Study On Automatic Modulation Recognition Algorithms Of Digital Modulated Signals

In non-cooperative communication systems, it is required to judge the modulation format of the received signals first before the signal information is intercepted. The objective of modulation recognition is to decide the modulation format of the communication signals by a certain algorithm without any priori knowledge about the signal information content. Therefore modulation recognition is the basis of non-cooperative communication, and it is very important to study on this subject for its wide application in many military and civil fields, such as communication counterwork, military threat analyses and frequency management.Modulation recognition usually composed of three parts: pre-processing of received signals, extracting of feature parameters and classification. In research of these parts, main works and innovation of this paper include:Firstly, parameter estimation of signals is mainly studied in the pre-processing part. Several common carrier frequency and symbol rate estimation methods are discussed, advantages and disadvantages of them are compared, applicable conditions of each method are analyzed. Theoretical analysis and computer simulation shows that for the eight digital modulated signals to be recognized in this paper, more precise estimation results will be obtained by using the phase difference method to estimate the carrier frequency, and using the method based on maximum peak of the power spectrum of delay product of the signal to estimate the symbol rate, and the estimation result can be used under low signal to noise ratio(SNR).Secondly, extracting of feature parameters and classification are regarded as a whole part to study on modulation recognition algorithms in this paper. The algorithm based on instantaneous information is mainly improved in two aspects: on the one hand, wavelet filter is adopted instead of traditional linear filter to de-noise the instantaneous information of the digital signals; on the other hand, three novel feature parameters that improved and analogized from the existing algorithms and three existing parameters are used and re-combined. As a result, the recognition probability of the algorithm in low SNR is improved.Finally, the algorithms based on wavelet transform and high-order cumulant are also studied. In the algorithm based on wavelet transform, the perfect examining characteristic of wavelet transform to transient information is utilized, modulus maxima of the wavelet transform of digital modulated signals is calculated, then good classification characteristic is obtained. In the algorithm based on high-order cumulant, MPSK signals are recognized under the environment of gauss white noise first, the multi-path environment of the mountainous area of standard COST 207 and in-door environment of standard M.1225 separately. Preferable classification result is obtained through computer simulation.