Parameter Estimation And Modulation Recognition Of Wideband Signals Based On Compressed Sampling

In recent years,the development of wireless communication technique is very rapid,this makes the limited spectrum resources become increasingly scarce,so the effective management of spectrum resources will have a far-reaching impact on the future development of wireless communication technology.In non-cooperative communication,the receiver lacks the relevant information of the transmitter,but it needs to effectively identify the type of the signal and estimate the parameters.In civil communication,in order to make better use of spectrum resources,it is necessary to detect the spectrum,so as to obtain the available spectrum quickly.This requires the estimation of the modulation parameters of the signal,but the existing techniques are mainly based on the Nyquist rate,which brings a lot of sampling pressure.Therefore,it becomes particularly urgent to realize the sampling rate lower than the Nyquist rate in the front end.In order to overcome the influence of high sampling rate,this dissertation introduced the multi-coset sampling framework,establish the relationship between the compressed samples with high-order statistics,analyzes in the cyclic frequency domain,and reconstructs the higher-order cyclic cumulants amplitude information of the signal at the cyclic frequency by directly using compressed samples,by using the position of the cyclic cumulants appeared to estimate carrier frequencies and symbol rates of signals,by using different characteristics of the cumulant amplitudes of different signals at the cyclic frequency,the modulation identification of the signals is carried out.The main research contents of this dissertation are:1.Three important directions of compressed sensing theory are studied,namely the sparse representation of the signal,the design of the measurement matrix and the signal recovery algorithm.In addition,several compressed sampling frames are analyzed,and the multi-coset sampling frame is selected as the way to obtain compressed samples by taking advantage of its simplicity.2.The characteristics of higher order statistics of signal are studied and the related theories are introduced.By using the high order cumulant which has the properties of additive and anti-gaussian white noise,the signal is modeled as a cyclostationary signal.The relationship between the compressed samples and the high-order cumulant is established,and the amplitude information of the signal cyclic cumulant is reconstructed by analyzing in the cyclic frequency domain.After the sampling frame is determined,the required inverse matrix can be calculated offline.Due to the limited samples obtained,The cyclostationarity of the signal is asymptotic,so the bottom noise will occur.Only the amplitude information at the cycle frequency is utilized,so the bottom noise is limited.The experimental results show that the operation can suppress the noise and has no influence on the useful amplitude information.3.Using the reconstructed high-order cyclic cumulant amplitude information and the particularity of the position where the high-order cyclic cumulant appears,the design algorithm estimates the carrier frequencies and symbol rates of the signals.Using different characteristics of the cumulant amplitudes of different signals at the cyclic frequency,an algorithm is designed to recognize the modulation of different signals.The modulation identification of the mixed signals is realized by using the peak information at the quadruple carrier frequency obtained when estimating the carrier frequency and the rate information.The experimental results show that the proposed algorithm can realize parameter estimation and modulation recognition of signals at low SNR.