Tata AIG General Insurance Company - Insurer Innovation Award 2024
Cr2012b
1. Cognitive Radio and its RF Challenges
Markku Renfors
Tampere University of Technology
Finland
2. Contents
1. Dynamic spectrum access and cognitive radio ideas
2. Elements of cognitive radio systems
3. Filter bank approach for cognitive radio physical layer
– Spectrum agility
4. Spectrum sensing
– Energy detection based spectrum sensing
– Wideband sensing using FFT or filter banks
5. RF challenges in cognitive radio
3. Dynamic spectrum access and
cognitive radio techniques
• Conventionally, the use of radio frequency bands has been
regulated through spectrum allocations with licensing procedure.
– However, measurements on the licensed bands show severe temporal
and / or spatial underutilization of the assigned spectral resources.
– This imposes a great challenge for future wireless communications
which attempts to satisfy the ever growing demands for new services
and ubiquitous broadband wireless access.
• In order to solve the imbalance between spectrum shortage and
spectrum underutilization an innovative spectrum access strategy
called spectrum pooling has been visioned.
• A concept of cognitive radio (CR) has been proposed as a
promising technology to fulfill the unique requirements of
intelligence and spectrum agility necessary for succesful
deployment of such dynamic spectrum access.
4. Dynamic spectrum access and
cognitive radio techniques
• Spectrum pooling
– Allows opportunistic secondary (unlicenced) access to spectral
resources unused by their primary (licensed) owners.
Significant improvement in spectrum utilization.
– Secondary transmission must avoid any harmful interference to
primary systems.
CRs have to regularly perform reliable radio scene analysis to
detect the presence of primary user signals with high detection
and low false alarm probability.
• Cognitive radio
– Smart and spectrally flexible radio (secondary user device, SU) that
monitors and senses its radio environment for potential spectrum
opportunities.
5. Cognitive radio
Definition by FCC:
“A radio or system that senses its operational electromagnetic
environment and can dynamically and autonomously adjust its radio
operating parameters to modify system operation, such as maximize
throughput, mitigate interference, facilitate interoperability, access
secondary markets.”
6. Dynamic spectrum access and
cognitive radio techniques
Concept of opportunistic spectrum sharing: secondary
utilization of the identified spectrum holes.
• Concepts of a spectrum hole
and opportunistic spectrum
sharing:
Spectral opportunity for secondary
access: a spectrum hole.
7. Contents
1. Dynamic spectrum access and cognitive radio ideas
2. Elements of cognitive radio systems
3. Filter bank approach for cognitive radio physical layer
– Spectrum agility
4. Spectrum sensing
– Energy detection based spectrum sensing
– Wideband sensing using FFT or filter banks
5. RF challenges in cognitive radio
8. Cognitive radio: Definitions
Spectrum gap, spectrum hole, white space
Spatially & temporally unused part of the radio spectrum, which is
considered for use by CR
Primary user (PU)
Licensed user/privileged user of a frequency band
Secondary user (SU)
Opportunistic user of a frequency band
Waveform
Signal model & other essential characteristics of the (PU or SU)
communication system
9. Types of CR systems
• Underlay systems use wide bandwidth and low-enough power spectral density
not to disturb PUs.
– Spread spectrum
– UWB
• Interweave CR is based on identification of spectral holes.
• Overlay: Here the SU RX and TX are assumed to know and utilize info about
the PU signal
– Example: SU TX sends a signal which consists of (1) relayed PU signal and (2) SU
signal, possibly with special coding (e.g. dirty paper coding). The relayed PU signal is
included to guarantee sufficient SNR for nearby PU receivers. SU receiver detects
both PU signal and SU signal, using interference cancellation techniques.
• Black space systems: Transmitting at relatively low power level on top of
powerful PUs. Detection based on interference cancellation, after detecting PU
data.
– Like overlay, but no PU relaying needed
• The focus in the following discussions is on interweave systems.
