Security Issues and Challenges of
Cognitive Radio Network
Dr. L. Murali
Associate Professor
Electronics and Communication Engineering
P.A. College of Engineering and Technology
Pollachi
Keysight Technologies
Cognitive Radio
Algorithm Development
and Testing
A basic cognitive radio
Motivation behind Cognitive Radio
2/22/2019
Filling the whitespace 3
Cognitive Radio Architecture
2/22/2019 4
CR Basic Building block
2/22/2019 5
Challenges and Barriers to Cognitive Radio
Development
Real-world spectral environments.
Difficult in emulating real-world CR spectrum to design and test CR algorithms.
Difficult to test methodically in field environments.
Testing with real-world impairments such as intermodulation distortion and
spurious signals are required.
CR requirements for maximum Performance and
minimum interference
Quick and accurate detection of signals under dynamic channel conditions.
Identifying underutilized spectrum – False Positive Detection.
Making intelligent decisions that maximize radio system performance.
2/22/2019 6
A Paradigm Flow in Algorithm Development
and Testing
• Capture live signals from more locations
• Synthesize signals using simulated
I sources
• Combine captured and synthesized
signals.
II • Test spectrum-sensing algorithms
• Test cognition engine performance.
• Test radio link performance.
III
2/22/2019 7
2/22/2019 8
2/22/2019 9
A Mobile WiMAX CR Whitespace Algorithm
Flow
2/22/2019 10
2/22/2019 Step 1 11
2/22/2019 Step 2,3 and 4 12
Security Threats of Physical Layer in
Cognitive Radio Networks
Primary User Emulation Attack
Objective Function Attack
Learning Attack
Jamming
Eavesdropping
Primary Users’ Location Attack
2/22/2019 13
2/22/2019 14
Results Evaluation Techniques
2/22/2019 15
DESIGNING CR PLATFORMS
The design and deployment of CR platform requires flexible
hardware and software packages.
SDR Software Packages SDR Hardware
GNU Radio Software Universal Software Radio Peripheral (USRP)
IRIS eFalcon
ASGARD COBRA
OSSIE
Hybrid Platforms
WARP
SORA
COgnitive RAdio Learning (CORAL)
Berkeley Emulation Engine (BEE2)
Lyrtech’s small form factor (SFF) SDR
2/22/2019 16
2/22/2019 17
Investigation of Primary User
Emulation Attack in Cognitive
Radio Networks
2/22/2019 18
Outline
Background
Cognitive radio technology
Security issues in cognitive radios
Spectrum sensing in cognitive radios
Primary user emulation attack
Cooperative sensing in the presence of primary user
emulation attack
Cooperative sensing in the presence of PUEA with channel
estimation error
Cooperative sensing with multiple PUE attackers
Cooperative Sensing with multiple antennas in the
presence of PUEA
2/22/2019 19
Wireless Communication Background
Wireless communication system design
higher data rate
larger channel capacity
quality of service and
spectrum utilization efficiency
Security issues
2/22/2019 20
Cognitive Radio Technology
1. Frequency spectrum —— a scarce resource
Frequency allocation chart in US as of 2003
2/22/2019 21
Cognitive Radio Technology
2. Spectrum access is a more significant problem than
spectrum scarcity.
Measurements of spectrum utilization
2/22/2019 22
Cognitive Radio Technology
2/22/2019 23
Cognitive Radio Technology
Main functions
Cognitive
Radios
Spectrum Spectrum Spectrum Spectrum
Sensing Management Sharing Mobility
2/22/2019 24
Security Issues in CR Networks
Spectrum occupancy failures
Policy failures
Location failures
Sensor Failures
Transmitter/Receiver failures
Operating system disconnect
Compromised cooperative CR
Common control channel attacks
2/22/2019 25
Spectrum Sensing in Cognitive Radios
Spectrum sensing
A classical signal detection problem
n(t ) H0
x(t )
hs(t ) n(t ) H1
channel gain
noise
primary signal
2/22/2019 26
Spectrum Sensing in Cognitive Radios
Spectrum sensing methods
Spectrum
sensing
Interference
Transmitter Cooperative
temperature
detection detection
detection
Matched Cyclostationary
Energy
filter
detection detection
detection
2/22/2019 27
Spectrum Sensing in Cognitive Radios
Transmitter detection
1) Matched filter detection
Advantages: Better detection performance and less time to achieve
processing gain
Disadvantages: Priori knowledge of primary signal is required (such as
pilots, preambles or synchronized messages).
