sensing. But the local reliability is only one-side and
sometimes may be inaccurate. Men et al. propose a robust
cooperative spectrum sensing method by applying D–S
theory. Furthermore, considering the faulty nodes in
CRSNs, the reliability of the sensor nodes and the mutual
support between different nodes are considered in [19]. Liu
et al. propose a novel cooperative spectrum sensing
scheme, which is different from traditional cooperative
spectrum sensing, based on D–S theory, using a multi-
modal spectrum sensing to combine the multi-modal
sensing data of the PU signal [20]. One method against
faulty nodes is proposed in [21]. Because of the difference
between unreliable nodes and honest nodes, the reliability
of these nodes can be calculated using the similarity
between sensor nodes, while the nodes with low reliability
should be eliminated from FC. Wang et al. propose a novel
cooperative spectrum sensing method which simultane-
ously considers the current difference of SUs and the sta-
tistical information of each node’s historical behavior. But
it does not consider the residual energy of nodes, which
may lead to nodes failure during data transmissions [22].
Monemian, et al. propose a heuristic algorithm which
periodically selects an appropriate set of sensor nodes with
minimum average energy consumption for CSS. Mean-
while, a sub-optimal algorithm is proposed to reduce the
computational complexity of the heuristic algorithm [23].
Because of the openness, dynamics and uncertainty of the
wireless environment, the channel conditions are dynami-
cally changing, a correlation-aware node selection
scheme is proposed in [24] to adaptively sel ect the
uncorrelated nodes for CSS. In [25], aiming at saving
energy consumption, an optimization framework has also
been proposed to jointly solve the problem of sensing node
selection and decision node selection. When there are
actual channel propagation effects, [ 26] has analyzed and
derived general criteria for decision-approach selection.
Considering that only part information of SUs and PUs is
available, Najimi, et al. propose an energy-efficient node
selection algorithm to minimize energy consumption while
still satisfying the average detection probability [27].
However, most of the above existing works utilize
current information to select nodes or estimate the relia-
bility of each node. Although the current information
reflects the reliability of nodes to some extent, it is only
one-sided, not always accurate, and it is also unnecessary
to estimat e the information of most nodes, due to the
similarity of the sensing results of densely deployed nodes.
Different from the existing efforts, in addition to the cur-
rent information, the historical information of nodes which
reflects their historical reliabilities should also be utilized
in the selection of R-Nodes. Furthermore, to improve the
reliability of spectrum sensing, the reliability evaluation is
introduced to recognize malicious nodes from R-Nodes
before the final decision making based on D–S theory, due
to the dynamic and uncertain character of wireless
environment.
Therefore, in this paper, we propose a novel node
selection scheme for CSS based on D–S theory. It is carried
out in three successive steps, which are node filtering
strategy, R-Nodes selection and decisi on making by the
combination rule of D–S theory.
3 Network model
In order to quantitatively analyze R-Nodes which are
suitable for performing spectrum sensing, this section
presents the considered cooperative spectru m sensing
model.
3.1 Cooperative spectrum sensing
We consider a cooperative spectrum sensing in a dis-
tributed CRSN network as shown in Fig. 1. The CRSN
network consists of a PU, n SUs, and a FC. We assume that
the FC is constant and energy-un constrained; all the bat-
tery-powered nodes can transmit with enough power to
reach FC if needed and the nodes can use power control to
vary the amo unt of transmit power.
In the area of interest, some R-Nodes are selected to
engage in CSS by the R-Nodes selection (details explained
at Sect. 4.2) and send their sensing results to the FC.
According to D–S theory, the FC makes a final decision
whether the channel is occupied by the PU or not and
allocates spectrum resource to sensor nodes.
The local spectrum sensing can be formulated as a
binary hypothesis testing problem, with the null hypothesis
H
0
represents that the PU is absent, and H
1
represents that
the PU is present, that is [28].
Primary User
Fusion Center
Secondary User
Representave Node
Fig. 1 Cooperative spectrum sensing in CRSN
Wireless Networks
123