10. Cognitive radio functionalities
• Higher layers
– Cognitive cycle, adapting to the environment, learning/predicting the PU
characteristics and spectrum usage patterns; game-theoretic approaches
– Real-time spectrum markets …
• Reliable information about primary usage
– Cognitive pilot channel
– Data base of primary users (PUs) & positioning of seconday user (SU)
stations
– Spectrum sensing
• Dynamic spectrum access (DSA) scheme
• Spectrum exploitation
– Flexible, spectrum agile waveforms and signal processing for secondary
communications
– Software defined radio technologies are essential in this context
11. Knowledge/detection of primary usage - 1
• Cognitive pilot channel
– Dedicated control interface ideally to all local users of the radio channel
– Could provide explicit information about primary usage
– In the first stage, expected to be used as a means for closer coordination of
wireles networks using existing technologies (e.g., sharing the same
frequency band for GSM/WCMDA/HSPA/LTE/WiMAX/... waveforms in a
flexible way)
– Means for exchanging information between spectrum sensing stations
(cooperative sensing)
• Data base of primary users & positioning of seconday user stations
– Database maintainance is critical
– Central element in the first CR standard IEEE 802.22, along with spectrum
sensing
12. Knowledge/detection of primary usage - 2
• Spectrum sensing
– Sensing the radio environment to detect on-going primary transmissions
Idea is to detect spectral holes as opportunities for secondary
transmissions
– Cooperative sensing by multiple stations helps to mitigate problems, e.g.,
due to shadow fading.
Reliable detection of primaries in a single mobile station is an extremely
challenging task
– Minimum infrastructure needed, suitable for ad-hoc networks.
– Most challenging scheme to implement from the RF point of view
– Discussed later in some more details
13. Spectrum sensing in general
• Occupancy sensing
– ‘white’ spaces or spectral holes
– ‘grey’ spaces
– ’black’ spaces
• Methods
– (matched filtering)
– energy-based detection (radiometer)
– feature-based detection
• CP autocorrelation for OFDM primaries
• Cyclo-stationary detection
• Covariance & eigenvalue based methods (within sample sequence
and/or between antennas)
• Many others
• Fixed sample size vs. sequential detection principles
• Cooperative sensing
14. Dynamic spectrum access scheme
• MAC protocol for the secondary user system
– Usually expected to support, in the same region, multiple independent SU
systems
– Cognitive pilots support coordinated use of the same resources by multiple
operators / SU systems
– Opportunistic dynamic spectrum access (DSA) for ad-hoc type operation
• Distributed coordination, usually no central control element between
different SU systems is assumed.
• Basically different SU systems are competing for the spectral resources
• To make the idea sensible, some form of distributed co-operation is
needed (e.g., good neighbour strategy, fairness)
• Game theory & other advanced concepts often considered in the
developments.
15. Spectrum agile waveforms for
secondary communications
• Very flexible waveforms are needed
– Waveforms need to be adapted to the radio environment
– Adjustable center frequency, bandwidth, SNR requirements, etc.
– Fragmented spectrum use is of interest: The overall transmission band may
consist of multiple non-adjacent frequency slots
– Well-contained spectrum for high spectral efficiency without leakage to
adjacent PU frequency channels.
• Multicarrier modulation has many of the desired features
16. Contents
1. Dynamic spectrum access and cognitive radio ideas
2. Elements of cognitive radio systems
3. Filter bank approach for cognitive radio physical layer
– Spectrum agility
4. Spectrum sensing
– Energy detection based spectrum sensing
– Wideband sensing using FFT or filter banks
5. RF challenges in cognitive radio
17. Multicarrier modulation: OFDM and FBMC
CP-OFDM
Simple and robust.
