Recognition Of Modulated Signals In Alpha-stable Noise And Fading Channels

At the receiving terminal of the communication system,the recognition of modulated signals is very critical.This process judges the modulation type of the received signal,which is of vital importance to some subsequent processing of the signal.In the civil communication and military communication,the application of the recognition of modulated signals is very common.Traditional digital modulated signal recognition methods mostly use Gaussian noise as the noise model,while the noise in the actual environment generally has the characteristics of spikes and bursts.Some of the most significant features of this type of noise are that its algebraic tail is significantly thicker than the exponential tail of the Gaussian distribution,and its probability density function is significantly higher than that of the Gaussian distribution.Relying on a single Gaussian distribution model cannot describe the kind of pulse burst process deficiently.The theoretical study has confirmed that the special properties of the alpha stable distribution can describe the noise in the real situation more closely,so it is practical to study the recognition problem under the model.But at present,there are very few studies on the recognition of modulated signals under the frequency selective channel in stable distributed noise model.And there is no good classification performance.Based on the above-mentioned problems,this paper studies the recognition of modulated signals under the frequency selective channel in stable distributed noise model.The main contents are as follows:Under the alpha stable distribution noise,a recognition method based on the Fractional Low-order Ambiguity Function of the digital modulated signal is proposed.The method first obtains the Fractional Low-order Ambiguity Function for all the signal types in the alternative set.Then the section in which the Doppler frequency shift is zero is intercepted and transformed into an image.Then fill the area below the edge of the image with a certain color,making it a color image.After a series of image processing to simplify the difficulty,the Zernike moments of the image are extracted as the feature vector.Finally,the probabilistic neural network classifier is utilized to identify the digital modulated signals.The simulation results show that the recognition rates of the candidate signals are all above 95% when the mixed SNR is 8d B.And the recognition rate of the method is not significantly decreased in the Alpha stable distributed noise multipath channel,indicating that the method can effectively combat the impact of multipath.Using the alpha stable distribution as the model of additive noise,a classification method based on likelihood function is proposed.Firstly,the kernel density estimation algorithm is used to estimate the probability density function of the alpha stable distributed noise.Secondly,the channel parameter is estimated by the expected condition maximization(ECM)algorithm.Thiedly the logarithmic likelihood function is constructed for each type of signal,and the recognition of the signal is identified by the composite hypothesis test.The recognition scheme corresponding to the assumption that the logarithmic likelihood function is maximized is judged as the signal type.Finally,the signal classification under the non-Gaussian noise and the frequency selective channel is realized.The simulation results show that the accuracy of the classification of each signal type in the candidate set is not less than 95% when the mixed SNR is 10 d B,which proves that the signal classification method has high accuracy,and that it has better robustness.