RESOURCE ALLOCATION IN COGNITIVE RADIO
NETWORKS
Presented by: Syeda Aqeela Naqvi
OUTLINE
Why we need CRN’s
Introduction to Cognitive Radio Networks (CRNs)
Function of CRN’s
Modes of CRN’s
Resource Allocation in CRNs
Joint Channel and Power Allocation Problem in
Multiuser CRNS
Recent Trends in Resource Allocation in CRNs
Why Cognitive Radio?
According to a FCC report [1],a large portion
of the licensed spectrum of various agencies remains
underutilized.
The concept of Cognitive Radio is introduced as a method
to improve the spectrum utilization.
Fig 1: Spectrum Usage [2]
INTRODUCTION TO COGNITIVE
RADIO NETWORKS
Cognitive Radio
Formally a cognitive radio network is defined as;
A radio that changes its transmitter parameters based on the
interaction with the environment.
The cognitive radio has been mainly proposed mainly to
improve the spectrum utilization by allowing unlicensed
(SU) to use under-utilized licensed frequency bands.
Terminology
Primary user (PU):users have license to use a certain
frequency spectrum.
Secondary user (SU): devices/ users that able to
sense and adapt licensed users allocated spectrum.
Spectrum hole: a frequency band licensed to a PU but not
utilized by that user at a particular time and at a specific
geographic location
Introduction to CRNs ( Contd «
In order to access u‡nder-utilized ·licensed bands
dynamically
and opportunistically, Cognitive Radio has to:
identify the spectrum opportunities (i-e free frequency bands)
in spatial and frequency domain.
or use the licensed spectrum with transmit power constraint so
that the interference created by secondary users is below the
tolerable limit.
Intro continue…
To share the spectrum the primary network should adapt the
parameters in order to protect the primary users.
The primary network should be unaware of the secondary
network and does not need to run any special function that is
known as transparency of secondary to primary.
Functions of CRN’s
In order to efficiently utilize the underutilized spectrum and
meet various QoS requirements and dynamic spectrum
utilization following functions of CRN’s should be performed
efficiently.
Spectrum sensing
Spectrum decision
Spectrum sharing
Spectrum mobility
Basic Cognitive Cycle
Fig 1 Cognitive Radio Cycle
Modes In CRN’s
Models for dynamic spectrum access
Cognitive Radio Paradigms
• Underlay
In spectrum underlay technique, the secondary users can
transmission the frequency bands used by the primary users as
long as they do not cause unacceptable interference for the
primary users. This approach does not require secondary users to
perform spectrum sensing; however, the interference caused by
the secondary user’s transmission must not exceed the
interference threshold.
• Overlay
In spectrum overlay, the secondary users first sense the spectrum
and find the location of a spectrum hole (vacant frequency band).
After locating the vacant frequency bands, the secondary users
transmit in these frequency bands.
Shared use model for spectrum overlay
and underlay
Underlay Interface
SpectrumPower
SpectrumPower
Threshold
SU
PU PU
Underlay Interface
SU Threshold
Frequency Frequency
(a) Spectrum overlay (b) Spectrum underlay
Underlay Systems
Cognitive radios determine the interference their
transmission causes to noncognitive nodes
Transmit if interference below a given threshold
IP
NCR
NCR CR CR
The interference constraint may be met
Via wideband signalling to maintain interference below the
noise floor (spread spectrum or UWB)
Via multiple antennas and beamforming
Underlay Challenges
Measurement challenges
• Measuring interference at primary receiver
• Measuring direction of primary node for beam steering
Policy challenges
Underlays typically coexist with licensed users
Licensed users paid $$$ for their spectrum
Licensed users don’t want underlays
Insist on very stringent interference constraints
Severely limits underlay capabilities and applications
Resource allocation for Cognitive radio
networks
Approaches in Resource Allocation
Centralized system. Distributed system.
Resource Allocation Process
Cognitive Radio Networks Architecture
1- Infrastructure-Based
In the Infrastructure architecture (Figure 1.1), a MS can only access a
BS/AP in the one-hop manner. MSs under the transmission range of the
same BS/AP shall communicate with each other through the BS/AP.
