Research On Key Techniques Of Spectrum Sensing In Cognitive Radio

At the early stage of wireless communication,the spectrum management organizations adopted static spectrum allocation to prevent interference.Spectrum resource is increasingly consumed as the result of fast development of wireless communication which makes it impossible to satisfy application.It has been shown in researches that many allocated frequencies are not effectively utilized or even idle in certain time interval or geographical space.Cognitive Radio technology has broken the convention of static spectrum allocation and proposed a novel method to “opportunistically” use idle spectrum to resolve the shortage of spectrum by improving the utilization efficiency.The first problem of cognitive radio is spectrum sensing which means finding the right frequency to use.Works have been done in this dissertation on spectrum sensing and several main contributions are listed as follows:1.Energy detection is simple and easy to implement but greatly influenced by noise uncertainty.A multi-antenna blind spectrum sensing algorithm based on distribution of eigenvalue limitation is proposed which is robust to noise uncertainty with low SNR.The sample matrix is constructed using the sample data of multi-antenna and channel cost is reduced more than cooperative sensing algorithm which constructs the sample matrix with multi-node's sample data.Closed form expression of decision threshold using distribution of maximum eigenvalue and constant false alarm rate is deduced when analyzing the relation of average power and the matrix rank of covariance data and considering the influencing of all eigenvalues to the detection performance.Simulation results show that the proposed algorithm outperforms the one which only considering the maximum and minimum eigenvalue of covariance matrix and the stable performance is achievable when noise variance estimation bias exists and the priori information of signal is unknown.2.An energy detection algorithm under non-cooperative networks is presented to get more available frequency and reduce the influence of other users and external interference to the result.Compared with traditional energy detection algorithm,the proposed one can improve detection accuracy by weighing individual array element coefficient to filter out interference and increase SNR.To counter the effect of noise uncertainty,a blind multi-antennas spectrum sensing algorithm is proposed which allows multi-users to coexist in the same space.The proposed algorithm forms the sample data matrix with output data after weighing antenna elements,analyzes the relation of received signal,maximum eigenvalue of covariance matrix and noise variance,and eliminates the effect of noise uncertainty through canceling the noise component in the receiver power and maximum eigenvalue.The closed form expression is deduced.Simulation results show that the proposed algorithm can reduce interference,get more available frequency and eliminate the influence of noise uncertainty.3.To achieve accurate spectrum usage state in low SNR and get more available spectrum,a double branch spectrum sensing algorithm is proposed.According to the linear property of FFT,the stochastic characteristic of each frequency point is deduced.The frequency band is divided into several segments of the same length under low SNR and under high SNR the frequency band is divided into several segments according the Wavelet singularity detection.The proposed algorithm takes the ratio of some segments' average power to the segments' minimum average power as decision statistics.The simulation results show that the proposed algorithm is robust to the noise uncertainty under low SNR and can provide some frequency band usage state.In high SNR,the algorithm can detect possible weak signals,decrease the false detection rate,and provide more frequency to reuse.4.A DOA algorithm based on virtual spatial smoothing is proposed for fully usage of real value charactetritic of non-circular signals.Analysing the real-valued characteristic of non-circular and the limitation of existing algorithms,a new virtual array antenna is constructed which extends the antenna aperture two times than the original array antenna using the conjugate data.Then the virtual array antenna is divided into two sub-arrays which have the same structure with a fixed distance.The two sub-arrays are spatially smoothed to estimate coherent signal's DOA.This method is efficient in calculation because of avoiding search the peak of spectrum.Simulation results show that the method can estimate the coherent signal's DOA through using the conjudate data to extend the antenna aperture and can get better performance than forword/backword spatial smoothing algorithm.And the applicability of the algorithm is obviously better than that of the Data Reconstruction C-ESPRIT Algorithm.