| Cognitive radio is a promising way to effectively improve the spectrum utilizationthrough the secondary use of a licensed frequency band. Spectrum sensing is theprerequisite and key technique of cognitive radio. Secondary users continuouslydetect the channel state and find out spectrum holes to use, without causinginterferes to the primary user.Energy detection is widely used and requires an accurate estimate of the noisepower. However there exists power uncertainty in the actual noise, which will bringdown the effect of energy detection. This paper analyzes the impact that noiseuncertainty has on energy detection, and discusses improved energy detection basedon double threshold.In recent years, random matrix theory (RMT) was introduced into the spectrumsensing areas, which can be very good at solving the problem of noise uncertainty.Current spectrum sensing algorithms based on RMT require a huge number ofsamples, which limits their application. In this paper, we introduce a new RMTalgorithm based on the Cholesky factorization of the covariance matrix (CMCF),which still performs well in situation of small sample size and less secondary usersThen a new cooperative spectrum-sensing algorithm is proposed based on SNRselection in the fusion center, called SNR-CMCF. Simulation results show thatSNR-CMCF inherits the advantages of CMCF and further reduces the number ofsecondary users involved in the final decision-making, which reduces thecomputational complexity indirectly.Although extensive literature focuses on how to improve the accuracy ofspectrum sensing, the security problemsof spectrum sensing receives little attention.Cognitive radio system requires a safe and reliable performance. This paperpreliminarily studies the security issues in cognitive radio and analyses the behaviorloss caused by malicious users. Finally, we propose a algorithm based on timewindow and reputation value. Simulation results show that our algorithm can resistmalicious attacks effectively. |
PDF Full Text Request