A Hybrid Spectrum Sensing Method for Cognitive Sensor Networks

Abstract

Existing spectrum sensing methods for cognitive radio do not consider the secondary network’s characteristics to obtain the frequency of spectrum sensing, i.e., the sensing period would be identical for secondary networks that have different traffic characteristics. In this paper, a hybrid sensing algorithm is proposed that finds the optimal sensing period based on both primary and secondary networks’ properties. A continuous-time Markov chain system is used to accurately model the spectrum occupancy, and a novel method is proposed that adaptively varies its parameters to avoid unnecessary sensing tasks, while guaranteeing the priority of the primary network. We conduct simulation work to evaluate the performance of the proposed method. It is shown that the proposed technique outperforms the non-hybrid approach with respect to sensing efficiency and energy consumption. A cognitive sensor network is also considered based on IEEE 802.15.4/ZigBee radios, and it is shown that significant energy savings can be achieved by the proposed method.

Appendix: Derivation of \(T_I\)

The Lambert W-function is used to derive $T_I$ from Eq. (15). Firstly, Eq. (15) is written as $\zeta = \frac{1-e^ {-\beta T_I}}{\beta T_I}$, where $\zeta$ is the minimum acceptable $AOR$. Assume $x=\beta T_I$, then

$$\begin{aligned} \zeta x = 1 - e^{-x}. \end{aligned}$$

(19)

Using the substitution $(t = x - \frac{1}{\zeta })$ yields to

$$\begin{aligned} te^t = \frac{-1}{\zeta e^{\frac{1}{\zeta }}}. \end{aligned}$$

(20)

Secondly, we use the following property of Lambert W-function [27]

$$\begin{aligned} Y = Xe^x \Leftrightarrow X = W(Y). \end{aligned}$$

(21)

Therefore, $t = \mathbf W \left( \frac{-1}{\zeta e^{\frac{1}{\zeta }}}\right)$, or $T_I$ is given by

$$\begin{aligned} T_I = \frac{1}{\beta }\left[ \mathbf W \left( \frac{-1}{\zeta e^{\frac{1}{\zeta }}}\right) +\frac{1}{\zeta }\right] . \end{aligned}$$

(22)