Cooperative spectrum sensing: performance analysis and algorithms

The employment of cognitive (intelligent) radios presents an opportunity to efficiently
use the scarce spectrum with the condition that it causes a minimal disturbance
to the primary user. So the cognitive or secondary users use spectrum sensing
to detect the presence of primary user.
In this thesis, different aspects related to spectrum sensing and cognitive radio
performance are theoretically studied for the discussion and in most cases, closedform
expressions are derived. Simulations results are also provided to verify the
derivations.
Firstly, robust spectrum sensing techniques are proposed considering some realistic
conditions, such as carrier frequency offset (CFO) and phase noise (PN).
These techniques are called the block-coherent detector (N2
-BLCD), the secondorder
matched filter-I (SOMF-I) and the second-order matched filter-II (SOMF-II).
The effect of CFO on N2
-BLCD and SOMF-I is evaluated theoretically and by simulation
for SOMF-II. However, the effect of PN is only evaluated by simulation for
all proposed techniques.
Secondly, the detection performance of an energy detector (ED) is analytically
investigated over a Nakagami-m frequency-selective (NFS) channel.
Thirdly, the energy efficiency aspect of cooperative spectrum sensing is addressed,
whereby the energy expenditure is reduced when secondary users report their test
statistics to the fusion center (FC). To alleviate the energy consumption overhead,
a censored selection combining based power censoring (CSCPC) is proposed. The
accomplishment of energy saving is conducted by not sending the test statistic that
does not contain robust information or it requires a lot of transmit power. The detection
performance of the CSCPC is analytically derived using stochastic geometry
tools and verified by simulation. Simulation results show that that the CSCPC
technique can reduce the energy consumption compared with the conventional techniques
while a detection performance distortion remains negligible.
Finally, an analytical evaluation for the cognitive radio performance is presented
while taking into consideration realistic issues, such as noise uncertainty (NU) and
NFS channel. In the evaluation, sensing-throughput tradeoff is used as an examination
metric. The results illustrate the NU badly affects the performance, but the
performance may improve when the number of multipath increases.