Communications between different cells are routed through
backbone/core networks. The BS/ AP may be able to run one or multiple
communication standards/protocols to fulfil different demands from MSs.
A cognitive radio terminal can also access various kinds of
communication systems through their BS/AP.
Cognitive Radio Networks Architecture
Figure 1.1 Infrastructure architecture of a CRN
Cognitive Radio Networks Architecture
2- Ad-hoc Architecture
There is no infrastructure support in ad-hoc architecture (Figure 2.1). The
network is set up on the fly. If a MS recognizes that there are some other
MSs nearby and they are connectable through certain communication
standards/protocols, they can set up a link and thus form an ad-hoc
network. Note that these links between nodes may be set up by different
communication technologies. In addition, two cognitive radio terminals
can either communicate with each other by using existing communication
protocols (e.g., WiFi, Bluetooth) or dynamically using spectrum holes .
Cognitive Radio Networks Architecture
Figure 2.1 Ad-hoc architecture of a CRN
Resource Allocation in CRNs
Based on local information on the spectrum band, CR users
need to determine the communication resources intelligently.
Each CR user tries to utilize spectrum resource as much as
possible.
Two main issues in Resource Allocation:
Power Allocation
Channel Allocation
Opportunistic Interference Avoiding
Scheduling for Underlay Cognitive
Radio Network
• An opportunistic scheduling scheme that avoids the
interference that a primary multi-antenna Base Station (BS)
generates in the secondary single-antenna links of an
underlay cognitive radio network.
• Primary BS apply a pseudo random sequence of transmit
beam forming vectors to serve its users.
• By combining the information with estimated value of slow
varying channel gain the secondary Rx can delay their
transmission until interference coming from primary is far
from its peak.
System Model
• Primary BS is fixed and has multiple antennas
• Primary BS serves primary mobile stations and secondary link
that coexists with primary system and uses same spectral
resources.
• Primary mobile station and secondary link have single
antenna.
• Interference introduced between primary BS and secondary
link is kept under control by restricting the maximum transmit
power of secondary transmitter.
• Objective: to improve overall data rate performance of
secondary links.
A) Signal Model for the Secondary link
∈
Secondary link is in the coverage of the primary BS. TTI attains
the form;
rk[i] = hk,k sk +(hp,k ・ w[i]) sp +nk
Desired signal Interference Noise
hk,k is the direct channel gain in the secondary link k,
hp,k is the vector that contains the complex
channel gains from primary BS antennas to secondary
RX k,
w[i] is the random TBF vector that primary BS apply at
time instant i.
b)Interference aware Scheduling
• To increase data rate in the secondary link transmission
should be scheduled in those TTI where instantaneous to
peak data rate satisfies given threshold.
• The transmission in secondary link takes place when the
interference is below the threshold value.
c) Codebook based Random Beam
forming Scheme
• In conventional beam forming scheme both phase and
amplitude of each transmit antenna vary continuously.
• Channel tracking is done by common pilot signal that is
broadcast in downlink and is affected by random TBF
vector.
• TBF vector vary slowly to accurately estimate
instantaneous channel condition.
• Rx update information of the channel gain every time TBF
vector changes. Results in excessive use of resources.
• To handle this situation codebook based RBF was designed.
• In this scheme large coherence time of the channel was
considered.
• Instead of keeping in track the overall gain of the slow
varying channel. Rx keeps track of the individual channel
gain.
• The Rx then determines the element that is closest to the
beamforming configuration.
• After having pseudo random TBF factor knowledge the Rx
inform Tx about TTI that is below threshold level.
Joint Channel and Power Allocation
Problems in Multiuser CRNs
But to achieve satisfactory performance from a multiuser
multichannel wireless network, such as CRNs, not just power
but joint channel and power allocation provides satisfactory
results.
System Model
A CRN is considered with K users and N channels (both K and N
varying dynamically based on the number of contending users and available
vacant channels) .
An access point (AP) controls the
transmission of CRs lying within its
range of coverage and also collects
reports about the activities of
primary users (PUs) that CRs may
interfere with.
AP CR Primary User
Fig. 7: Cognitive Radio Network [8].
