grows
– Does not require much memory to run; requires long
execution time
• Expert Systems
– Decisions determined offline and stored in radio memory
– Decision making time is very fast
– Interesting trade-off exists between rule base size and
the efficiency of decision
#16 22
Example: GA Convergence
Converges
to an
overall
fitness
score of
0.8
GA Convergence for a cognitive radio operating in emergency mode
T. R. Newman et al., “Cognitive Engine Implementation for
Wireless Multicarrier Transceivers”, To appear in the Wiley
Wireless Communications and Mobile Computing Journal, 2007.
#16 23
Example: GA Solution
Subcarrier channel attenuation, throughput, and transmit power levels
T. R. Newman et al., “Cognitive Engine Implementation for
Wireless Multicarrier Transceivers”, To appear in the Wiley
Wireless Communications and Mobile Computing Journal, 2007.
#16 24
Presentation Overview
•
•
•
•
•
•
Motivation
What are Cognitive Radios?
How are they “cognitive”?
Agile Transmission
Kansas University Agile Radio (KUAR)
Conclusion
#16 25
Transmission Approaches for DSA
• Transmission in licensed spectrum
classified into three categories
– Cooperative Approach
• Primary and secondary users coordinate with each
other regarding spectrum usage
– Underlay Approach
• Secondary signals transmitted at very low power
spectral density; undetected by primary users
• e.g., ultra wideband (UWB)
– Overlay Systems
• Secondary signals fill in the spectrum unoccupied by
primary users
#16 26
NC-OFDM Transmission
• Based on conventional orthogonal frequency
division multiplexing (OFDM)
• Uses spectrum sensing measurements to
“turn off” potentially interfering
subcarriers
#16 27
FFT-Pruning for NC-OFDM
Pruning an FFT employed in an NC-OFDM Transceiver
R. Rajbanshi et al., “An Efficient Implementation of NC-OFDM
Transceivers for Cognitive Radios”, Proc. CrownCom, June. 2006.
#16 28
Example: FFT Execution Time
Mean execution times for a 1024-point FFT
R. Rajbanshi et al., “An Efficient Implementation of NC-OFDM
Transceivers for Cognitive Radios”, Proc. CrownCom, June. 2006.
#16 29
Presentation Overview
•
•
•
•
•
•
Motivation
What are Cognitive Radios?
How are they “cognitive”?
Agile Transmission
Kansas University Agile Radio (KUAR)
Conclusion
#16 30
KUAR
•
•
•
•
Programmable, agile radio platform
for networking (and other) research
Enabled by support from NSF and
DARPA
Flexible foundation for experimental
research
– Agile platform for research at
physical, link, MAC layers
– Capability to sense and act across
layers
– Enables building new network
architectures
Evolving into a cognitive radio
platform
– Provide sufficient computing
resources for cognition experiments
Front view of a KUAR unit
#16 31
KUAR Team
• Principal Investigators
– Gary J. Minden, Joseph B. Evans
• Investigators
– Arvin Agah, James Roberts, Alexander M. Wyglinski
• Design Engineers
– Leon Searl, Dan DePardo
• Graduate Research Assistants
– Rakesh Rajbanshi, Qi Chen, Tim Newman, Rory Petty,
Ted Weidling, Brett Barker, Jordan Guffey, Dinesh
Datla,
Levi Pierce, Megan Lehnherr, Brian Cordill
#16 32
KUAR Schematic
#16 33
KUAR RF and Digital Boards
• RF Board
– Frequency Range: 5.25 – 5.85 GHz (includes UNII and
ISM bands)
– SW controls Tx Power, Rx Front-end attenuation and IF
gain
– 30 MHz Baseband Bandwidth
• Digital Board
– PC employing industry standard COMeXpress formfactor
• Pentium-M @ 1.4GHz, 1 GB SDRAM, 6GB CF+ Disk
– FPGA: Xilinx Virtex II Pro P30
• FPGA External Memory: 4 Mb SRAM
– Dual ADC (14 bits parallel, 105 MSPS)
– Dual DAC (16-bits parallel, 160/400 MSPS)
#16 34
KUAR Software/Firmware
• PC runs Linux 2.6 kernel
• Software measures radio power usage
• Radio Net scripts automate multi-radio
experiments
• KUAR Radio Systems
– BPSK with phase and timing recovery
– Multi-carrier demo
• KUAR VHDL components:
– Energy Detector, Digital Sampler, Absolute
Value, Clocks, Sin Generators, Control
Processor, Bus Utilities, Delay, Register
controls, etc…
#16 35
KUAR System Diagram
System diagram of a KUAR unit (Version 3.0)
#16 36
KUAR Transmit Performance
#16 37
KUAR Receiver Eye Diagram
#16 38
Presentation Overview
•
•
•
•
•
•
Motivation
What are Cognitive Radios?
How are they “cognitive”?
