Sie sind auf Seite 1von 13

Energy Harvesting in Cognitive Radio using

Neymyan Pearson Lemma


K. Swaminathan, Kushagra Goutam and Rajat Ligade

Abstract

Cognitive Radio is one of the trending areas of technical collage of various engineering disciplines.
Specifically speaking, Green Cognitive Radio is the area which is assumed to be the fuel which powers
modern research in this domain. Cognitive radio, in general, has been portrayed in three basic paradigms-
Underlay, Overlay and Interleave. Our work primarily is focussed on overlay paradigm of green cognitive
radio. Earlier, methods of spectrum detection, estimation and duty cycle optimisation were done for either
Primary user or secondary user as Energy Harvesting capable node only. In our project, we propose a
system model which allows both the Primary and secondary users to be capable of Energy Harvesting,
Green Radio. We discuss, in this project, the methods proposed to demonstrate the energy harvesting
capability of the individual nodes and the optimal duty cycle necessary for the same.

I. M OTIVATION

Wireless spectrum is one of the most important resources required for radio communications.
Throughout the world, spectrum utilization is regulated so that essential services can be provided
and also protected from harmful interference. There are Spectrum regulatory bodies such as the
Federal Communications Commission (FCC) in the US or the European Telecommunications
Standards Institute (ETSI) in Europe have always allocated spectrum frequency blocks for specific
uses, and assigned licenses for these blocks to specific groups or companies, this registration
for a spectrum cost a huge amount of money and not financially possible for a small start-up
or a small organization. There have been several studies and reports over the years that show
that spectrum is in fact vastly underutilized. A report presenting statistics regarding spectrum
utilization showed that even during the high demand period of a political convention such as the
one held in 2004 in New York City, only about 13% of the spectrum opportunities were utilized.
So this is a motivation to use the spectrum efficiently, unlicensed users can use the licensed
spectrum for transmission of data.

The authors are with the Dept. of Electrical Communication Eng. at the Birla Institute of Technology and Science, Pilani.

December 7, 2018 DRAFT


1

Devices using advanced radio and signal processing technology should be able to exploit the
underutilized spectrum. Much of the early motivation for cognitive radio technology was indeed
to accomplish such opportunistic spectrum use and to also alleviate the artificial scarcity of
prime spectrum. If successful, this technology could revolutionize the way spectrum is allocated
worldwide.

II. I NTRODUCTION

Cognitive radio is a generic term used to describe a radio that is aware of the environment
around it and can adapt its transmissions according to the available spectrum In their simplest
embodiments, cognitive radios can recognize the available systems around them and adjust their
frequencies, waveforms, and protocols to access those systems efficiently. An important feature
that differentiates cognitive radios from normal radios is their agility Which are mentioned below:
Spectrum agility or frequency agility refers to the discovery strategies for available spectrum as
well as opportunistic transmission in the identified spectrum. In simple words, it can adjust its
frequency bandwidth according to the availability of frequency spectrum.
Technology agility refers to the operation of a single radio device across various access
technologies. In simple words, all the parameter is adjustable in cognitive radio as mentioned
cognitive radio is software-defined radio and it gets all the flexibility possible.
Protocol agility refers to constituting a dynamically reconfigurable protocol stack on radio
devices so that they can proactively and reactively adapt their protocols depending on the devices
they interact with. Cognitive radio networks basically extend the software defined radio (SDR)
framework to the development of dynamic spectrum access (DSA) algorithms

III. D IFFERENT PARADIGMS OF C OGNITIVE R ADIO

Depending on the usage scenario, cognitive radio network operation can be classified as
interweave, overlay, and underlay paradigms,

A. The interweave paradigm

In this paradigm, the secondary users should not interfere with the communication between
the already active primary users. This mandates the secondary users to be able to detect (sense),
with very high probability, the primary user transmissions in the network. Once the cognitive
radio successfully detects the primary user transmissions, it can opportunistically communicate

December 7, 2018 DRAFT


2

only if it is able to do so without harming the primary transmissions. The temporary space-time-
frequency void in the transmission of primary users is referred to as a spectrum hole or a white
space.

B. Overlay Paradigm

In the overlay paradigm, the secondary user needs to know the channel between the primary
transmitter and the primary and secondary receivers as well as the channel between the secondary
transmitter and the primary receiver. With the channel knowledge of both the primary and
secondary users, the secondary user can then choose appropriate transmission strategies so that
the communication in the secondary network causes the least interference to the primary network.
The point should be noted that unlike interweave paradigm the secondary user won’t be searching
for a void, in spite will be using the same spectrum along with the primary user.

