In The Cognitive Radio Spectrum Perception Research Based On Covariance Matrix

Cognitive radio can exploit and reuse the licensed but underutilizedspectrum, which is recognized as the most promising next-generation solutionto improve spectrum efficiency. Spectrum sensing is a fundamental componentin cognitive radio, through which the cognitive radio users find and access thespectrum hole and monitor the activity of the primary user in order to avoidinterfering with primary user. Up to now, many spectrum sensing algorithmshave been proposed, but most of the existing algorithms suffer from lowdegree of detection accuracy and long detection time due to noise uncertaintyand the mobility of the communication terminal. Spectrum sensing is still thehottest research field in cognitive radio.In this thesis, we propose two type of cooperative spectrum sensingalgorithm based on the statistical covariance matrices of the received signal toimprove the detection performance in the environment of noise uncertainty andcommunication terminal mobility.Firstly, we proposed a cooperative spectrum sensing based on averageeigenvalue detection (AED) algorithm. In this scheme, the sample covariancematrix is first computed according to the received signals, then the averagevalue of the eigenvalues of the sample covariance matrix is obtained to be thestatistic of the detector. AED algorithm has the inherent advantages ofeigenvalue detection algorithm, such as no needs of prior knowledge of theprimary user, being independent of the noise power and being easy to beachieved. Furthermore the AED algorithm is more robust in the actualtime-varying noise power environment than other eigenvalue based algorithm.Secondly, we proposed a sequential cooperative spectrum sensingalgorithm by combining the sequential test with AED algorithm. In thisscheme the numbers of the cognitive radio users and samples are adjusteddynamically according to the change of the environment. This scheme canachieve the detection accuracy require by using the least cognitive radio users and samples, therefore it can shorten detection time and improve detectionperformance.