Research On Recognition Theory And Algorithm For Signal Reconnaissance

With the continuous progress of modern communication and radar technology,signal modulation recognition and sorting has become more significant in the field of non-cooperation communication, such as radio monitoring and battlefield communication. How to detect the interest signal effectively becomes an important topic that the scholars both at home and abroad are interested in.In this thesis, the main research and analysis are focused on the signal modulation recognition and sorting algorithms in the non-cooperative communication systems.These contents are separated to several aspects, such as parameters estimation issues during the signal pre-processing, recognition algorithm research of typical communication signal based on clustering algorithms and fractal theory, the signal sorting algorithms of radar pulse signal, and the influence analysis on the received signal brought by phase noise.Firstly, this thesis discusses the problems of parameter estimation during signal pre-processing. Several key feature parameters estimation algorithms are studied and analyzed detailly. Aiming to two important parameters, the symbol rate of digital signal and the signal’s signal-to-noise ratio (SNR), two improved methods are proposed respectively for higher estimation accuracy. During the blind estimation of symbol rate,a low wavelet scale always introduces strong noises in the signal’s low frequency part.For avoiding this problem, a joint estimation method based on wavelet transform and two different types of median filters is proposed. In addition, an improved method is proposed for estimating the SNR under real communication systems. Simulation results show the effectiveness of these two proposed methods in this thesis.Clustering algorithms are usually adopted in image processing. This thesis introduces two kinds of clustering algorithms, subtractive clustering and fuzzy C mean clustering algorithm in signal modulation recognition and classification. Based on the phase periodicity of PSK signals, phase clustering is introduced for modulation order classification, and its improved version is proposed for enhancing the clustering speed.On the other hand, a few statistical parameters are introduced for the signals’recognition with the same modulation order. What’s more, an improved autonomous supervised method based on fuzzy C mean clustering algorithm is proposed for reconstructing the constellation of QAM signals. The clustering process is more accurate and efficient under autonomous directing, and the QAM signals can be identified effectively.Fractal theory is also a hot spot in recent years. In this thesis, the typical digital communication signal modulation recognition algorithm based on the perspective of fractal theory is studied. For some common digital communication signals, an modulation recognition method based on fractal box dimension and information dimension is proposed and verified by the simulation results.Radar systems also have huge demands and prospects of signal’s recognition and sorting in the field of non-cooperation communications, such as monitoring military activity. In this thesis, radar pulse signal recognition and sorting technologies are studied and analyzed from the two aspects, including the intra-pulse modulation characteristics and the inter-pulse characteristic parameters of radar signals. Several algorithms of time-frequency analysis are adopted for radar signal’s recognition based on the intra-pulse modulation feature. On the other hand, the radar pulse signals’ sorting is researched and achieved by the inter-pulse characteristic parameters. In which, an improved Sequence Difference Histogram (SDIF) sorting algorithm based on the pulse repetition interval is proposed. Via the proposed selection method of potential PRI and the new retrieval method, the novel method can effectively improve the performance of radar pulse signal sorting.Numbers of non-ideal factors will be inevitably introduced in the signal’s transmission and reception processes in the non-cooperative communication systems. In this thesis, phase noise is chosen for studying which may possibly influence on the received signal. Besides, a phase noise correction scheme is proposed for the multiple channel communication systems and it has been patented. This calibration system,which adopts intelligent convergence and reback modules, can improve the effectiveness of phase noise correction, and it has been verified by the following simulation results.