C. Underlay Paradigm

In the underlay paradigm, the secondary transmitter keeps the interference levels below a
certain threshold. The primary receiver sees a higher noise level if the primary and secondary
transmission overlap in the same band. While the primary transmission is decoded with an
enhanced noise floor, the secondary receiver despreads the data to decode the transmissions.
In this project, we will be using Overlay Paradigm, since this paradigm makes the best use
of the spectrum during off time of the primary user, on the other hand, the underlay paradigm
reduce the ouptut signal noise ratio of the licensed user.

IV. GREEN COGNITIVE RADIO

The concept behind using Green cognitive radio is making Both primary user and secondary
user self-sufficient, self-sufficient here means using harvested energy to transmit the signal and
does not depend on some external power source. The harvested unit can be anything from a
solar panel, wind turbine or using the EM radiation to generate power. Here we assume that
there will be a harvesting period out of the total period, and using this in our advantage the
harvesting period of the primary user can be used as Transmitting period for the secondary user
and similarly using transmission period of the primary user as harvesting period for secondary
user. One logical assumption we have to take is that we cannot change any protocol for the

December 7, 2018 DRAFT


3

primary user since it the licensed user and have legal authority of that spectrum, so all the
changes will be made in secondary user protocol.
As Illustrated in the diagram, there is a recharging Unit or harvesting Unit connected to the
battery to make the system sufficient as mentioned below, sensing unit and memory is used to
detect the signal of the primary user so that synchronization can take place.

A. PROBLEM STATEMENT

The primary objective is to design a model to efficiently use the energy harvested by the
primary and secondary users for the radio transmissions and subsequently, reduce the probability
of false alarms and missed detection.

B. Approach to the Detection Model

Different techniques for Detection


Bayesian Detection theory:
A decision rule that minimizes this average error probability Pe is called a Bayesian optimal
decision rule. The distinguishing feature of the Bayesian approach is the averaging of an assumed
cost function with respect to the a priori probabilities. Hence, the Bayesian approach presumes
that:
• Hypotheses can be modeled as endowed with a certain prior distribution.
• This prior distribution is known.

December 7, 2018 DRAFT


4

Neyman-Pearson(NP) theory:
The two key assumptions of the Bayesian approach can be justified: (i) Hypotheses have certain
prior distribution (ii) prior distribution is known. There are situations in which either or both
of these assumptions may not be valid. There are many problems in which this is the case,
including our model i.e cognitive radio(CR) spectrum sensing problem. Hence we go with the
Netman Pearson(NP) approach where no prior knowledge of destribution is required. Our system
model deals with energy harvesting in CR using NP.

C. Project Outline

We have defined a cognitive radio system consisting of the primary and secondary user. Having
this pre-requisite, the below are our objectives:
• Detection of the primary user using Neyman-Pearson Detection Theory Analysis and sub-
sequent reduction in the Probability of misses and false alarm in the detection process.
• Using Probability distribution to come up with a more realistic design to take care of
uncertainity in detecting the primary user.
• Parametric analysis of the Probability of Detection vs. SNR of the received signal
• Efficiently using the harvested energy for the transmission of a signal by the Secondary
and Primary user. To do so we come up with a bound for harvesting period that will be
necessary for the self-sufficient system in terms of energy.

December 7, 2018 DRAFT


5

V. S YSTEM M ODEL

Fig. 1. System Model

A. Assumptions

• The primary signal is considered to be a constant signal here. The channel Noise is assumed
to be a standard Additive White Gaussian noise ∼ N(0,1).
• For a Fixed PFA (Probability of false Alarm) we will be calculating PD (Probability of
detection) for various value of ENR.
• We have assumed that the harvesting factor τ is an unknown constant and secondary
user(SU) will determine it using pilot sequence.
• N1 and N2 are the samples of harvesting and transmitting period of primary user(PU)
respectively.

B. Solution to our problem

We approach our problem using NP.


