Automatic Modulation Classification And Parameter Estimation Of OFDM And SC-FDE In Multipath Fading Channel

In the wireless communication systems, people has paid more and more attention to theability of information transmission speed and adapting the vile condition of channels. OrthogonalFrequency Division Multiplexing (OFDM) and Single Carrier with Frequency DomainEqualization (SC-FDE) has become the focus and been widely applied to the communicationsystems because of their advantage of the capability to mitigate the side-effect of multipathpropagation and easily implementing. The aim of modulation recognition unknowing themodulation information and modulation parameters are to judge the modulation type of thesignals and estimate the modulation parameters offering the foundation to the signal processing(such as the demodulation) afterward. Therefore, it has been considered as one of the focal pointof study. Delves into the particularity of OFDM and SC-FDE system structure, this dissertationmainly focuses on the key techniques in automatic modulation classification and parameterestimation for OFDM and SC-FDE signals. The main work and innovative achievements can besummarized as follows:1、 Based on analysis of signal characteristic and modulation principle, the paperestablishes the mathematical models for OFDM and SC-FDE signals. This is the foundation forthe research afterward.2、The paper discusses the automatic modulation classification problem of the cyclic prefixOFDM (CP-OFDM) and linear modulation single-carrier (SCLD) signals. Aiming at the problemthat the algorithm which is based on the Guassian detection will fail in the multipath fadingchannel, this paper presents an algorithm which based on energy normalized autocorrelation torecognize the modulation of CP-OFDM and SCLD. The algorithm normalizes theautocorrelation with energy to remove the scale problem and the impact of frequency offset.Meanwhile, morphologic filtering is used to suppress colored-background noise effectively.Finally, the algorithm designs the classifier based on support vector machine (SVM) andachieves good recognition performance.3、The modulation recognition problem of zero-padding SCLD (ZP-SCLD), cyclic prefixSCLD (CP-SCLD), zero-padding OFDM (ZP-OFDM), CP-OFDM and SCLD signals has beendiscussed. Based on the model and the structure of the signal, the expressions of cyclicautocorrelation function (CAF) for ZP-SCLD and ZP-OFDM signals have been deduced. Andthe location information of the peaks on the CAF profile is exploited to classify the five kinds ofsignal with the cyclic statistical detection method. The simulations in the Gaussian channel andthe multipath channel sustain the theoretical analysis.4、The problem of the parameters estimation for CP-OFDM and CP-SCLD signal isdiscussed. Firstly, based on the analysis of the autocorrelation for the CP-OFDM, the usefulsymbol duration and the cyclic prefix are estimated according to the relativity of the cyclic prefix.Secondly, based on the analysis of CAF profiles for the CP-SCLD, the useful symbol durationand the cyclic prefix are estimated according to the location information of the discrete spectrum lines feature that on the CAF profile at cycle equals to zero, the CAF magnitude is not zero onlyat delay We have to say that some peak of CAF may be submerged by the colored-backgroundnoise. So this paper extends the searching region from the positive frequency domain to thewhole frequency domain and achieves more accuracy result. These two algorithms also applie tomulti-path fading channel. The Computational complexity of the former is lower, but the abilityof suppressing noise of the later is better.