2/22/2019 28
Spectrum Sensing in Cognitive Radios
Transmitter detection
2) Energy detection
Decision statistic Y follows Chi-square distribution
2/22/2019 29
Spectrum Sensing in Cognitive Radios
Transmitter detection
2) Energy detection
False alarm probability and detection probability
( M , T / 2)
Pf Pr (Y T | H 0 )
( M )
Pd Pr (Y T | H1 ) Qm ( 2 , 2T )
T is decision threshold
2/22/2019 30
Spectrum Sensing in Cognitive Radios
Transmitter detection
3) Cyclostationary detection
Exploits built-in periodicity of modulated signals
couple with sine wave carriers, hopping sequences,
cyclic prefixes and etc.
Advantages: better performance than energy detection
Disadvantages: more computational complexity and
longer observation time.
2/22/2019 31
Spectrum Sensing in Cognitive Radios
Cooperative detection
Transmitter detection problem
2/22/2019 32
Spectrum Sensing in Cognitive Radios
Cooperative detection
Figure 6. Cooperative detection model
2/22/2019 33
Spectrum Sensing in Cognitive Radios
Cooperative detection
Fusion rules:
Hard combination (1 bit): AND rule, OR rule, majority
rule …
Soft combination (n bits): soft sensing information
(e.g., signal energy) [3].
2/22/2019 34
Spectrum Sensing in Cognitive Radios
Interference temperature detection
Figure 7. Interference temperature detection
2/22/2019 35
Spectrum Sensing in Cognitive Radios
Challenges
Hardware requirement
Hidden primary user problem
Primary users detection in spread spectrum
Detection capability
Decision fusion in cooperative detection
Security issues
2/22/2019 36
Primary User Emulation Attack
An attacker occupies the unused channels
by emitting a signal with similar form as the
primary user’s signal so as to prevent other
secondary users from accessing the vacant
frequency bands.
2/22/2019 37
Primary User Emulation Attack
Detection of PUEA
Distance ratio test & distance difference test
Wald’s sequential probability ratio test
2/22/2019 38
Primary User Emulation Attack
Defense against PUEA
Localization based transmitter verification
procedure
Channel identification
Dogfight and blind dogfight
2/22/2019 39
1. Cooperative Spectrum Sensing in
the Presence of PUEA
System model
PUE attacker Primary user
Primary user
Secondary user
…...
SU1 SU2 SU3 SUN PUE attacker
…...
Channel between primary
and secondary user
Channel between attacker
Fusion Center
and secondary user
2/22/2019 40
Cooperative Spectrum Sensing in the
Presence of PUEA
System model
The signal received by the ith secondary user
at the kth time instant is
yi (k ) Pp hpi (k ) x p (k ) Pm hmi (k ) xm (k ) ni (k ), i 1, 2,..., N
x p (k ) : primary user’s signal with power Pp
xm (k ) : attacker’s signal with power Pm
hpi (k ) : channel gain between primary and ith secondary user
hmi (k ) : channel gain between attacker and ith secondary user
2/22/2019 41
Cooperative Spectrum Sensing in the
Presence of PUEA
System model
The combined signal in the fusion center at the
kth time instant is,
2/22/2019 42
Cooperative Spectrum Sensing in the
Presence of PUEA
System model
When there is a PUEA, i.e., β = 1, the detection
problem is reformulated as,
After energy detector,
2/22/2019 43
Cooperative Spectrum Sensing in the
Presence of PUEA
Optimal combining scheme
Objective: To design optimal weights to
maximize the detection probability under the
constraint of a prefixed false alarm probability
wopt arg max{Pd | Pf }
w
where
w [w1 , w2 ,..., wN ]
wopt [ w1opt , w2opt ,..., wNopt ]
2/22/2019 44
Cooperative Spectrum Sensing in the
Presence of PUEA
Optimal combining scheme
Assumption: Block fading k is omitted in hpi (k ) and hmi (k )
For given h pi and hmi , the combined signal y(k ) is also a
complex Gaussian distributed random variable,
where,
2/22/2019 45
Cooperative Spectrum Sensing in the
Presence of PUEA
Optimal combining scheme
Decision statistic Y is compliant with central chi
square distribution for both H0 and H1,
And Pd and Pf are expressed as,
2/22/2019 46
max
Cooperative Spectrum Sensing in the
Presence of PUEA
Optimal combining scheme
Optimal solution:
max is the largest eigenvalue of 1
2/22/2019 47
Cooperative Spectrum Sensing in the
Presence of PUEA
Optimal combining scheme
Remarks:
1) if Pm = 0,
wopt Pp n2 hpH MRC
2) virtual antenna array
3) average detection probability over fading
channel
Pd Pd ( wopt ) f (hp ) f (hm )dhp dhm
2/22/2019 48
Cooperative Spectrum Sensing in the
Presence of PUEA
Optimal combining scheme
Remarks:
4) acquisition of channel information
a. priori knowledge such as pilots,
synchronization messages, preambles...
