Research On Modulation Analysis Of Digital Receiver In α Stable Distribution Noise

Modulation analysis technology is a key technology in the signal processing process of the digital receivers,which is between detection and demodulation,and plays a vital role in communication reconnaissance,electronic countermeasures,spectrum supervision,and other fields.Traditional modulation analysis is usually studied with Gaussian distribution as the background noise model,while in many practical communication scenarios,the channel noise exhibits significant impulsive characteristics.Numerous studies have confirmed that αstable distribution is the most suitable model to describe this type of impulsive noise.Underα stable distribution noise,the performance of the modulation analysis algorithm designed based on the Gaussian distribution assumption degrades or even fails completely.Therefore,from the perspective of non-Gaussian signal processing,this paper aims to improve the accuracy and robustness of parameter estimation and modulation identification at the digital receiving end based on α stable distribution as the noise model,and deeply studies and optimizes the modulation analysis algorithm.Firstly,aiming at the problem that traditional α stable distribution noise suppression methods have high dependence on the accuracy of channel estimation and limited noise suppression ability,this paper proposes an improved normalized compression transformation function based on the idea of nonlinear mapping,which does not require any noise prior information,and can suppress large outliers while retaining the effective information of the signal.Secondly,this paper combines the improved normalized compression transformation function with cyclic correlation spectrum to construct a new generalized cyclic statistic,and introduces the weight function to reduce the influence of truncation noise on discrete spectral line extraction.The simulation results show that the proposed algorithm has better estimation performance in low mixed signal-to-noise ratio,strong impulsive noise,and few sampling points without increasing computational overhead,and it provides a solid technical support for the subsequent research of modulation identification.Finally,aiming at the problems of dynamic changes of the channel environment and an insufficient number of obtained signal samples in the non-cooperative communication system,which leads to the decline of the recognition accuracy and the weak generalization ability of the classifier,a small-sample modulation recognition optimization algorithm based on the optimal selection of impulsive noise robustness features is proposed.This paper constructs the evaluation indexes of the sensitivity of features to impulsive noise and extract multi-dimensional raw features.The robust feature screening is carried out by combining K-means clustering and neighborhood rough set,which can reduce the number of feature dimensions and eliminate redundancy between features to ensure that the indexability between each signal is maximized.The support vector machine(SVM)for quantum particle swarm optimization is integrated into a strong classifier,which reduces the dependence of the base classifier on hyperparameters and improves the learning ability and stability of a single classifier.The simulation results show that the proposed algorithm only requires a small number of training samples,which can greatly improve the generalization ability of the classifier in the dynamic impulsive noise environment and have higher recognition accuracy.The results of this paper are helpful to promote the development of modulation analysis under dynamic impulsive noise and small sample conditions,enrich and expand the research ideas and methods in the field of non-Gaussian signal processing.Meanwhile,the research results have important theoretical significance and application value for communication reconnaissance and intelligence processing at non-cooperative digital receivers.