UNIT-III

SPECTRUM SENSING AND DYNAMIC SPECTRUM ACCESS

Introduction – Primary User Detection Techniques – Energy Detection, Feature Detection, Matched
Filtering, Cooperative Detection And Other Approaches, Fundamental Tradeoffs In Spectrum
Sensing, Spectrum Sharing Models Of Dynamic Spectrum Access - Unlicensed And Licensed
Spectrum Sharing, Fundamental Limits Of Cognitive Radio.

Reference Books

1. Cognitive Radio Communications and Networks (Alexander M. Wyglinski , Maziar Nekovee, Y.

Thomas Hou)

2. Cognitive Radio Networks (Kwang-Cheng Chen, Ramjee Prasad)

INTRODUCTION (REF. BOOK: 1)

Multiple measurement campaigns reveal that much of the licensed spectrum remains
unused—both in time and in frequency: traffic in wireless networks tends to be bursty.
Hence, efficient exploitation of the spectrum requires the ability to exploit instantaneous
opportunities at a rather fine time scale. For cognitive networks to operate efficiently,
secondary users should be able to exploit radio spectrum that is unused by the primary
network.

A critical component of cognitive networking is thus spectrum sensing.

The secondary user (SU) should sense the spectrum efficiently, quickly seize opportunities to
transmit, and vacate the spectrum should a primary user (PU) reoccupy the spectrum.

The goal of spectrum sensing is to determine spectrum status and the licensed user’s
activity by periodically sensing the target frequency band.

In particular, a cognitive radio transceiver detects a spectrum which is unused or spectrum

hole (i.e. band, location, and time) and also determines the accessing method of it (i.e.
transmitting power and access duration) without interfering of a licensed user’s transmission.

Spectrum sensing may be either centralized or distributed.

In the centralized spectrum sensing, a sensing controller (e.g. access point or base station)
senses the target frequency band, and the information obtained is shared with other nodes in
the system. For example, the sensing controller may be unable to detect an unlicensed user at
the edge of the cell.

In distributed spectrum sharing, unlicensed users sense the spectrum independently, and the
spectrum sensing is achieved either used by individual cognitive radios (non-cooperative
sensing) or shared with other users (cooperative sensing).

In this chapter, we discuss

 how to detect spectrum opportunities by detecting primary signals and highlight the
difference between these two.

 present different types of detectors for primary signal detection.

 consider the detection of spectrum opportunities based on the detection of primary signals

PRIMARY SIGNAL DETECTION

The spectrum sensor essentially performs a binary hypothesis test on whether or not there are
primary signals in a particular channel.
 The channel is idle under the null hypothesis and busy under the alternate:

o H0 (idle) vs. H1 (busy)

Under the idle scenario, the received signal is essentially the ambient noise in the radio
frequency (RF) environment, and under the busy scenario, the received signal would consist
of the PU’s signal and the ambient noise; thus,

o H0: y(k) = w(k)

o H1: y(k) = s(k) + w(k)

for k = 1, . . . , n, where n is the number of received samples,

w(k) represents ambient noise, and s(k) represents the PU signal.

It seems natural that the received signal will have more energy when the channel is busy than
when it is idle; this is the underlying concept in the energy detector.

When aspects of the signal structure are known, one can exploit the structure; a special case
leads to the cyclostationary detector.

When the PU’s signal s(n) is fully known, one can use the matched filter

TECHNIQUES

Energy detection

Energy detection is the simplest sensing technique, which does not require any information
about the PU signal to operate.

It performs by comparing the received signal energy with a threshold. The threshold depends
only on the noise power.

The decision statistic of an energy detector can be calculated from the squared magnitude of
the FFT averaged over N samples of the SU received signal as illustrated in Figure
 In many cases, the signaling scheme of the PU may be unknown to the SU; this may
correspond to the case where an agile PU has considerable flexibility and agility in choosing
its modulation and pulse shaping.

In this setting, the energy detector is optimal in the Neyman-Pearson sense.

The spectrum sensing problem is now one of distinguishing between two mutually
independent and identical Gaussian sequences:

o H0: y(k) = w(k)

o H1: y(k) = s(k) + w(k)

for k = 1, ..., n. Here w(k) and s(k) are zero-mean complex Gaussian random variables with
variances σ2w and σ2s per dimension.

