Study On Automatic Modulation Recognition Algorithms Of Digital Modulated Signals

Digital communication signal's modulation can be automatically recognized by modulation recognition method. It is critical to identify different modulation signals effectively in the non-cooperative communication occasions including military and civilian and in the cooperative communication occasions including adaptive OFDM systems and wireless automatic routing distribution network and so on. In a word, automatic digital modulation recognition plays a vital role in digital communication.It is shown in literatures published at home and abroad that the theory of automatic digital modulation recognition becomes increasingly rich and perfect. The best-studied question is how to identify modulation mode in low signal to noise ratio (SNR). And, most of literatures focus on artificial neural network, higher-order accumulation, wavelet improvement, support vector machine (SVM), instantaneous information. In consideration of complexity and performance, the recognition algorithm based on instantaneous which uses wavelet improvement is discussed in the paper. The paper is mainly about:Firstly, the background and development status of the automatic digital modulation recognition are introduced, the algorithms of wavelet transfer are presented, and the methods of modulation are analyzed and proven by constellation diagrams. Based on the theory of parameter estimation of signals which belongs to pre-processing, simulations are given and the method which fits the essay best is chose.Secondly, the algorithms of wavelet modulus maxima de-nosing and traditional wavelet threshold de-noising are analyzed and simulated. The wavelet threshold de-noising algorithm is key duty in this segment and the improved method proposed by the paper involves selection of optimal wavelet, non-adaptive method to obtain optimal decomposition level, adaptive algorithm to get suitable threshold function.Lastly, both the recognition algorithm based on instantaneous information and the algorithm based on wavelet transfer are researched. The proposed method is used to de-noise instantaneous information for the first time and two improved features are applied for the first time in recognition algorithm based on instantaneous information. Results indicate that the recognition performance in low SNR is improved greatly with the innovations mentioned above. Meanwhile, it is evident from simulations that the correct recognition rate when using the algorithm based on wavelet transfer is receptible, for input SNR is equal or greater than 10 dB.