Aerospace and Electronic Systems Magazine May 2018 - 21

(USRP) hardware platform and the GNU Radio software. Both of
these methods have straightforward analytical definitions and do
not differ much in terms of implementational complexity. The energy detection method is known to be computationally light but
lacking efficiency in very low SNRs [25], [26]. The cyclostationary detector has, in general, the opposite characteristics. The two
detectors are included into a framework, which adapts the sensing
period by using a precalculated efficiency-accuracy trade-off and a
method for estimating the pattern of the PUs operation.
Our novel contributions presented in this article are the following:
1. A performance analysis of real-time practical implementations
of known cyclostationary [27] and energy [28] detectors, including Rayleigh fading channel.
2. Method for the estimation of the channel parameters and analysis of the detection performance integrated with the spectrum
sensing procedure for an obstructed non-line-of-sight (NLOS)
scenario where the receiver has a limited mobility.
3. Functionality for adaptive change of the sensing time through
the addition of a simplified method for the estimation of the PUs
states and analysis of the efficiency-accuracy trade-off for three
different PU configurations implemented in the transmitter.
4. A detailed outline of the characteristics of implementing an
adaptive spectrum sensing algorithm using USRP and GNU
Radio.

BACKGROUND AND RELATED WORK
Spectrum sensing algorithms based on energy and cyclostationary
detectors and their respective advantages and weaknesses, have
been widely researched. In this section, a summary of existing
advances in these topics including practical implementations, is
presented.

ENERGY DETECTION-BASED SPECTRUM SENSING
The energy detector proposed in [29] obtains the decision threshold by setting the probability of false alarm as constant and using
the standard expression for the constant false-alarm method ((13)
in [23]).
MAY - JUNE 2018

Pfa = Q

(

( )

)

2γ + 1Q −1 Pd + M γ ,

where Pd is the target probability of detection, γ is the SNR, M is
the number of observed samples and Q(.) is the complementary
distribution function of the Gaussian variable, i.e.
Q( x) =

1
2π



∞

 t2 
exp  − .
 2

It also implements a method for noise variance estimation. Further improvement is achieved by deriving a new threshold, which
is built upon the traditional one but depends on the signal-to-noise
ratio (SNR) as well [30]. Another approach is proposed in [31]
which introduces upper and lower thresholds, which define a "confused region", in which the algorithm attempts to identify whether
the signal power is closer to the noise-only region or the PU signal region. In [32], the threshold is bound and the bounds are dynamically defined on the basis of the noise uncertainty in order
to increase the efficiency, both, in terms of Pd as well as Pfa. This
proposal for an energy detector (ED) outperforms the traditional
forms of two other more complex spectrum sensing techniques.
The algorithm in [33] combines an energy and a cyclostationary
detectors to improve the overall performance by using the former
in high SNR and the latter in low SNR. The above-mentioned solutions all follow the traditional derivation of the performance of the
ED, which assumes a Gaussian distribution of the received signals
in additive white Gaussian noise (AWGN) channel and very low
SNR (between −20 and 0 dB). They do not implement adaptivity
in terms of the sensing time and only study the detection accuracy.
A generalized ED with adaptations for several fading channels is
derived in [34]. The analysis in this article focuses on assessing the
performance of the detector in the case of diversity reception and
cooperation between multiple SUs. Alternatively, accurate spectrum occupancy evaluation can be achieved through cooperative
spectrum sensing and a decision fusion center with multiple antennas, which processes the observations of the SUs [35]. The authors
in [35] devise multiple access channel protocols which take advantage of the multiple-input multiple-output concept to obtain more
reliable spectrum sensing decision.
In this article, we assume the ED proposed in [28]. It is suitable
for practical implementations, like ours, because it presents deri-

IEEE A&E SYSTEMS MAGAZINE

Table of Contents for the Digital Edition of Aerospace and Electronic Systems Magazine May 2018