The Neyman-Pearson detector is a threshold detector on the likelihood ratio or equivalently

the log-likelihood ratio (LLR):

where τ is a suitably chosen threshold. Given the independent and identical assumption, the
detector is easily seen to be equivalent to deciding H1 if

CYCLOSTATIONARY FEATURE DETECTOR

In some standards, the PU network uses a pilot tone frequency that can be exploited by the
SU.

The use of a cyclic prefix also leads to periodic signal structures. The means and correlation
sequences of such signals exhibit periodicity and are, hence, called cyclostationary.

The test statistic in a cyclic detector is

 If the received signal y(n) can be written as y(n) = ∑k sk(n) exp(j2πfkn) + w(n), where sk(n)
are mutually independent zero-mean wide-stationary processes, independent of the circularly
symmetric white noise sequence w(n), then for a large N

where Rk(τ ) = E{sk(t)sk(t + τ)}.

MATCHED FILTER

Matched filter detector is a coherent pilot sensor that maximizes the SNR at the output of the
detector.
It is an optimal filter that requires the prior knowledge of the PU signals.
This sensing technique is the best choice when some information about the PU signal are
available at the SU receiver.
Assuming that the PU transmitter sends a pilot stream simultaneously with the data, the SU
receives the signal and the pilot stream.
Matched filter detection is performed by projecting the received signal in the direction of the
pilot, xp, as illustrated in Figure

(Image Source:IEEE paper)

Often the pilot sequences used in the primary network are known to the SU. the WRAN 802.22
standard specifies these sequences.
Let s(n), n = 1, ...,N, denote the known pilot sequence. Assuming perfect synchronization,
The received signal at the SU can be written as y(n) = hs(n) + w(n), where w(n) is additive
white Gaussian noise (AWGN) and h represents an unknown channel gain. For this AWGN
setting, the optimal detector is the matched filter
COOPERATIVE SENSING
The performances of a single detector can be severely degraded due to fading, shadowing, or a
faulty sensor.
This is one motivation for cooperative sensing, where observations from multiple SUs are
combined to improve detector performance.
The performance of such a system has been well studied in the context of bit detection (e.g.,
s(n) = 1 versus s(n) = −1).
Our detection problem is to test whether θ = 0 or 1.
The cooperative scheme requires a control channel (which could be in-band and frequency
hopped) and a trusted spectrum broker.
Latency is an important issue:
The time required to sense the channel, report the measurements to the FC (Fusion centre),
and for the FC to detect white space and allocate spectrum to a user must be considerably less
than the channel free time.
Distributed implementations are also possible in which SUs exchange information with all
their neighbors and each separately makes a decision.
In addition to its ability to cope with the hidden node problem, cooperative sensing can be
used to localize the active transmitters.

OTHER APPROACHES
The implicit frequency nonselective flat fading assumption made so far in this chapter
essentially assumes a narrowband channel model. When the spectrum to be sensed is
wideband, there are multiple challenges.
First, one may want to consider (partially overlapping) sub channels for each of which the flat
fading assumption would be reasonable.
How many channels should be monitored? This is dictated partly by the affordable complexity
of the receiver, the traffic usage in the primary network, and the desired rates for the
secondary user.
If the primary traffic is heavy, SU would seek to monitor multiple bands.
A related issue is whether the SU can sense (and transmit on) multiple (possibly well-)
separated channels or whether the channels should be contiguous.
Multiresolution- and wavelet-based methods have been proposed to deal with the wideband
problem.
By using the wavelet transform, the discontinuities can be identified and thus spectrum
activity detected.
FUNDAMENTAL TRADEOFFS IN SPECTRUM SENSING
Spectrum sensing is a fundamental in CR operation. Its performance decides the level of
interference with PUs and spectrum utilization efficiency.
Fundamental trade-offs in spectrum sensing, includes
 performance versus constraint
 sensing accuracy versus sensing overhead