b. blind channel estimation
2/22/2019 49
Cooperative Spectrum Sensing in the
Presence of PUEA
Simulation results
1 1
Conventional MRC,no attack,N=2 Conventional MRC, no attack, N=4
Proposed combining scheme,attack,N=2 Proposed combining scheme, attack, N=4
0.9 0.9
Conventional MRC, attack, N=4
Conventional MRC,attack,N=2
No cooperation, no attack No cooperation, no attack
0.8 0.8 No cooperation, attack
No cooperation, attack
0.7 0.7
d
d
Detection probability P
Detection probability P
0.6 0.6
0.5 0.5
0.4 0.4
0.3 0.3
0.2 0.2
0.1 0.1
0 0
-3 -2 -1 0 -3 -2 -1 0
10 10 10 10 10 10 10 10
False alarm probability P f False alarm probability P f
(a) N = 2 (b) N = 4
N is the number of secondary user
2/22/2019 50
Cooperative Spectrum Sensing in the
Presence of PUEA
Simulation results
1
0.9
0.8
0.7
Pm
d
Detection probability P
0.6
0.5 Proposed combining scheme, =0.1 Pp
Proposed combining scheme, =1
0.4 Proposed combining scheme, =10
Conventional MRC, =0.1
0.3
Conventional MRC, =1
Conventional MRC, =10
0.2
0.1
0
-15 -10 -5 0 5 10 15
SNR between primary and secondary user p (dB)
2/22/2019 51
Cooperative Spectrum Sensing in the
Presence of PUEA
Different network scenarios of PUEA for two
users case
PUE attacker Primary user Primary user
SU2
SU2
SU1 SU1
PUE attacker
(a) (b)
PUE attacker Primary user
Primary user
PUE attacker
SU2
SU2
SU1 PUE attacker SU1
(c) (d)
2/22/2019 52
Cooperative Spectrum Sensing in the
Presence of PUEA
Simulation results
1
0.9
0.8
0.7
d
Detection probability P
0.6
0.5
0.4
case a,def ense
0.3
case a,attack
case b,def ense
0.2 case b,attack
case c,def ense
case c,attack
0.1 case d,def ense
case d,attack
0
-3 -2 -1 0
10 10 10 10
False alarm probability P f
2/22/2019 53
2. Cooperative Spectrum Sensing in the
Presence of PUEA with Channel Estimation
Error
System model
PUE attacker Primary user
estimation error Primary user
Secondary user
…...
SU1 SU2 SU3 SUN PUE attacker
…...
Channel between primary
and secondary user
Channel between attacker
Fusion Center
and secondary user
2/22/2019 54
Cooperative Spectrum Sensing in the
Presence of PUEA with Channel
Estimation Error
System model
hˆp hp e p
hˆm hm em
hˆ : estimated channel gain
h : actual channel gain
e : estimation error
2/22/2019 55
Cooperative Spectrum Sensing in the
Presence of PUEA with Channel
Estimation Error
Average detection probability
Pd Pd ( wˆ opt ) f (h p ) f (hm ) f (e p ) f (em )dhp dhm dep dem
2/22/2019 56
Cooperative Spectrum Sensing in the Presence
of PUEA with Channel Estimation Error
Simulation results
1
0.9
0.8
Proposed combining scheme
0.7
d
Detection probability P
0.6
0.5
0.4 Conventional MRC
0.3
Proposed combining scheme, 2em = -15dB
0.2
Proposed combining scheme, 2em = -10dB
0.1 Proposed combining scheme, 2em = -5dB
Conventional MRC
0
-3 -2 -1 0
10 10 10 10
False alarm probability P f
2/22/2019 57
Cooperative Spectrum Sensing in the Presence
of PUEA with Channel Estimation Error
Simulation results
1
2
Proposed combining scheme, ep = -15dB
0.9 2
Proposed combining scheme, ep = -10dB
2
Proposed combining scheme, ep = -5dB
0.8
Conventional MRC, 2 = -15dB
ep
0.7 Conventional MRC, 2 = -10dB
ep
d
Conventional MRC, 2 = -5dB
Detection probability P
ep
0.6
0.5
0.4
0.3 Proposed combining scheme Conventional MRC
0.2
0.1
0
-3 -2 -1 0
10 10 10 10
False alarm probability P f
2/22/2019 58
3. Cooperative Spectrum Sensing in the
Presence of Multiple PUE Attackers
System model
Primary user
PUE attacker 1
PUE attacker 2
Primary user
Secondary user
…...
