Research On Detection And Modulation Analysis Techniques For Co-Channel Multi-Signal Of Communication

As the development of modern wireless communication technique, there’s a problem ofconcentrated of communication signals in the limited bandwidth, the condition of time-frequencyoverlapped is occurs generally and seriously, and the traditional way to single signal can’t satisfydetection and modulation analysis. So, this paper is based on the research of single channelmulti-signal, accomplished blind detection and modulation analysis. The main work andinnovative algorithms proposed are summarized as follows:1. For the characteristics of single-channel multi-signal, adopted a mathematical model ofhybrid signal, then the selected multi-signal’s characteristics such as wavelet, cyclostationary,high-order statistics etc is in analysis and demonstration.2. For the problem of blind detection and estimation of number source of multi-signal. Firstmade use of blind detection algorithm which based on eiqenvalue decomposition, researched onthe relation between detection performance and estimated dimension,simplified the algorithmprocess,this method reduced the complexity and without losing effection. And using wavelettransform, wavelet of noise reduction, SSWIA(Square Sine Wave Interpolating Algorithm) suchalgorithms to estimate the source number. This can detect the signal singular point moreeffectively and conducive to the extraction of spectral lines, promoting the property of algorithm.3. For the problem of single-channel multi-signal parameter estimation. First analysed thecyclostationarity of multi-signal, discussing the relationship between the signal carrier frequencyand symbol rate of different orders of cumulants and different profile of cyclic spectrum. Andfurther analysed the parameter estimation based on cyclic spectrum, started from the relevancebetween the spectrums so that solving the problem of pairing parameters. And usingmorphological filtering to optimize the extraction lines, and simulation proved the cyclicspectrum estimation algorithm.Finally proved that making use of mixed methods in a lower SNRcan estimate the parameters of MPSK, MQAM any mixed case accurately, the mixed methodscould pair parameters and expand the type of signals.4. For the problem of single-channel multi-signal modulation recognition. After carriersynchronization and timing synchronization, through analysis of multi-signal’s high-orderstatistical properties is equivalent to single signal, then extracting the characteristic value basedon high-order cumulants. Then derived Kernel Logistic Regression classifier process, analysedand performanced comparision of different types of classifiers, then using mixed-kernel SupportVector Machine to do classification, finally achieved modulation recognition of MPSK, MQAMmixed.