Opportunistic Spectrum
Access in Cognitive Radio
Networks
Project Team:
Z. Ding and X. Liu (co-PIs)
S. Huang and E. Jung (GSR)
University of California, Davis
(Well known) Motivations for
Cognitive Radio Networks
• Spectrum scarcity.
• More wireless services.
• Inefficient static spectrum
allocation.
• Existence of a large
amount of under-utilized
spectrum.
• Advantage of flexible and
cognitive spectrum
access scheme needed:
cognitive radio.
Wireless Sensor Network
AP
Smart House
Public Safety Station tower
AP
Smart House
Wireless Sensor Network
Cellular tower
TV tower
WiMAX Base Station
Smart House
Wireless Sensor Network
Wireless Sensor Network
AP
Smart Car
Smart House
Traditional Static Spectrum Allocation
100MHz
10GHz
Opportunistic Spectrum Access
•
Design Objectives:




Non-intrusiveness
Spectral efficiency
Cost efficiency
Decentralized
Primary User
House
Secondary User
Radio tower
AP
Three basic access schemes
PU Xmit
Virtual Xmit
SU Xmit
Vacation
Sensing Point
Overlapping time
PU:
Collision!
Success
Collision!
Success
Collision!
Success
SU: VX
SU: KS
SU: VAC
PU -- primary user (licensee of the channel)
SU -- secondary user (cognitive ratio)
Problem Formulation
•




Assumptions:
Exponentially distributed idle period
General primary busy period distribution
Perfect sensing
Knowledge of average idle time/busy time
• Constraint Metrics:
 Bounded collision probability
 Bounded overlapping time
max C2
• Optimization problem:
s.t.
P1c   , or,
P1r  
Fundamental limits of opportunistic
spectrum access
• Primary channel with exponentially distributed idle
period
• Bounded collision probability constraints
• Maximum achievable throughput of a secondary user
C2   
 --- collision probability bound
 --- percentage of idle time (by primary users)
Comparison of VX and VAC
Comparison of VX and KS
Observations
• VX, VAC and KS schemes have
indistinguishable throughput performance, under
collision probability constraint;
• The smaller the packet length, the larger the
throughput.
• The result can be extended to systems with
multiple primary users and multiple secondary
users (treat all secondary users as a “super”
secondary user)
Fixed length packet wins
• Under the collision probability constraint, the secondary
user achieves the maximum throughput when it transmits
fixed length packets
Overhead Consideration
• Optimal packet length achieves trade-off
between overhead and collision probability
Relation between two constraint metrics
Multi-band multiple secondary
systems
• No synchronization between secondary
users and primary users
• No control channel for secondary users
• Collision probability constraint
• Perfect sensing
Two sensing strategies
All-Channel-Sensing
Random-Sensing
Virtual
Transmit
Vacation
Vacation
Randomly
choose a
Channel to sense
Sensing All
channel
Y
Busy?
Virtual
Transmit
Y
All channel
busy
N
Transmit a
packet
N
Randomly
choose an
idle channel
Transmit a
packet
Simulation result for Multi-band
competitive systems
Smart Antenna Technique Applied
in Cognitive Radio Networks
• Design Objective:
 Maximize the QoS of SUs while protecting PUs
 Design MAC Protocols to take advantages of smart
antenna technologies
• System Setup:
 One primary Tx (PT), one primary Rx (PR)
 One cognitive Tx (CT) , one cognitive Rx (CR)
 PT and CT transmit simultaneously to PR and CR,
respectively
• Performance metric:
 talk-able zone of CR
System Model
Cognitive Rx
d cc
d pc
 cp
 cc
d cp
Cognitive Tx
Primary Rx
 pp
d pp
 pc
y p  hpp s p  hcp sc  n p
Primary Tx
ys  hcc sc  hpc s p  nc

ij
hij  d w v ( ), i, j  p, c
H
i i
Optimal Beamforming Problem with
Constraints
min
wc
max
 cp    ci  cp  
| Gc ( ci ) |
s.t.
| Gc ( cc ) | 1
| Gc ( cj ) | 1 / 2,  cj  [ cc   ,  cc   ]
Gc ( )  w cH v ( )
v( ) : array manifold
• Can be solved efficiently by convex optimization
method
A Typical Beamforming pattern of a
Secondary TX
Beamforming Pattern of Cognitive Tx
0
-10
|Gs(2i)| in dB
-20
-30
-40
-50
-60
-70
-80
Primary Rx
Cognitive Rx
0
50
100
150
2i
200
250
300
350
Simulation Results (1)
• PT uses omni-directional
antenna
• PRs are evenly distributed
over the area centered at
PT
• Interference to PR is less
than 0.1 of the received
signal power
• Spectrum efficiency
increased at least by:
Shaded Area
p
 40.64%
Total Area
4500
p = 0.9
4000
p = 0.7
3500
p = 0.5
3000
2500
CT
2000
1500
1000
PT(omni-directional Ant.)  = 15dB
c
T = 6dB
p = Pr[SINRc  T]
500
0
0
500
1000
1500
2000
2500
3000
3500
4000
4500
Simulation Results (2)
• PT uses Transmit
beamforming
• PRs are evenly distributed
over the area centered at PT
• Interference to PR is less
than 0.1 of the received
signal power
• Spectrum efficiency
increased at least:
Shaded Area
p
 45.15%
Total Area
4500
p = 0.9
4000
p = 0.7
3500
p = 0.5
3000
2500
CT
2000
1500
PT (TXBF)
1000
c = 15dB
T = 6dB
p = Pr[SINRc  T]
500
0
0
500
1000
1500
2000
2500
3000
3500
4000
4500
Integration of MAC/PHY design in
Cognitive Radio Networks
• Design Objective:
Under the collision probability constraint,
increase the capacity of secondary users
A cross-layer approach
• Channel models
 Rich scattering environment: Rayleigh fading
MISO channel from CT to CR and PR
 Rayleigh SISO fading channel from PT to PR
and CR
Received signal model
• Idea:
– when overlapping happens, primary user can decode
its signal as long as the interference power from
secondary user is very small.
– Transmit beamforming helps in this scenario, since it
can mitigate the interference to primary users;
• Collision probability:
v1
P P 
l2  v2
c
1
c
2
P  P  Pr[ I cp  I 0 ]
c
1*
c
1
I cp : Interferen ce from CT to PR
I 0 : Interferen ce threshol d
Simulation Result
Conclusions
• Opportunistic spectrum access of secondary
users can increase the spectrum efficiency of
system
• Smart antenna technique enables concurrent
transmission of primary users and secondary
users, and reduces interference to primary user
• Integration of PHY/MAC layer can improve
system’s spectrum efficiency