A Channel Allocation Algorithm for Cognitive Radio Users Based on Channel State Predictors

A Channel Allocation Algorithm for Cognitive Radio
Users Based on Channel State Predictors
Nakisa Shams, Hadi Amirpour, Christian Timmerer, and Mohammad Ghanbari
ICICT 2021- February 26, 2021
ICICT 2021, 6th International Congress February 25 – 26, 2021

Introduction
Outline
CCAA using Channel State Prediction
CR Performance with Channel State
Prediction
Performance Evaluation
Conclusion
ICICT2021

Introduction
Cognitive radio (CR) techniques provide the capability to share the
spectrum in an opportunistic manner.
1
The main task in a CR system is to determine the available frequency
bands by exploiting spectrum sensing.
2
[1]
[1] I. F. Akyildiz, W.-Y. Lee, M. C. Vuran, and S. Mohanty, "A survey on spectrum management in cognitive radio networks," IEEE Communications magazine, vol. 46, no.
4, pp. 40-48, 2008.
ICICT2021

Introduction
The dynamicity and decision making required by the CR will be
supported by AI technique.
1
Learning is the process of gathering knowledge after applying decisions,
observing their effects and then adding new information to awareness.
2
CR exhibits sensing, reasoning, and learning, and adapting to the
spectrum environment
3
ICICT2021 [2] https://www.engineersgarage.com/tech-articles/cognitive-radio/
[2]

CCAA using Channel State
Prediction
ICICT

Secondary users in a CR system try to find and use the idle
slots to reduce its interference with the primary users.
1
ICICT2021
Channel Allocation
There is a need to employ a CSP to identify and choose the best
communication channel.
2
The knowledge requirements of primary users can be eliminated
by using the channel state prediction.
3
Channel state prediction is performed on two types
of the neural networks
Time Delay Neural Network (TDNN)
Recurrent Neural Network (RN N)

A centralized channel allocation algorithm for secondary users
Channel Allocation
ICICT2021
Start
Assign initial values i=1, m=1, n=1
Number of time slots< i
Select a channel by the secondary
users based on predictor output
NO
Number of users < n
Allocate channel to the secondary user in
this time slot based on the waiting time
YES
i = i +1
Number of channels < m NO
Traffic_PU(T,m,n) == 0 NO
NO m = m +1
n = n +1
YES
Stop
YES
YES

There are “n” secondary users on the network , i.e., n∈{1, , . . . , N}.
1
Each secondary user uses the channel state prediction.
2
ICICT2021
Channel Allocation
The distribution of traffic on the channels of the primary user is the
same for all secondary users.
3
In each time slot, channel is free with a probability of “Pm” and
occupied with a probability of “1−Pm”.
4

ICICT2021
Channel Allocation
Interference between a primary
user and secondary users
Interference between secondary
users
Each secondary user tries to learn the traffic model of the primary user
by using its CSP.
1
By considering several secondary users in the system, the concept of
secondary interaction is defined.
2
This interference occurs when the secondary user selects the channel
occupied by the primary user and transmits data on that channel.
1
This interference occurs when more than one secondary user select the
same channel for data transmission
2

CR Performance with using
Channel State Prediction
ICICT

ICICT2021
The benefits of channel state prediction are
expressed using
The percentage of improvement in
spectrum usage of SUI(%)
The percentage of reduction in
sensing energy of SER(%)
Considerable energy is consumed by sensing the spectrum.
1
Secondary users can save energy detection by preventing the occupied
channels from being detected during sensing.
2
3
Spectrum usage can be improved by achieving a low probability of false
prediction in an idle channel.

ICICT2021
Improvement in spectrum
usage
expressed using
Because of channel state prediction, the percentage of improvement in
spectrum usage can be expressed as:
SUI
Isense and Ipredict represent the number of sensed idle time slots and the
number of the sensed time slots which are predicted to be idle, respectively.
SUI

Sensing Energy Reduction
ICICT2021
expressed using
Overall sensing energy for all the time slots, and the total sensing energy for those
channels whose states are idle during the respective time slot can be expressed by SEP.
Bp is the total number of the time slots predicted to be busy.
SER
SER

Sensing Energy Reduction
ICICT2021
expressed using
The percentage of reduction in sensing energy SER can be given by:
SER
It means that the sensing process is not performed if the state of the timeslot
is predicted to be busy.
SER

Performance
Evaluation
Performance Evaluation of the CCAA
CR Performance in Spectrum Usage
Improvement and in Sensing Energy
Reduction
ICICT2021

The performance of the CCAA will be evaluated by two parameters,
(i) fairness and (ii) channel switching.
Employing a CCAA helps secondary users minimize switching
channels to figure out the idle channels.
3
1
A network with five secondary users is considered under stationary
traffic condition (N= 5).
2
Comparing number of channel switchings by
five secondary users under static traffic
condition in T = 20,000.
Performance
Evaluation
ICICT2021
Number of channel switching using TDNN predictors, and RNN predictors, is
less than 400 out of 20000 time slots and 200 out of 20000 time slots,
respectively.
4

Performance
Evaluation
ICICT2021
The proposed algorithm ensures sufficient fairness among
secondary users and maintains fairness among them.
1
Comparing access to the channels by five
secondary users under static traffic
condition in T = 20,000
Secondary users using RNN-2HL can access more channels than
other users.
2
The art of channel access is reduced by increasing the number of
secondary users on the network because of increased interference.
3

Performance
Evaluation
ICICT2021
By using TDNN and RNN predictor, sensing time slot that has been
predicted to be idle can detect more idle slots than sensing all time slots.
2
Performance of the four CSPs as percentage
of the improvement in spectrum usage.
Consider eight primary channels (M= 8) with different traffic
distributions.
1
It can be seen that the number of sensed timeslots with a predicted idle
state improves as the number of channels increases.
3

ICICT2021
Performance
Evaluation
Comparing the percentage of spectrum usage improvement
 Percentage of improvement in spectrum
usage for the predictors of TDNN-1HL and
TDNN-2HL is more than 77% and
80%,respectively.
 by using the predictors of RNN-1HLand RNN-2HL,
the percentage of improvement in spectrum
usage is more than 82% and 89%, respectively.

ICICT2021
Performance
Evaluation
 The proportion of sensing energy reduction as a percentage for different traffic intensities.
 It presents various mean ON + OFF times on the channel when secondary users use different
predictors.
 The highest percentage of sensing energy reduction is 73%, and the lowest percentage of
sensing energy reduction is 50%.

Conclusion
A centralized allocation algorithm based on the channel state
predictor is presented.
1
The channel allocation is evaluated by the number of channel
switching on the network and the fairness between secondary users.
2
The channel switching probability in the multiple secondary users is
less than 2% (less than 400 out of 20000 time slots).
3
The channel status prediction saves the sensing energy and improves
the spectrum usage.
4
ICICT2021

Thank You
For Your Attention
Questions?
ICICT

A Channel Allocation Algorithm for Cognitive Radio Users Based on Channel State Predictors

More Related Content

What's hot (20)

Similar to A Channel Allocation Algorithm for Cognitive Radio Users Based on Channel State Predictors (20)

More from Alpen-Adria-Universität (20)

Recently uploaded (20)

A Channel Allocation Algorithm for Cognitive Radio Users Based on Channel State Predictors