| As an important part between signal detection and signal demodulation, the identification of the modulation types of modulation signals aims to determine the modulation type based on some Analysis and processing without any priori knowledge. Its extensive application and application prospects in military and civilian communication fields are endowed with the great value of the study. With the development of digital communication technology, the system and modulation of communication signals are becoming more and more complex and diversified, so it is very significant and difficult to determine the modulation mode of signal.In recent years, many scholars have carried out a lot of research on the field of communication signal modulation recognition, and many achievements have been made. Many classical algorithms have their unique advantages, but also have their limitations, such as the dependence on prior knowledge, the adaptability of the SNR and the small scope of application. So how to overcome these shortcomings is the research object and the main idea of this article.Based on this premise, in order to cover the commonly used modulation signal,the research objects of the algorithm are many kinds of modulation signal, including 14 kinds of digital modulation signals: BPSK, QPSK, OQPSK, 8PSK, 16 PSK, 2ASK, 4ASK, 16 QAM, 64 QAM, 256 QAM, 2FSK, MSK, 5 kinds of analog modulation signals: AM, USB, DSB, LSB, FM and carrier signal CW. It is the core content of all kinds of modulation recognition algorithms to find the appropriate characteristic parameters or statistical characteristics of various signals. In this paper, the feature extraction of the structural features of various signals is mainly based on the characteristics of the signals.Based on the analysis of the characteristics of the signal structure, according to the characteristics of the unique or common features can determine the relative optimal level and order, and in order to establish the automatic identification algorithm of 20 kinds of signals of the treelike classification structure, in order to determine, from large to small class of thinking way. There are 14 kinds of feature parameters, the principle and the method of extracting and the analysis and implementation of the two matching algorithms are the main research contents of this paper.For characteristic parameters, in this paper researchs mainly on the symbol features, frequency domain features, constellation diagram characteristics,etc, involving signal time domain symbols change time extraction, spectrum and high order spectrum unimodal existence judgment, the power spectrum peak number judgment, spectrum symmetry judgment. In addition to the threshold decision based on feature parameters, the algorithm is also involved in the constellation graph matching algorithm based on clustering algorithm, and the most suitable clustering algorithm is introduced. The HCM algorithm and the improved FCM algorithm are described in detail. Finally, we can get more accurate signal constellation in the condition of high SNR.For the recognition of the high order QAM modulation signal, the clustering algorithm is very easy to be affected by noise, so the algorithm can achieve a good performance only under the condition of high noise ratio. In order to expand the scope of application, this paper proposes a method of SNR estimated value matching algorithm based on feature decomposition and polynomial fitting for low SNR QAM modulation signal, and combines this 2 algorithms to build the structure of the whole algorithm. |
PDF Full Text Request