Performance of centralized data-fusion cooperative eigenvalue-based spectrum sensing under correlated shadowed fading

Electromagnetic spectrum scarcity is undoubtedly an issue nowadays given the demand for quality of service of wireless communication systems and its development. It makes necessary to use this resource efficiently. One possible solution is the spectrum sensing task in cognitive radio networks. This paper aims to analyze some cooperative eigenvalue-based spectrum sensing techniques under correlated shadowed fading, once in practice the correlation of the shadowing effect is present depending on the distance of the cognitive radios. To generate such channel it was used the three-dimensional model based on grid points, in which it is possible to create shadowing correlation among the cognitive radios. The spectrum sensing techniques analyzed are the energy detection (ED), the maximum-eigenvalue to geometric-mean (ME-GM); and the arithmetic-to-geometric-mean (A-GM). The results show that the ED technique outperforms the remaining techniques, however it requires the knowledge of the thermal noise power, whereas the ME-GM and the A-GM techniques are noise power estimation independent, and the ME-GM outperforms the A-GM. Additionally, all techniques presents approximately the same performance variation for a given variation in the correlation parameter.