Radio Signal Modulation Recognition And System Design Based On Big Data

With the development of the fifth-generation mobile communication technology,in the face of spectrum reuse in limited spectrum resources,it is particularly important to develop new technologies for signal modulation recognition and classification.Based on the deep learning algorithm,this paper proposes two signal recognition and classification models to replaces the traditional feature extraction method designed by experts.In addition,based on these two models,this paper designs a signal recognition and classification software system that can efficiently process massive data.Utilizing Hadoop and distributed clusters,the system is highly available,highly reliable,highly secure,highly scalable and low latency.The main innovations of this paper are as follows:1?Proposed a signal recognition and classification model based on Inception.The Inception network structure is used to extract data features of different sizes,which makes the network more capable of fitting.At the same time,the dense structure is used to approximate the optimal local sparse structure,so that the network can maintain good sparsity and utilize the high computational performance of dense matrix.The Inception network model proposed in this paper can accurately recognize and classify 17 kinds of mixed analog and digital signals.2?Proposed a signal recognition and classification model based on LSTM.The LSTM network introduces the loopback structure and the timing of the input signal sequence,making the network memorable.This feature of the LSTM network allows for better feature extraction and sample fitting of analog signals.Compared with the Inception network model,the LSTM network model can improve the recognition accuracy of analog signals in condition of the high recognition accuracy of digital signal.3?Designed a signal recognition and classification software system capable of efficiently processing massive data.The software system is designed based on Hadoop and SSM,including signal recognition and classification,standard signal library,authority management and etc.Based on the two signal recognition and classification models proposed above,Hadoop combines HDFS and Spark to provide high-performance computing services for the whole system.The Java Web application built by SSM framework based on MVC principle provides interactive interfaces and business logic processing.At the same time,in order to ensure the ability of software system to process massive amounts of data,the system uses distributed clusters,including interface service clusters,MySQL data storage clusters,Redis cache clusters,and so on.