Spectrum leakage to adjacent subcarriers is a problem, e.g., in non-contiguous
OFDM for fragmented spectrum use.
o Various attempts for reducing the sidelobes can be found in the literature
Filter bank multicarrier (FBMC)
Improved spectral efficiency due to lack of CPs and reduced guardbands
Very good spectral containment → reduced spectral leakage
Possibility to non-synchronized transmission of different groups of subcarriers
with relatively small loss in spectrum efficiency
18. Spectral containment of FBMC vs. OFDM
Frequency response of a subchannel filter
= Transmited subcarrier spectrum
• OFDM:
• FBMC:
19. Filter banks
• We consider efficient uniform, modulation-based filter banks,
where subchannel frequency responses are obtained as
frequency-shifted versions of a prototype.
• In our designs, the overall subchannel bandwidth (with transition
bands) is 2 x subchannel spacing (i.e., roll-off factor = 1).
– Only immediately adjacent subchannels are significantly interacting.
– One unused subcarrier is sufficient as a guard-band.
6 7 8 9 10 11 12 13 14
-80
-70
-60
-50
-40
-30
-20
-10
0
10
Subchannel index
AmplitudeindB
20. Filter bank structure
• Reduced guardbands between users
• No CP's
Improved spectral efficiency
• The transmultiplexer system achieves nearly perfect reconstruction.
Distortion is small compared to noise in practical SNR range with ideal
channel.
6 7 8 9 10 11 12 13 14
-80
-70
-60
-50
-40
-30
-20
-10
0
10
Subchannel index
AmplitudeindB
21. Transmultiplexer system model
OQAMpre-processing
+
2M↑ 0 ( )G z
1( )G z
1( )MG z−
2M↑
2M↑
Synthesis filter bank
2M↓
0 ( )F z
1( )F z
1( )MF z− 2M↓
2M↓
Analysis filter bank
OQAMpost-processing
D
z−
0
ˆ ( )X z
1
ˆ ( )X z
1
ˆ ( )MX z−
0 ( )X z
1( )X z
1( )MX z−
( )Y z
To achieve orthogonality in a spectrally efficient manner, offset-QAM signal
model is crucial.
Each QAM symbol is mapped to two consecutive subcarrier samples.
Subcarrier sample sequences are oversampled by a factor of 2.
22. Receiver side: Efficient polyphase structure for
analysis filter bank
2
0 ( )B z
2
1( )B z
2
1( )MB z−
FFT
1−
z
1−
z
2M↓
2M↓
2M↓
×
×
×
Subchannel
processing
Subchannel
processing
Subchannel
processing
*
0,nβ
*
1,nβ
*
1,M nβ −
×
*
0,nθ
×
*
1,nθ
Re
×
*
1,M nθ −
0,nd
1,nd
1,M nd −
Re
Re
Analysis filter bank OQAM post-processing
kR2C
kR2C
kR2C
Proper subchannel processing restores the orthogonality of subcarriers in case of
frequency-selective channels.
- Synchronization & channel equalization
- 2x oversampled subcarrier processing => Fractionally spaced equalization
23. Prototype filter design based on frequency
sampling approach
• High frequency selectivity
• Exact stopband zeros
6 7 8 9 10 11 12 13 14
-80
-70
-60
-50
-40
-30
-20
-10
0
10
Subchannel index
AmplitudeindB
24. Spectrum efficiency, FBMC vs. OFDM
• Link-level simulations with FEC & HARQ
• (PHYDYAS (FP7-ICT-2007-211887) , TUT & ALUD):
25. PHYDYAS:
FBMC based cognitive radio physical layer
• The idea of FBMC has existed since the 60'ies, but still there are signal
processing and communication theoretic issues which are not mature
enough for practical use:
– Synchronization
– Channel estimation and equalization
– Adaptation to multi-antenna & MIMO concepts
– Multiple access specifics
– etc.
• EU FP7 project PHYDYAS project is focusing on these open topics.
