The Research Of Radio Signal Classification Based On BP Neural Network

With the rapid development of modern science and technology and information technology, the radio spectrum resources has become an important resource for the widespread use of human society. Increasingly wide range of radio applications, the radio spectrum resources are becoming more and more nervous, and the supply and demand contradictions have become increasingly prominent. At the same time also cause that the interference phenomenon gradually increased. In order to can better maintain radio spectrum monitoring and management, we should timely recognition and classification, and good anti-jamming measures for the detected radio signals. Therefore, the accurate classification and recognition of the radio signal has important practical significance in radio monitoring.The applications of pattern recognition and neural network in many technological fields have achieved rapid promotion, and promoted the development of artificial intelligence systems. The classification and recognition of the radio signal is a typical pattern recognition applications, because of the existence of a large number of abnormal signal of radio signals, these abnormal signals were accurately classified recognition is crucial. Therefore, classification and recognition of radio signal has important realistic significance and practical application value on radio monitoring. The main achievements of the thesis are summarized as follows:(i) We summarize the relevant theoretical knowledge of pattern recognition and neural network, introduce the design of the structure of BP neural network model, the principle of BP algorithm and its advantages and disadvantages, and also detailed instruction the algorithm improvements;(ii) In view of the complexity and particularity of the radio environment, we analyze the data preprocessing before the classification model design in the actual background of radio signal monitoring;(iii) Based on the improved BP algorithm and characteristics of radio signals, we construct a suitable classifier model and verify the effectiveness and feasibility of the model by comparing experimental.