Study On Automatic Modulation Recognition Algorithms Of Digital Communicate Signals

Digital communication signal’s modulation technology is the key technologies ofcommunication system applied to a wide range. It’s has very important value both in thecooperation and non-cooperation communication occasions. Modulation recognition isthe basis of non-cooperation communication and is widely used in many military andcivil fields. And also widely used in cooperation communication such as adaptiveOFDM systems and wireless automatic routing distribution network and soon.Therefore automatic modulation technology of digital communication signal plays avital role in digital communication, has important value and significance in research.Pre-processing of received signals, extracting of feature parameters andclassification are the three parts of modulation recognition. In search of these parts,main works and innovation of this paper include:Firstly, the background, significance and development status of the automatic digitalmodulation recognition are introduced. And the results of simulations verified thecorrectness of the modulation principle.Secondly, parameter estimation of signals is 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 eachmethod are analyzed.Thirdly，studing on two de-noising metnods---butterworth digital filter de-noisingand wavelet de-noising, then analysising the development and the principle of them.Through simulations to achieve de-noising by use these two de-noising methods, andcompare consumernoise performance.of both.Lastly, the recognition algorithm based on instantaneous information is researched.For this identification method, this paper using the Hilbert transform to extract theinstantaneous parameters of the identification signals, using both Butterworth digitalfilter and wavelet threshold filter to de-noise the noise of the identification signals’instantaneous, using binary tree classifier of decision tree classifier to identify thesignals. In the decision tree classifier has used the improved characteristic parameter.The simulation results show that using the inproved characteristic parameter has greatlyimproved the recognition rate under the low signal-to noise rate.