The Research Of Efficient And Robusrt Spectrum Sensing Algorithm Based On Goodness Of Fit

Spectrum sensing is one of the key aspects of the cognitive radio technology,and efficient and robust spectrum sensing technologies are extremely crucial for the whole cognitive radio system to work stably.Goodness-of-fit is widely used in spectrum sensing.Usually,the fitting object and the fitting criteria influence the performance of the detection algorithms based on goodness-of-fit.In this paper the spectrum sensing technologies are studied from the two points mentioned above.The specific work is summarized as follows:The classical unilateral right-tail Anderson-Darling(URAD)test is not accurate enough when the sample points are limited.Based on the analysis of the principle of URAD test,the accurate threshold of URAD test is derived by using the theorem that the cumulative distribution function of any continuous function obeys the uniform distribution of features from 0 to 1 and properties of Gamma distribution.Using of the accurate threshold in the detection algorithm overcomes the shortcoming of URAD detection algorithm under the condition where the sample points are limited.The classical URAD test constructs the fitting object only through a single sample point,which leads to the certain limitation in application.Aiming at this problem,the fitting object is constructed by multiple sample points,and the expression of the decision threshold is derived by using the properties of the chi-square distribution.Meanwhile the influence of the number of packets on the detection performance is studied.The classical URAD test algorithm requires the power of noise and is susceptible to noise uncertainty,which leads to the limitation in practical applications.A fitting object that does not require the power of noise is constructed in this paper using the knowledge of random matrix theory and a detection algorithm is derived based on the ratio of the maximum eigenvalue to the mean of the trace,which overcomes the influence of noise uncertainty while the good detection performance is ensured.Theoretical research and simulation results show that the accurate threshold derived in this paper effectively improves the accuracy of the URAD fitting criterion and breaks the limitations of the classical URAD detection algorithm.The performance of the proposed algorithm is efficient and robust.The research in this paper provides a reliable spectrum sensing strategy for cognitive radio systems,which is of the certain theoretical and practical significance.