LAUSR

Spectrum sensing is an indispensable technology for cognitive radio networks, which enables secondary users (SUs) to discover spectrum holes and to opportunistically use under-utilized channels without causing interference to primary users. Aim at improving the sensing performance, a multi-antenna spectrum sensing scheme based on main information extraction and genetic algorithm clustering (MIEGAC) is proposed in this paper. Specifically, in order to reduce the amount of signal that is transferred to the fusion center, an information pre-processing scheme based on principal component analysis (PCA) is presented. Main information from the sensing signal is extracted via PCA, which reduces the cost of the reporting channel and the impact of interfering information on detection result. Furthermore, an information fusion method is described in this paper, which takes the place of complicated matrix decomposition algorithms. Moreover, inspired by machine learning, a clustering scheme based on genetic algorithm is introduced to classify signal features, which implements the spectrum sensing decision and avoids calculating the decision threshold. Simulation results illustrate that the MIEGAC can considerably improve the sensing performance for spectrum sensing. Significantly, this paper provides a novel approach for the design of centralized spectrum sensing algorithms in cognitive radio technologies.