Agile Transmission
Kansas University Agile Radio (KUAR)
Conclusion
#16 39
Conclusion
• DSA approach to spectrum management is
a reality
– FCC Proposed Rule-Making with respect to TV
bands
• Cognitive Radios can help us realize DSA
networks
– Increased spectral efficiency
– Enhanced transmission performance
• Much work still required before deploying
reliable DSA networks
– Continue work on developing communication
techniques that enable DSA
#16 40
References
•
•
•
•
•
S. Haykin, “Cognitive Radio: Brain-Empowered Wireless
Communications”, IEEE Journal on Selected Areas in
Communications, Feb. 2005.
William Krenik and Anuj Batra, “Cognitive Radio Techniques from
Wide Area Networks”, Proceedings of the 42nd Design Automation
Conference, pages 409-412, 2005.
Upcoming May 2007 Issue of the IEEE Communications Magazine
(Feature Topic on Cognitive Radios for Dynamic Spectrum Access)
KUAR Wiki: https://agileradio.ittc.ku.edu/
DARPA XG Website:
http://www.darpa.mil/ATO/programs/XG/index.htm
• T. R. Newman et al., “Cognitive Engine Implementation for
Wireless Multicarrier Transceivers”, To appear in the Wiley
Wireless Communications and Mobile Computing Journal,
2007
• R. Rajbanshi et al., “An Efficient Implementation of NCOFDM Transceivers for Cognitive Radios”, Proc. CrownCom,
June. 2006.
#16 41
Additional Slides
#16 42
Current Spectrum Allocation
• “Command-and-control” Approach
– License holders maintain exclusive rights to their
allocated spectrum
• Purchased during a spectrum auction, e.g., 3G auctions
• Allocated via government decree, e.g., military, television
– Unlicensed devices not permitted to transmit in licensed
bands
• Allocated unlicensed bands (with transmit constraints)
– Industrial, Scientific, Medical (ISM) bands
» 900 MHz, 1.8 GHz, 2.4 GHz, 5.8 GHz
– Unlicensed National Information Infrastructure (UNII) band
» 5.15 GHz – 5.825 GHz
#16 43
Spectrum Sensing
• Required by agile modulation process
• Classification of spectrum into either signal or
noise
– Recursive One-Sided Hypothesis Testing (ROHT)
recursively performs hypothesis test on the
measurement data and classifies a portion of data as
signal
– Otsu’s algorithm segments data into 2 classes to achieve
maximum separation between classes
– Adaptive thresholding uses a sliding window approach
that classifies blocks of data separately and then
combine the classification results
#16 44
Channel Sounding
• Need to identify spectrum worth transmitting
across
– Unoccupied spectrum may be severely attenuated
• Simultaneously, sounding process cannot interfere
with signals from primary users
– Sounding a large bandwidth with several primary users
requires the power spectral density to be low
• Adapt current sounding techniques to DSA
scenario
– Swept Time Delay Cross-Correlator (STDCC)
#16 45
KUAR RF Board and Antennas
• TX and RX Active Antennas
– 5.250 - 5.850 GHz, -100 dBm min Rx, +25 dBm max Tx
– Independent Tx and Rx antennas & frequencies
• RF Board
– Frequency Range: 5.25 – 5.85 GHz (includes UNII and ISM
bands)
– SW controls Tx Power, Rx Front-end attenuation and IF gain
• Useful for fading channel experiments
• Accommodates variety of experiments and test environments
– Superheterodyne Hybrid direct conversion
• IF range of 1.85 – 2.45 GHz controlled by SW
• Quadrature Direct conversion between baseband and IF
– 30 MHz Baseband Bandwidth
– Microcontroller converts Digital Board I2C bus to RF device
SPI bus, control/status lines
#16 46
KUAR Digital Board
• PC in industry standard COMeXpress form-factor
– Pentium-M @ 1.4GHz
– 1 GB SDRAM
– 6GB CF+ Disk
• FPGA: Xilinx Virtex II Pro P30
– FPGA External Memory: 4 Mb SRAM
– PC<->FPGA Buses: PCI Express / PCI / USB->Parallel
• USB controller or PC software programs FPGA
• Dual ADC (14 bits parallel, 105 MSPS)
• Dual DAC (16-bits parallel, 160/400 MSPS)
#16 47
KUAR Software/Firmware
•
•
•
•
•
PC runs Linux 2.6 kernel
FPGA firmware registers addressable as PCI registers
Software measures radio power usage
Radio Net scripts automate multi-radio experiments
KUAR Radio Systems
– BPSK with phase and timing recovery LFR-QPSK
– Multi-carrier Demo
• KUAR VHDL components:
– Energy Detector, Digital Sampler, Absolute Value, Clocks, Sin
Generators, Control Processor, Bus Utilities, Delay, Register
controls, etc…
• RF board configuration through RFControl API
#16 48