SU1 SU2 SU3 SUN PUE attacker
…...
Channel between primary
PUE attacker 3 PUE attacker K and secondary user
Channel between attacker
Fusion Center
and secondary user
2/22/2019 59
Cooperative Spectrum Sensing in the
Presence of Multiple PUE Attackers
System model
The signal received by the ith secondary user
at the kth time instant is
K
yi (k ) Pp h pi (k ) x p (k ) Pmj hmij (k ) xmj (k ) ni (k )
j 1
i 1,2,..., N
2/22/2019 60
Cooperative Spectrum Sensing in the
Presence of Multiple PUE Attackers
Optimal weights
2/22/2019 61
Cooperative Spectrum Sensing in the
Presence of Multiple PUE Attackers
Simulation results
1 1
Conventional MRC, no attack Conventional MRC, no attack
0.9 Proposed combining scheme, tw o attackers 0.9 Proposed combining scheme, four attackers
Conventional MRC, tw o attackers Conventional MRC, four attackers
0.8 0.8
0.7 0.7
d
d
Detection probability P
Detection probability P
0.6 0.6
0.5 0.5
0.4 0.4
0.3 0.3
0.2 0.2
0.1 0.1
0 0
-3 -2 -1 0 -3 -2 -1 0
10 10 10 10 10 10 10 10
False alarm probability P f False alarm probaility P f
(a) K = 2 (a) K = 4
2/22/2019 62
Cooperative Spectrum Sensing in the
Presence of Multiple PUE Attackers
Normalized attacking power
1
0.9
0.8
0.7
d
Detection probability P
0.6
0.5
0.4
Proposed combing scheme, two attackers
0.3 Proposed combing scheme, four attackers
Proposed combing scheme, six attackers
Proposed combing scheme, eight attackers
0.2
Conventional MRC, two attackers
Conventional MRC, four attackers
0.1 Conventional MRC, six attackers
Conventional MRC, eight attackers
0
-3 -2 -1 0
10 10 10 10
False alarm probaility P f
2/22/2019 63
4. Cooperative Spectrum Sensing with
Multiple Antennas in the Presence of PUEA
Multiple antenna technology
2/22/2019 64
Cooperative Spectrum Sensing with
Multiple Antennas in the Presence of PUEA
System model
PUE Primary
Attacker User
…...
... ...
Secondary …... Secondary
User 1 User N
…...
Fusion Center
2/22/2019 65
Cooperative Spectrum Sensing with
Multiple Antennas in the Presence of PUEA
System model
The received signal at ith user at the kth detection
instant is,
M
yi (k ) wij Pp hpi (k ) x p (k ) Pm hmi (k ) xm (k ) ni (k ) , i 1,2,..., N
j 1
the final combined signal at the fusion center
is given as,
N M
y(k ) wi wij Pp hpi (k ) x p (k ) Pm hmi (k ) xm (k ) ni (k )
i 1 j 1
2/22/2019 66
Cooperative Spectrum Sensing with
Multiple Antennas in the Presence of PUEA
2/22/2019 67
Cooperative Spectrum Sensing with
Multiple Antennas in the Presence of PUEA
Simulation results
1
0.9
0.8
0.7
d
Detection probability P
0.6
0.5
0.4
0.3
Conventional MRC,no attack, N=4, 1 antenna
0.2 Conventional MRC,no attack, N=2, 2 antennas
Proposed combining scheme,attack,N=4, 1 antenna
Proposed combining scheme,attack,N=2, 2 antennas
0.1
Conventional MRC,attack, N=4, 1 antenna
Conventional MRC,attack, N=2, 2 antennas
0
-3 -2 -1 0
10 10 10 10
False alarm probaility P f
(a) 2 antenna case
2/22/2019 68
2/22/2019 69
