Study On The Identification Of Wavelet Packet Modulation Signals

Modulation identification technology is very important in the non-cooperative communication and software radio application areas. In the field of modulation identification, the new modulation patterns also give challenge to people constantly. With the development of wireless communication, multi-carrier modulation(MCM) technology is concerned with people greatly for its good anti-fading performance and easy implementation. Wavelet packet modulation(WPM) is a new type of multi-carrier modualtion technologies which based on wavelet packet transform(WPT), it has the advantages of high spectrum utilizaion, good anti-multipath interference and anti-intersymbol interference(ISI) ability, also easy to realizing multi-rate transmission etc. This thesis studies the inter-class identification of multi-carrier and single-carrier modulation(SCM) signals, the within-class identification between WPM and OFDM signals, by extracting proper parameters, realizing the classification in Gaussian, Rayleigh and frequency selective channels, respectively.1.According to property that multi-carrier modulation signals are asymptotic Gaussian, whereas sigle-carrier modulation signals don't have this property, the Gaussian detection method based on empirical distribution function is introduced to realize the classification between multi-carrier and single-carrier modulation signals in Gaussian channels, the new method based on higher-order moments is introduced to realize the identification in Rayleigh and frequency selective channels. Computer simulation results show that these methods have good performance.2.For wavelet packet modulation singals and OFDM signals, by studying the characteristic in frequency domain with power spectrum estimation theorem, a new identifying method based on the feature of power spectrum envelop is proposed. According to the difference of power spectrum envelop of these two signals, the feature parameter which reflecting the envelop change of power spectrum is extracted to classify wavelet packet modualtion signals. Computer simulation results show that, this method can identify wavelet packet modulation signals in three different channels and the performance is well.3.Using the cyclostationary theory to discuss the identification of wavelet packet modulation signals, this thesis proposes the identification method related with cyclic correlation. Firstly, through the process based on cyclic correlation, the peak in cyclic frequency domain which reflects the symbol rate is detected, then the feature parameter about the peak amplitude is extracted, which can classify wavelet packet modulation signals from OFDM signals. Computer simulation results verify that the classification performance of this parameter in Gaussian, Rayleigh and frequency selective channels is good.