Research On Joint Spectrum Sensing And Primary User Localization In Cognitive Radio

In recent years, with the rapid development of mobile communications, the demand forspectrum resources is also growing. At the same time, the spectrum resources which can’t be fullyutilized is led by the fixed spectrum allocation policies, so the spectrum resources is becomingscarce. Cognitive radio can greatly improve the efficiency of utilization of spectrum by finding thespectrum holes and dynamic spectrum access. Spectrum sensing is the premise and basis of thecognitive radio. However, in a more complex cognitive radio networks, the secondary user needs toknow more than just the status of the occupied spectrum. The main user locations and transmitpower and other information are also needed to detect. Without this information, the normalcommunication of primary users will inevitably be affected. This thesis mainly studies the problemof spectrum sensing in cognitive radio and localization of the primary users.In this thesis, the wideband spectrum is divided into adjacent but disjoint narrowband spectrum,the primary user’s location information and transmit power is fused into spectrum sensing modelthrough the free path loss model. A cooperative algorithm based on clustering and bayesiancompressive sensing is proposed to solve the problem of broadband spectrum detection andlocalization of the primary users. The method of clustering was used to reduce the computationalburden of the fusion center and time delay of the system. The method of bayesian compressivesensing was used to recover the original signal. It can alleviate the sampling rates and reduce theinterference of the noise to the detection results. The simulation results show that this algorithm caneffectively recover the original signal even when the compressive ratio is low. We can obtain notonly the spectrum information but also the accurate position of the primary users.In addition, in order to improve the detection performance, an adaptive measuring algorithm isproposed based on the error bars. When the spectrum environment changes, the error bars which isgenerated in the process of bayesian compressive sensing is also changing. The error bars will becompared with a preset value in the calculation process. This algorithm can adaptively adjust thenumber of measurements according to the result of comparison. Finally, it can select a compromisebetween reconstruction time and detection error. The simulation results show that this algorithm canrealize the adaptive spectrum detection and the localization of the primary users without sacrificingthe reconstruction error. It can also improve the performance of the spectrum sensing system.