– WiMAX -like system concept as the starting point
– Special focus on dynamic spectrum use and cognitive radio
28. Application of non-contiguous multicarrier:
Broadband – narrowband co-existence
Professional Mobile Radio (PMR), Tetra family evolution
BB
UE#2 BB
UE#3
BB
UE#4
BB
UE#1
Reserved for narrow-
band network
LTE channel bandwidth
Frequency
29. Fast convolution based flexible channelization filter
bank
• Supports different waveforms in
different channels
- Single-carrier
- FBMC
- Filtered multitone
• See [Renfors ECCTD2011]
Independent tunability of the center frequencies of individual
subchannels or multiplexes
o Individual frequency offset compensation of different users’ signals.
o Individual timing offset compensation.
o Possibility to process non-synchronized groups of subchannels in a
single filter bank.
30. Multiuser synchronization aspects - 1
OFDM assumes quasi-synchronous
operation of different transmissions
participating in a multiplex (e.g., uplink):
– Delay spread + timing uncertainty within CP
– CFOs < 0.01∆f
– Tight base-station control required
– Tight time-domain
multiplexing is possible
32. FBMC as cognitive radio physical layer
Advantages:
• Spectral efficiency: no CP's, one empty subchannel is sufficient as a guard-
band to isolate different secondary users.
• The same filter bank can be used for receiver data signal processing and
flexible, high-resolution spectrum sensing with high dynamic range.
The possibility of simultaneous sensing and reception (at different parts of
the spectrum) facilitates the coexistence of primary and secondary users.
• Filter bank of an FBMC receiver as a part of the decimation filtering
chain => very flexible channelization
• Reduced synchronization requirements; possibility to asynchronous operation
with high spectral efficiency
Challenges:
• High linearity for transmitter power amplifier needed to maintain the clean
spectrum provided by the synthesis filter bank.
• Filter bank impulse response "tails" (i.e., time-domain overlap of subcarrier
symbols) introduce overhead in tightly
time-multiplexed operation.
33. Contents
1. Dynamic spectrum access and cognitive radio ideas
2. Elements of cognitive radio systems
3. Filter bank approach for cognitive radio physical layer
– Spectrum agility
4. Spectrum sensing
– Energy detection based spectrum sensing
– Wideband sensing using FFT or filter banks
5. RF challenges in cognitive radio
34. Spectrum sensing specifications
• Frequency resolution: subchannel spacing
– smallest spectral hole
– spectral granularity for transmission
• Spectral dynamic range: > 50 dB
• Noise floor: thermal noise + interference
• Out-of-band interference rejection performance of spectrum analyzer:
> 80 dB
• Sensing latency
Max signal level (SU & PU)
SU sensitivity level
Noise level
Spectrum sensing sensitivity
RX interference level
50 dB
80 dB
Dynamic range is significantly larger
than in data reception!
Important to understand the effect of
RF imperfections on different sensing
methods
Exemplary numeric values!
35. Energy detection for spectrum sensing
0
1
[ ], :noise only
[ ]
[ ] [ ], :signal present
n l H
Y l
s l n l H
=
+
( ) 0( | )FAP P T Hγ= >Y
0.8 1 1.2 1.4 1.6 1.8 2 2.2
0
100
200
300
400
500
600
700
Energy detector: P
FA
= 0.1, N = 100, σ
n
2
= 1, SNR = -3.0103 dB
γH0
H
1
P
MD
P
FA
PD
= 1-PMD
fY|H
0
(y|H0
) f
Y|H
1
(y|H
1
)
Sensing decision is a binary
hypothesis testing problem:
Test statistic:
( ) 1( | )MDP P T Hγ= <Y
Probability of false alarm:
Probability of missed detection:
Interference to primary system.
Lost secondary opportunity.
36. 36
• The actual PDF’s are chi-square or gamma distributions
• Gaussian approximation is usually accurate enough (but care must
be exercised). For a complex vector of N samples, the distributions in
the absence and presence of the PU signal are:
• P: signal variance; σ2: noise variance.