PERFORMANCE VERSUS CONSTRAINT


MAC Layer Performance Measures
 Global Interference Model
 Local Interference Model
A fundamental question in designing the spectrum opportunity detector is how to choose the
detector operating point (P∗ FA, P∗ MD) to achieve the optimal trade-off between false
alarms and miss detection.
Such a trade-off, however, should be addressed in terms of MAC layer performance: the
throughput of the secondary user and the probability of colliding (PC) with primary users.
A translation from the physical layer performance in terms of {PFA, PMD} to the MAC layer
performance in terms of throughput and interference constraint is therefore crucial in
choosing the optimal operating point of the spectrum opportunity detector.
(PFA-Probability of false alarms; PMD-Probability of miss detection)
MAC LAYER PERFORMANCE MEASURES
The MAC layer performance is measured by the throughput of the secondary user and the
interference to the primary users.
The design objective is to maximize the throughput under a constraint on the maximum
outage probability ζ that the interference at an active primary receiver exceeds the noise floor
ρ. We refer to such events as collisions with primary users.
The objective and constraint at the MAC layer is thus given by max PS subject to PC ≤ ζ.
first consider PS, which is application dependent. For applications requiring guaranteed
delivery
an acknowledgment (ACK) signal from B to the secondary transmitter A is required to
complete a data transmission
In a successful data transmission, the following three events should occur in sequence:
  A detects the opportunity [I(A, rD, tx)] and transmits data to B;
 B receives data successfully [I(B, RI, tx)] and replies to A with an ACK;
 A receives the ACK [I(A, RI, tx)], which completes the transmission.
We thus have

The probability of collision is defined as

 PC = Pr{A transmits data | I(A, rI, rx)}.

GLOBAL INTERFERENCE MODEL

Consider first a global interference model where the transmission from every primary user of
interest affects the reception at B and the transmission from A affects the reception at every
primary user.
Under this condition, an opportunity occurs if and only if no primary users are transmitting.
Spectrum opportunities are thus symmetric, and detecting primary signals is equivalent to
detecting spectrum opportunities.
Furthermore, we have the following properties, assuming that A transmits in a slot if and only
if the channel is detected as an opportunity at the beginning of this slot.

Property : PHY-MAC Translation under the Global Interference Model

P.1 Successful transmissions from A to B can result only from opportunities (i.e., H0).
P.2 Every correctly identified opportunity leads to a successful transmission.
P.3 Every miss detection results in a collision with primary users.
These properties lead to the following simple relationship between {PFA, PMD} and {PS,
PC}.
PS = (1 − PF ) Pr[H0], PC = PMD.
With this relationship, to maximize PS under the constraint of PC ≤ ζ, we can obtain the
optimal operating point (P∗ FA, P∗ MD) for the spectrum sensor.
 LOCAL INTERFERENCE MODEL
When the transmissions from primary and secondary users have local effect, the
statements and the relationship between {PFA, PMD} and {PS, PC} given in Equation (PS =
(1 − PF ) Pr[H0], PC = PMD) no longer hold.
The relationship between PHY and MAC has complex dependency on the applications and the
use of MAC handshaking.

Impact of Application
Immediate acknowledgment is required at the end of each slot to complete a successful data
transmission. For the latter, acknowledgments are not necessary.
Due to the asymmetry of spectrum opportunities and the local effect of transmissions, we have
the following relationship between {pfa, pmd} and {ps, pc}.

Property : PHY-MAC Translation under the Local Interference Model

P.1 For both types of applications, PC = PMD.
P.2 For applications with guaranteed delivery, correctly detected opportunities may lead to
failed data transmission, and miss detections may lead to successful data transmission; that
is, Pr[success | H0] ≤ 1 − PFA, 0 < Pr[success | H1] ≤ PMD.
P.3 For best-effort delivery, correctly detected opportunities always result in successful data
transmission, and miss detections may also lead to successful data transmission; that is,
Pr[success | H0] = 1 − PFA, 0 < Pr[success | H1] ≤ PMD.

Impact of MAC Handshaking

Detecting spectrum opportunities from detecting primary signals resembles the hidden and
exposed terminal problem in the conventional ad hoc networks of peer users.
It is therefore natural to consider the use of a RTS/CTS handshaking to enhance the detection
performance of LBT.
RTS/CTS signaling can improve the performance of opportunity detection at the physical
layer, it may lead to decreased throughput at the mac layer for best-effort delivery
applications.

