IEEE Signal Processing Magazine - January 2018 - 158

1.4

× 105
4

0
-5

1

-10
-15

0.8

-20

0.6

-25

0.4
50
f

-50
-100
100

50
0
0
-5
00
-1
50
-1

0

0
15
0
10

3
2
1
0
100

1.2

-30

0.2
0
-150 -100 -50

α

(a)

-35
0
α
(b)

50

100

150

-40
-10 -8 -6 -4 -2 0 2 4
f
(c)

6

8 10

FIGURE 21. The carrier frequency and bandwidth estimation from the cyclic spectrum: (a) preprocessing, (b) thresholding and clustering, and (c) parameter estimation.

1

0.8

0.6

0.4

0
f
(c)

0.2

-0.2

× 109

Power Spectrum (µ = 0)

-0.4

f
(a)

× 105

-0.6

1
× 109 0.5
α 0

-1
-0.8
-0.6
-0.4
-0.2
0
0.2
0.4
0.6
0.8
1

0
2
1.5

4.5
4
3.5
3
2.5
2
1.5
1
0.5
0

-1

0.5

Cyclic Spectrum for f = 0
1
0.9
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0
-2 -1.5 -1 -0.5 0 0.5 1 1.5 2
α
× 109
α = 2fc
(b)

-0.8

Cyclic Spectrum
1

× 109

FIGURE 22. A screenshot from the MWC with cyclostationary detection. The input signal is composed of N sig = 3 transmissions (or K = 6 bands) with carri-

ers f1 = 320, f2 = 600, and f3 = 860 MHz. (a) The recovered cyclic spectrum from low-rate samples; (b) the cyclic-spectrum profile at the angular frequency
f = 0, where the cyclic peaks are clearly visible at twice the carrier frequencies; and (c) the power-spectrum recovery failing in the presence of noise.

sensing success is still achievable at SNRs that are 15-dB
lower than those allowed energy detection performed on the
recovered spectrum or power spectrum. Representative results
shown in Figure 22 demonstrate the advantage of cyclostationary detection over energy detection in the presence of
noise. The figure presents the reconstructed cyclic spectrum
from samples obtained using the MWC prototype as well as
cross sections at f = 0 and a = 0, which correspond to the
power spectrum. This increased robustness to noise comes at
the expense of more involved digital processing on the lowrate samples. The additional complexity stems from the higher
dimensionality involved, because we reconstruct the twodimensional (2-D) cyclic spectrum rather than the one-dimensional (power) spectrum.

Collaborative-spectrum sensing
Collaborative model
Until now, we assumed a direct observation of the spectrum.
In practice, the task of spectrum sensing for CR is further
complicated due to physical channel effects, such as path loss,
158

fading, and shadowing [34]-[36], as described in "Channel
Fading and Shadowing." To overcome these practical issues,
collaborative CR networks have been considered, where different users share their sensing results and cooperatively
decide on the licensed spectrum occupancy.
The different collaborative approaches can be distinguished
according to several criteria [34]. First, cooperation can be
either centralized or distributed. In centralized settings, the
data streams are sent to a fusion center that combines them
to jointly estimate the spectrum or determine its occupancy.
In the distributed approach, the CRs communicate among
themselves and iteratively converge to a common estimate or
decision. While centralized cooperation does not require iterations and can reach the optimal estimate based on the shared
data, convergence to this estimate is not always guaranteed in
its distributed counterpart. However, the latter is less power
hungry and more robust for node and link failures, increasing
the network survivability. An additional criterion concerns the
shared data type; the CRs may share local binary decisions on
the spectrum occupation (hard decision) or a portion of their
samples (soft decision).

IEEE SIGNAL PROCESSING MAGAZINE

|

January 2018

|
Table of Contents for the Digital Edition of IEEE Signal Processing Magazine - January 2018