Energy detection for spectrum sensing
0.8 1 1.2 1.4 1.6 1.8 2 2.2
0
100
200
300
400
500
600
700
Energy detector: PFA
= 0.1, N = 100, σn
2
= 1, SNR = -3.0103 dB
γH
0
H
1
PMD
PFA
P
D
= 1-P
MD
fY|H
0
(y|H
0
) f
Y|H
1
(y|H
1
)
37. Sensing in the presence of noise uncertainty
-40 -35 -30 -25 -20 -15 -10 -5 0
0
2
4
6
8
10
12
14
16
SNR in dB
log
10
N
Sample complexity (N) of the radiometer under noise uncertainty
x = 0.001 dB
x = 0.1 dB
x = 1 dB
1 1 2
2
2[ ( ) (1 )(1 )]FA MDP P SNR
N
SNR
− −
Φ − Φ − +
=
1 1 2
2
2[ ( ) (1 )]
1
( )
FA MDP P
N
SNR ρ
ρ
− −
Φ − Φ −
≈
− −
2
1energy
wallSNR
ρ
ρ
−
=
[1] R. Tandra and A. Sahai, ”SNR Walls for Signal Detection,” IEEE J. Selected Topics in
Signal Processing, vol. 2, No. 1, Feb. 2008
The sensing time depends on the
primary signal SNR and the noise
uncertainty ρ (x in dB units).
The noise uncertainty introduces
an SNR wall in energy detection
[1]:
38. About noise estimation/calibration
• Switching off antenna during calibration.
– Various limitations …
• If there is a good reason to believe that the channel is initially free of PUs, then
we can get a reliable noise reference for spectrum monitoring
– For example, using more powerful methods in initial sensing, and energy detection
for monitoring.
• With frequency hopping PU systems, we should be able to obtain observations
in the presence and absence of a transmission burst.
• Noise only vs. noise + interference
– Interference, e.g., from adjacent channels due to TX PA nonlinearity or RX
nonlinearity.
– How to distinguish interference due to adjacent channels from a co-channel PU?
39. Wideband/multichannel spectrum sensing
Alternative approaches for analysing the spectrum scene:
• Narrowband (per frequency channel) frequency scanning receiver
– Simple to implement
– Smaller RX dynamic range => less critical to RF imperfections
– Relatively slow process when targeting at high sensitivity
• Wideband receiver with DSP-based spectrum analysis functionality
– Fast: Sensing simultaneously for a high number of possibly available spectral slots.
– Frequency resolution determined by subband spacing.
– Integration over multiple subbands in case of wideband PUs.
– Wideband, high dynamic range ADC needed
– Sensitive to RF imperfections
– FFT / windowed FFT / filter bank approaches
40. Commonality of wideband spectrum sensing and
Spectrum agile waveforms
• Spectrum sensing will be an important element in future networks.
• There is also need for devices which can be used for both sensing and
data reception.
Commonality of sensing and data reception functions is important.
Similar requirements for spectral agility.
• Basically, we need a highly configurable filter bank to do wideband
spectrum sensing, simultaneously for a high number of possibly
available spectral slots.
• Multicarrier techniques can provide the needed commonality and
configurability.
OFDM is the common choice
- Limited spectral containment introduces various challenges for both
sensing and transmission functions
Filter bank based multicarrier (FBMC) techniques have some very
interesting characteristics.
41. FBMC receiver as energy detector
-1 -0.5 0 0.5 1
10
-5
10
-4
10
-3
10
-2
10
-1
10
0
Normalized frequency [ω/π]
PSD
OFDM
FBMCPU1
PU3
PU2
SHSH
• Here, classical energy
detection is considered.
• M = 1024 subchannels.
• FBMC receiver is not blinded
by the presence of high level
neighboring signals and is
able to identify accurately the
spectrum holes.
42. 42
Multiband spectrum sensing
• High flexibility:
– Energy measurements for each subcarrier symbol.
– Summation over used time-frequency window(s).
– The total number of statistically independent (maximally decimated)
samples determines sensitivity, together with target PFA & PMD.
• Multiple-dwell approaches easy to implement using different window
sizes:
– Fast reaction to new strong PU signals using short integration.
