Research On Modulation Classiifcation And Parameters Estimation Technologies Of Communication Signals

With the rapid development and widespread use of wireless communication technologies,the electromagnetic environment becomes more complex, resulting the lack of electromagneticspectrum resources. In the optimized management of electromagnetic spectrum resources, theelectromagnetic spectrum monitoring appears increasingly important. As key technologies of theelectromagnetic spectrum monitoring, modulation classification and parameter estimation arefacing more difficulties and challenges. In this paper, aiming at the requirements of theelectromagnetic spectrum monitoring under low SNR and limited data, modulation classificationand parameter estimation of digital communication signals have been researched andimplemented. Combined with structure of the electromagnetic sensors network, distributedmodulation classification and parameter estimation have been researched and implemented. Themain content of this paper is summarized as follows:1．To classify digital modulation signals in the case of low SNR and small sample,automatic modulation classification based on cyclic spectrum is investigated, and a scheme ofdistributed modulation classification is designed. Firstly, aimming to distinguish a signal type inASK, FSK, MSK, PSK and QAM, a set of simple cyclic spectral characteristics is extracted tocomplete identification of seven kinds of signals. Secondly, targeted at disadvantages ofmodulation classification method based on decision tree in the case of low SNR and smallsample, an improved modulation classification method based on feature selection and supportvector machines is provided to improve signal average recognition accuracy via selectingeffective features for classification along with the variation of SNR and sample. Finally, madeuse of the network structure of the electromagnetic spectrum monitoring, a multi-nodecollaboration modulation classification scheme is designed to reduce the average energyconsumption of nodes and improves the recognition accuracy.2．A symbol rate estimation method of MPSK signals based on cyclic spectrum andprincipal component analysis(PCA) is investigated in the case of limited data. Firstly, to solvethe problem that peak features of cyclic spectrum are corrupted by colored background noise inthe case of limited data, a symbol rate estimation algorithm which combines PCA with cyclicspectrum is brought forward, improves the estimation precision and is suitable for MPSK signalswith distinct raised cosine roll-off factors. Secondly, in the network of electromagnetic spectrummonitoring, make PCA transformation of cyclic spectrum sections from multi-node and extractthe first principal component to estimate symbol rate. Simulation results show that distributedestimation method based on PCA performs better than single node when the estimation capabilities of multiple nodes are analogous.3．The overall program of electromagnetic spectrum monitoring system is designed basedon signal processing unit’s hardware platform, related hardware modules are implemented, andthe algorithms of modulation classification and parameter estimation are realized on the DSPplatform of signal processing unit. Firstly, the implementation program and hardware structure ofplatform are provided. According to the requirements of electromagnetic spectrum monitoring,signal processing solution and related modules are designed and implemented. Secondly, theDSP processing flows of modulation classification and parameter estimation modules aredesigned and implemented. A distributed modulation classification and parameter estimationprogram is designed, and a software control platform for signal processing unit is completed.Finally, the system test scheme was designed and the algorithms were tested. The test results areconsistent with the conclusion of research.