Property : PHY-MAC Translation with RTS/CTS Signaling

P.1 PC = [Pr[H1] / Pr{I(A,rx)}] PMD ≥ PMD.
P.2 Correctly detected opportunities always result in successful data transmission, as well as
miss detections; that is, PS = (1 − PFA) Pr[H0] + PMD Pr[H1].
SENSING ACCURACY VERSUS SENSING OVERHEAD
Increasing the sensing time improves the fidelity of the sensing outcomes, thus reducing
overlooked spectrum opportunities.
On the other hand, increasing the sensing time results in less transmission time.
The trade-off between sensing accuracy and sensing overhead depends on the SNR level, the
duration of spectrum opportunities, and the interference constraint ζ
Consider a slotted primary network. Let N represent the slot length in some arbitrary units,
and n the duration of the sensing window. Let PFA(n) and PMD(n) denote the performance
metrics based on a sensing window of length n.
Assuming that the channel is free, the fractional time that the channel will be accessed is 1 −
PFA(n), and the fractional slot available for such transmission is (N − n)/N.
A spectral efficiency metric defined as η(n) = [N − n / N ] [1 − PFA(n)]
For a specified PD (interference constraint), PFA can be written as PFA(n) = Q [(1 +
snr)Q−1(PD) + snr√n] which decreases monotonically in n.
The metric thus consists of two terms:
The first one decreases as n increases and the second one increases with n

(Image source : Cognitive Radio Communications and Networks)

Figure shows throughput η versus sensing window duration n for various values of the
interference constraint, δ, and various values of SNR: −15, −10, −5, 0, 5 dB depicted by
markers x, o, +, , and ♦, respectively.
 As the interference constraint becomes tighter (i.e., as the specified δ decreases), the optimal
n∗ increases and η ∗ decreases.
As SNR increases, the required sensing window length decreases and efficiency increases.
SPECTRUM SHARING MODELS OF DYNAMIC SPECTRUM ACCESS -
UNLICENSED AND LICENSED SPECTRUM SHARING
UNLICENSED SPECTRUM SHARING
Unlicensed frequency bands are chunks of spectrum set aside to be used by devices that wish
to operate in a way that is not constricted by licenses and the associated complicated
transmission/ownership rules and are therefore prone to interference.
The most commonly used unlicensed bands are the 2.4 GHz ISM band, used by
IEEE 802.11 b/g/n and Bluetooth devices, and the 5GHz UNII band, as used by IEEE 802.11a
and the European Hyper LAN standards.
Various spectrum measurement campaigns have shown unlicensed bands to be among the
most heavily used frequency bands.

LICENSED SPECTRUM SHARING

Licensed bands are frequency bands assigned exclusively to a licensee, for instance, a specific
mobile operator. Traditionally, such a license award also stipulates a specific technology to be
used in the band, for example, GSM or UMTS.
More recently, however, regulators such as ofcom in the united kingdom have shown an
interest in assigning spectrum bands as “technology neutral.
” Depending on the technology used in the licensed band of a specific service provider,
various mac techniques are used to allow end users to share the medium.

FUNDAMENTAL LIMITS OF COGNITIVE RADIO (REF. BOOK: 2)

Spectrum leasing: Allowing unlicensed users to lease any part of or all the spectrum of a
licensed user.
Dynamic spectrum leasing: Temporary and opportunistic usage of spectrum rather than a
longer term sublease.
‘Private commons’: A licensee could allow unlicensed users access to his/her spectrum
without a contract, optionally with an access fee.
Interruptible spectrum leasing: Suitable for a leaser that wants a high level of assurance that
any spectrum temporarily in use, or leased, to an incumbent cognitive radio could be
efficiently reclaimed if needed.
(Image Source: Cognitive Radio Networks)
Cognitive radios have the ability to listen to the surrounding wireless channel, make decisions
on the fly and encode using a variety of schemes. In order to exploit these abilities fully, first
consider the simplest example, shown in Figure , of a channel in which a cognitive radio
device could be used in order to improve spectral efficiency.

As shown on the left, suppose sender X1 is transmitting over the wireless channel to receiver
Y1, and a second incumbent user, X2, wishes to transmit to a second receiver, Y2.

In the current secondary spectrum licensing proposals, the incumbent user X2, a cognitive
radio that is able to sense the presence of other transmitting users, would either wait until X1
has finished transmitting before proceeding, or possibly transmit over a different frequency
band.

Rather than forcing X2 to wait, it has been suggested to allowX2 to transmit simultaneously
with user X1 at the same time in the same band of frequencies.

The wireless nature of the channel will make interference between simultaneously
transmitting users unavoidable. However, by making use of the capabilities of a cognitive
radio, it is shown that the cognitive radio is able to potentially mitigate the interference.