– Fast detection of possible white spaces; longer integration to reach
adequate PFA & PMD.
Symbol index
Subchannelindex
Moderate resolution,
moderate latency
High resolution, high latency
Low resolution, low latency
2
,
,∑=
nk
nkrT
Analysis through FFT
or filter bank
43. Sensing time vs. sensing bandwidth
• Sensing time as a function of bandwidth for different primary signal SNR's:
• With higher SNR's, high bandwidths, and high number of subcarriers, the
sensing time is only a few multicarrier symbols, and filter bank impulse
response length becomes the limiting factor.
0 1 2 3 4 5 6 7
x 10
5
10
-4
10
-3
10
-2
10
-1
10
0
Sensing bandwidth (in Hz)
Sensingtime(inseconds) M = 2048
M = 512
M = 128
M = 32x = 0,1 dB
SNR = - 3 dB
SNR = - 12 dB
SNR = - 6 dB
No uncertainty
0.1
0.01
FA
MD
P
P
=
=
44. A practical WLAN scenario - 1
Two WLAN signals spectra in 2.4 GHz ISM band with
smallest spectral gap of 3 MHz.
• We can see the ideal OFDM
signal spectrum, which has a
deep hole in the considered 3
MHz frequency band.
• In the worst case situation
allowed by the 802.11g
specifications, the power
spectral density in the gap can
be at about -20 dBr in the
considered case.
• Figure shows also a third case
with modest spectral regrowth
at -30 dBr level.
45. A practical WLAN scenario - 2
Actual false alarm probability of WLAN signals with target PFA=0.1 for 3
MHz sensing bandwidth in AWGN case:
46. Contents
1. Dynamic spectrum access and cognitive radio ideas
2. Elements of cognitive radio systems
3. Filter bank approach for cognitive radio physical layer
– Spectrum agility
4. Spectrum sensing
– Energy detection based spectrum sensing
– Wideband sensing using FFT or filter banks
5. RF challenges in cognitive radio
47. RF challenges in CR
• Basically the CR idea, in the most aggressive form, leads to extremely
high dynamic range requirement for the receiver
In traditional RF systems, the nearby adjacent channels and blockers are
assumed to be at a reasonable level from the RF implementation point of
view.
In the CR case, there are no specifications for spectral components close to
the potential white spaces in the sensing phase or next to the frequency slot
to be used for communication once it has been determined to be empty.
• Due to the requirement of flexibility and spectral agility, advanced SDR
technologies are expected to be used for the RF functions
Wideband sampling with the related ADC-dynamic range and sampling jitter
issues.
Also the effect of all other RF imparements becomes more critical.
• High spectral purity requirement for the transmitter
SU’s spectral leakage would introduce interference to nearby primary
transmissions.
PU’s spectral leakage destroys SU’s transmission
51. Possible ways to handle RF challenges in CR
• Co-operative sensing
• Beam-forming
• Spectral awareness
• DSP-enhanced RF techniques
52. Reducing the spectral dynamic range
observed by CR
Radio resource management (RRM)
– Cellular mobile radio systems have effective RRM functions, which limit the power level
differences in adjacent frequency channels
– Due to the distributed nature in opportunistic DSA, we cannot expect such effective
control.
• It will be a great challenge for CR system to reach the capacity/efficiency of a
cellular mobile system! Also the spectral dynamic range is not so well under control.
In multiantenna stations, beamforming is an efficient way to control interferences.
– Should be implemented on the analog side, in order to relax the ADC dynamic range
requirements.
In cooperative sensing, dynamic range requirements are reduced
Overlay CR
Robust waveforms, like MC-CDMA
53. Coping with RF imperfections
• Choice of receiver architecture
Wideband multichannel receiver inevitably needs high dynamic range
• DSP-assisted calibration of analog circuits
• Interference mitigation => DSP enhanced radio, ‘dirty RF’
• PA nonlinearities (TX)
• I/Q imbalance (TX&RX)
• oscillator phase noise (TX&RX)
• timing jitter in IF/RF sampling (RX)
• mixer, LNA and ADC nonlinearities (RX)
54. Co-operative sensing
• Due to large-scale fading (radio shadows), it is very difficult to reach
sufficient sensitivity in a single mobile station or access point
The target sensitivity for single-station spectrum sensing is typically deep
below the thermal noise level.
