Research On Blind Multi-band Spectrum Sensing Algorithm In The Communication System Of Smart Grid

Compared with the traditional power grid,the data and information to be transmitted in smart grid is growing explosively,and the traditional power communication system can not meet the requirements of smart grid for data and information transmission.To solve this problem,scholars apply cognitive radio technology to smart grid communication system.Therefore,various data and information generated randomly in smart grid can be transmitted timely and accurately.However,the spectrum sensing methods used in smart grid communication systems are all narrow-band detection methods,which can not meet the real-time requirements of the smart grid for information transmission system.So multi-band spectrum detection technology utilized to simultaneously detect the spectrum holes in multiple frequency bands,has attracted wide attention.In this paper,a series of research work on how to design blind multi-band spectrum sensing algorithm for smart grid communication system is carried out,and the chapters are arranged as follows.Accuracy is an important performance indicator for spectrum sensing in smart grid communication systems.Applying the mature narrowband spectrum sensing algorithm to the multi-band spectrum sensing technology by single band serial processing,the purpose of fast and accurate detection for the multi-band spectrum is achieved.Hence,in the second chapter focuses on the improvement of the traditional spectrum sensing algorithm based on the eigenvalues of covariance matrix.Firstly,in order to obtain the more accurate decision threshold,improved the existing semi-blind spectrum sensing algorithm based on the maximum eigenvalue of the covariance matrix.Secondly,the traditional semi-blind spectrum sensing algorithms using the maximum and minimum eigenvalue are not suitable for the color noise sensing scene,which need to prewhiten the color noise,and their detection results are no longer reliable,so a new semi-blind spectrum sensing algorithm is proposed.At last,by the latest research conclusions of the large dimension random matrix theory,a more accurate calculating method for decision threshold is proposed to improve the existing blind spectrum sensing algorithm based on the ratio of maximum and minimum eigenvalue.The new method obtained the thresholds using the approximate distribution of the maximum eigenvalue and the minimum eigenvalue respectively,and then the average value of the two thresholds is taken as the final decision threshold.Thus the more reliable spectrum sensing result is obtained.Fast sensing of idle spectrum is another important requirement for spectrum sensing in smart grid communication systems.The multi-band spectrum sensing technology by serial processing has stable and reliable detection effect,but the disadvantage of low detection efficiency makes it not suitable for high real-time sensing scene.In the third chapter,a parallel blind multi-band spectrum sensing algorithm based on principal component analysis is presented.In view of the shortcoming of low detection efficiency in multi-band spectrum sensing algorithm by serial processing,a continuous hypothesis testing model H_k and a decision measure using generalized likelihood ratio are established to detect multiple subbands at the same time.Owing to spectrum sensing should be completed in a very short time to meet the real-time requirements of the detection system,an improved parallel multi-band blind spectrum sensing algorithm based on principal component analysis is put forword.The results of theoretical analysis and simulation experiments show that the improved algorithm still maintains the stable detection performance with low computational complexity and high detection efficiency.Even in the sensing scene of low SNR and small sample number,the detection probability of the improved algorithm is greater than that of the original algorithm.Compared with white noise,it is more realistic to use color noise to characterize the noise environment of smart grid communication systems.Parallel blind multiband spectral sensing algorithm based on information theory or principal component analysis is not suitable for color noise sensing scenarios.Combining the matrix perturbation theory and spatial distance measurement with the spectrum sensing,a parallel multi-band blind spectrum sensing algorithm utilizing rank criterion is presented.In the white noise sensing scene,when the SNR is low,the detection probability of the spectrum sensing algorithm by the information theory criterion is greater than that based on the rank criterion,but the latter shows robust detection performance in the color noise sensing scene.What’s more,with the increasement of sample number and SNR,its detection performance becomes better.It is theoretically proved that the optimal estimation value of the eigenvalue of the sampling covariance matrix is the power of the subband signal,so the spectrum sensing algorithm based on the rank criterion is improved according to this.This algorithm no needs to calculate the eigenvalues of the sampling covariance matrix,but directly uses the power of the subband signal to construct the decision rule,thus it greatly reduces the computational complexity.In case of white and color noise sensing scenarios,comparing with the other three algorithms,the improved spectrum sensing algorithm has optimum detection performance when the sample number is small and the SNR is low.Considering the low SNR problem in smart grid communication systems,a parallel blind multi-band spectrum sensing algorithm utilizing clustering analysis is proposed.When the transmitted powers of the primary user signals are exactly the same,the parallel blind multi-band spectrum sensing algorithm by K-means clustering is used to classify the subbands as the occupied subbands and the idle subbands.It is simple to calculate,and shows robustness in complex sensing scenes,such as noise power uncertainty,color noise,low signal to noise ratio,and small sample sensing scenes.The multi-band spectrum sensing algorithm by K-mean clustering has been greatly affected when the transmission powers of the primary user signals are not completely consistent,which shows sharp decline about detection performance.Threrfore,a parallel blind multi-band spectrum sensing algorithm based on fuzzy C-means clustering is put forward.It has stable and efficient detection performance when the transmission powers of the primary user signals are not completely consistent.Considering the complexity and variability of the detection environment,the signals are transmitted through the Rayleigh fading channel,and then analyze the adaptability of the proposed algorithm.Compared with other algorithms,it shows robust spectrum sensing performance in the scenario of Rayleigh fading and low SNR.