Assumptions
The following knowledge [8] is available at the AP:
(i)the set of vacant channels that are not currently utilized
by PUs and are free for CRs to use.
(ii)the power gains of this channel set corresponding to
each of the contending users.
Channels are assumed to be independent and identically
distributed (IID). The strength of each is assumed to be
Rayleigh distributed.
Problem Formulation
Let
N0: one-sided noise power spectral density
B: unit bandwidth
Pnk: uplink power when channel n is assigned to the user k
Capacity associated with user k (depends on the number of channels assigned) is
N Pnk gnk (14)
C B log
k nk N0 B
2n1 1 nk assignment of channel n to user k,ϵ {0,1}
The total sum capacity is:
K
(15)
K N
nk B 1 Pnk g nk
C
k Ck k
1
n1
log
2
N 0 B nk
1
The objective function is:
max (16)
Problem Formulation (contd…)
Sum transmit power constraint available power budget
N
n 1
Pk k
(17)
P nk
The constrained optimization problem is :
m ax from equation (16)
s.t.
C
C1: sum transmit power constraint from equation (17)
0 n k 1,P nk 0
The above optimization problem comprises both continuous
and discrete variables and thus belongs to the class of mixed
integer programming.
Problem Formulation (contd…)
To make the problem tractable, discrete nature of αnk is relaxed
and channel sharing among different users is allowed, i.e. αnk
is allowed to take on continuous values in the range from 0 to 1.
Thus, now the optimization problem is :
m ax from equation (16)
s.t.
C
C1: sum transmit power
0 constraint
1,P 0
nk nk from equation (17)
C2: assignment constraint, i.e.
K
k
nk 1 (18)
1 n
Problem Formulation
In this new formulation, channel sharing is allowed among
different users under the condition stated in C2.
The importance of having αnk in the denominator of C (15)
becomes clear now as αnk can be a fraction (channel
sharing).
The methodology used here is to start solving the modified
problem (with channel sharing) and then find the condition that
allows for only 1 user utilization per channel.
Above formulated problem being convex, convex
optimization has
been used to solve it.
max C
0 nk 1,Pnk
0
Recent Trends and Future
recommendation for Resource
Allocation in CRNs
In a highly dynamic environment (as in CRN), finding a reasonably good solution (i.e., a
suboptimal solution) fast enough is the only practical goal. Otherwise, spectrum holes may
disappear before they can be utilized for communication. In such a situation, the concept of
equilibrium is very important, and here comes the advantage of using Game Theory in this
context.
Due to this advantage of the very idea of Game Theory , it is now being quite interestingly
used to solve the Resource Allocation Problems in CRNs. For instance, [12] shows the
investigation of distributed power control for CRNs, based on a cooperative game-theoretic
framework.
Network Mobility: Static networks are almost assumed in
most literature works, and network mobility is greatly
. ignored. It is desirable to study a CRN with mobile SUs and
investigate the impact of nodes mobility on the performance
of the RA techniques. In particular, in a mobile network model
the SUs can move during the operation changing their space
and time coordinates. Certainly, this will affect the analysis
derived for the static case. For instance, the SUs may be closer
to the primary receivers which could result in IT violation
and harming the primary transmission. Therefore, the spectrum
underlay etiquette is violated and the SUs should act quickly
to solve this issue. Additionally, the influence of network
mobility on the optimality of the RA process for underlay
CRNs is still an open problem. Also, the consideration of
realistic user and traffic distributions is beneficial.
Hybrid Users: A hybrid user is a PU with cognition
capability . These users could have hybrid access to both
primary and secondary networks. Hybrid users have good
potential to be adopted in a number of applications in the
future such as cognitive femtocells where a user can select to
be connected to either macro BS as a PU or to a femtocell
as a SU to enhance its performance and coverage. Even, it
is possible for these hybrid users to access both secondary
and primary networks simultaneously (using different radio
interfaces depending on the applications), and thus maximizing
their benefit.
Experimental Testbeds: To effectively evaluate the performance
of CRNs techniques, it is very essential to test these
techniques on a realistic environment. This would also help in
assessing their practicality and highlight real implementation
issues that require further investigation. Designing CR testbeds
can provide a tool for researchers to experimentally approve
their analytical results in a real environment