• In co-operative sensing, the sensing requirement can be relaxed, which
directly reduces the dynamic range requirement.
Enhanced decision statistics
’Sensing diversity’: all stations don’t need to be in good positions to do
sensing
• Further, in co-operative sensing, different stations are in different radio
environments, and some of them might be in particularly good position
for securing each specific spectrum hole.
For example, using sensing stations with high antennas.
How to secure that the local sensing network has sufficient coverage &
sensitivity?
55. Beam-forming
• Beam-forming is an effective tool for
interference management
– In a spectrum sensing station, notches
can be turned in the direction of the
sources of strong interfering spectral
components. The same applies during
data reception.
– Directing the beam towards the actual
receiver reduces the general
interference level.
56. Spectral awareness
• A specific SU system would usually have multiple white spaces at its
disposal. But they might be quite different from the RF requirements
point of view.
Choose the one leading to the easiest requirements.
• Now we see that the radio scene analysis should include, in addition to
identification of spectrum holes, also evaluation of the signal strengths
at the nearby frequencies, etc.
57. Energy efficiency and spectrum awareness
• Energy efficiency is important at all levels of radio implementation
Analog, digital, RF, baseband, HW, SW, …
• Dependancy also on the radio scene:
Highly loaded network → complicated signal processing to maximize
efficient spectrum use
Less congested → relaxed signal processing, e.g., relaxed filter specs →
reduced energy consumption
• What is needed:
Capabilities to operate with very wide dynamic range (say 80…100 dB)
Ability to relax the signal processing algorithms based on the results of the
radio scene analysis
58. References
Survey papers on spectrum sensing:
• Y. Zeng, Y.C. Liang, A. T. Hoang, and R. Zhang, “A Review on Spectrum Sensing for Cognitive Radio:
Challenges and Solutions,” EURASIP Advances in Signal Proc., vol. 2010, pp. 1-15, January 2010.
• T. Yucek and H. Arslan, “A survey of spectrum sensing algorithms for cognitive radio applications,” IEEE
Communications Surveys & Tutorials, vol. 11, no. 1, pp. 116–130, March 2009.
General references:
• J. Mitola, “Cognitive radio for flexible mobile multimedia communications,”in IEEE Int. Workshop Mobile
Multimedia Communications, San Diego, USA, Nov. 1999, pp. 3–10.
• T. A. Weiss and F. K. Jondral, “Spectrum pooling: An innovative strategy for the enhancement of spectrum
efficiency,” IEEE Commun. Mag., vol. 42, pp. S8–S14, Mar. 2004.
• C. R. Stevenson, C. Cordeiro, E. Sofer, and G. Chouinard, “Functional requirements for the 802.22 WRAN
standard,” https://mentor.ieee.org/802.22/file/05/22-05-0007-48-0000-draft-wran-rqmts-doc.doc, November 2006.
• A. Kuzminskiy and Y. Abramovich, “Adaptive antenna array interference mitigation diversity for decentralized
DSA in licence-exempt spectrum”, in Proc. of the Int. Conf. on Communications, Dresden, Germany, June 2009.
• R. Tandra and A. Sahai, “SNR walls for signal detection,” IEEE J. Select. Topics Signal Processing, vol. 2, pp. 4-
17, February 2008.
• B. Farhang-Boroujeny, “Filter bank spectrum sensing for cognitive radios,” IEEE Trans. on Signal Processing,
vol. 56, pp. 1801-1811, May 2008.
• INFSO-ICT-211887 Project PHYDYAS, Deliverables available at: http://www.ict-phydyas.org