Research On Intelligence Spectrum Sensing Algorithm Based On Signal Energy Distribution

In recent years,wireless communication has provided convenience to human beings,and the proportion of wireless services in people's lives has gradually increased.Therefore,wireless spectrum resources have become increasingly tense.Cognitive radio technology uses idle spectrum to provide more access opportunities for increasing spectrum utilization.Spectrum sensing is an important part of cognitive radio.The signal-to-noise ratio has a greater impact on the performance of traditional spectrum sensing algorithms,and the reduction of the signal-to-noise ratio will reduce the performance.This paper studies the spectrum sensing algorithm.In the aspect of signal feature distribution,combined with neural network technology,it proposes a research on spectrum sensing algorithm based on long short-term memory network and goodness of fit of distribution of signal energy and a neighbor blind sensing algorithm based on signal energy distribution.The main contributions of the thesis are as follows:First,this thesis proposes a spectrum sensing algorithm which ueses long short-term memory network to train the goodness of fit of signal energy distribution.This method calculates the distance value form the energy distribution of the received signal when primary user signals exist.Then the feature vector consists of the distance value is input into a LSTM network training to obtain a model.Finally,test data is input into the model for prediction to achieve sensing.From the results,the new method proposed can reach a detection probability of 96.21% when the SNR is-13 d B and the number of sampling points is 28,which is significantly better than the traditional energy detection algorithm and the traditional goodness-of-fit algorithm.Secondly,this thesis further explores the characteristics of signal energy distribution to improve spectrum sensing.Most sensing algorithms require prior knowledge,such as signal-to-noise ratio,signal energy,signal type in advance.The specific model is only suitable for the corresponding environment,so the adaptability is not strong.This thesis proposed,when prior knowledge is not required,by adding the proximity algorithm in the machine learning,blind perception of the signal can be realized.From the results,the algorithm's detection probability can go to 0.9 with a signal-tonoise ratio of-13 DB,a K value of 7,a sampling number of 300,and a training sample number of350,which is better than the traditional blind perception algorithm.