P [X; H1 ]
>γ (1)
P [X; H0 ]

December 7, 2018 DRAFT


6

The test Statstics we get is

T (X) = P T1 (x) + (1 − P )T2 (x) (2)

where PN −1
n=0 x(n)
T1 (X) = x̄ = (3)
N

PN2 −1 P 1 −1 2 PN2 −1
n=0 x2 (n) − N n=0 x (n) x(n)
T2 (X) = + n=0 (4)
2N2 A N2
The above can be undersood as assigning probabilites for two cases:
1. The scenario when τ is 0.5
2. The scenario when τ > 0.5
The probability p is for the first case and 1-p is for the second.
Now let us arrive at the Expectation and variance of this test statistics for the alternate and
null hypothesis proposed in the beginning
We get the Expectation for T(x) given Ho and it’s variance as follows

2
 
σN (N1 − N2 )
E[T (X); H0 ] = 1 − P (5)
2N2 A

 
2 N1 − N2 4 2
V ar[T (X); H0 ] = (1 − P ) 3σN + σN P2 (6)
4N22 A2
Now let us look at the Expectations and variance for hypothesis H1
 2 2
 
(A + σN )(N1 − N2 )
E[T (X); H1 ] = P A + (1 − P ) +A (7)
2N2 A

(A4 + 4A2 σN
2 4
  
+ 3σN )(N1 − N2 )
V ar[T (X); H1 ] = P 2 σN
2
+ 2 2
2
+A + 2
σN (1 − P )2 (8)
4N2 A
Deflection coefficient for our secondary user is derived as follows
s
[E[T (X); H1 ] − E[T (X); H0 ]]2
d=
V ar[T (X); H0 ]
v
uh  h
A(N1 −N2 )
ii2 (9)
P A + (1 − P ) + A
u
u 2N2
=u t h i
(1 − P )2 N4N1 −N 2
2 A2 3σN4 2
+ P 2 σN
2

December 7, 2018 DRAFT


7

Since our model is for energy harvesting nodes, We have come up with the harvesting parmater
in relation to the Harvesting rate. First we define our harvsting rate for the primary user as Kpu
joules/Sample(Sample becoz we are handling discrete functions in problem.It can be generalized
to continous time). Then we assign Ksu joules/Sample as the energy harvesting rate for secondary
user. Precisely what we are trying to achieve is to formulate a relation ship for both these energy
harvesting rates with the harvesting parameter(τ ).
Now we all know that in a particular time period T, the energy harvsted by a primary user
in the harvesting time should be greater than or equal to the energy it spends in transmission.
Similar is the prerogative for the secondary user. So now we can formulate the relation as follows

Kpu τ T fs ≥ (1 − P )A2 (1 − τ )T fs + P A2 0.5τ (10)

Using this equation, we can calculate τ

(1 − P/2)A2
τ≥ (11)
Kpu + (1 − P )A2
And using this value, we calculate energy harvesting rate for secondary user.

((1 − P )τ 2 ) + P (0.5)A2
Ksu ≥ (12)
1−τ

December 7, 2018 DRAFT


8

VI. S IMULATIONS A ND R ESULTS

Fig. 2. P=0.1

Fig. 3. p=0.2

December 7, 2018 DRAFT


9

Fig. 4. p=0.3

Fig. 5. p=0.4

December 7, 2018 DRAFT


10

Fig. 6. p=0.5

Fig. 7. p=0.6

December 7, 2018 DRAFT


11

VII. C ONCLUSION AND I NFERENCES

From the above plots we can arrive upon the following inferences:
• As we increase the Energy to Noise ratio, we see the traditional increase in the Probability
of detection at the reciever end.
• While, we have the value of the Energy Harvesting parameter (τ ), varied, we can see that
the probability of detection is higher for a lower value of τ .
• Another prime observation is the behavior with respect to the change in the bernouli
probability in the equality case of Test Statistic. When we have a higher probability of
the τ becoming 0.5, the probability of detection gets degraded for a given value of Energy
to noise ratio.
• There is a specific band of ENR values where a higher value of τ can be tolerated compared
to the lower values and consequently gives a better detection performance.

VIII. L IMITATIONS

• Our model assumes a bernoulli distribution for the harvesting parameter indirectly. This
would not be the case in practical radio environment.
• The channel fading is not considered.The results would have substantial changes in proba-
bility calcuations.
• Initial states of the Nodes(PU and SU) are taken to be the same and set to 0. The results
will be accordingly modified for non-zero and unequal initial states.

December 7, 2018 DRAFT


12

R EFERENCES

[1] .P. Praveen and N.B. Mehta, “Trade-offs in Analog Sensing And Communication in RF Energy Harvesting Wireless Sensor
Networks”,2016.
[2] .K Jayaweera, “Signal Processing for CR”, Edition 1, Aug. 2014.
[3] .W Poor, N.B Mandayam, “Principle of CR”, 2013

December 7, 2018 DRAFT

Das könnte Ihnen auch gefallen