Spectrum Detection And Parameter Estimation Methods Based On Nonuniform Sampling

Information interception system can get the functional parameter, model and other key information of the emitters by sampling and processing of non-cooperative emitter signals. About the spaceborne Information interception system, since the capacity of data sampling, storage and transmission on the satellite is very limited, non-uniform sampling techniques which can realize the wideband detection in the lower sampling rate have a very broad application prospect. Currently, techniques which complete the signal processing directly using non-uniformly sampled data without reconstructing signal are imperfect. This paper focuses on the key technologies on spectrum detection and parameter estimation of sparse multiband signal using non-uniformly sampled data and other expand research. The content of the dissertation is as follows.In chapter 2, the spectrum detection problem of sparse multiband signal using non-uniformly sampled data is investigated. On the diffusion of eigenvalues of autocorrelation matrix by shorter samples, a spectrum detection algorithm based on MUltiple SIgnal Classification(MUSIC-like) and improved Akaike Information Criterion(AIC) is proposed. The proposed algorithm can improve the estimating accuracy of signal subspace dimension with shorter samples, which has better spectrum detection performance.In chapter 3, the problem of frequency estimation of multiple sinusoidal signals using non-uniformly sampled data is investigated. Multiple sinusoidal signals can be considered as one form of sparse multiband signal. On the dictionary mismatch problem on estimating the signal frequency of non-uniformly sampled data, an algorithm based on sparse Bayesian learning(SBL) and dictionary optimization is proposed. The proposed algorithm not only retains the advantages of SBL algorithm of low signal-to-noise ratio(SNR), much limited samples, but also dynamically adjusts discrete Fourier transform(DFT) dictionary for sparse representation. So the error due to dictionary mismatch is effectively reduced and the frequency estimation accuracy of the algorithm is improved. In addition, theoretical analysis and simulation results reveal the effect that non-uniform sampling technique have on the frequency estimation precision and the adjacent frequency resolution of the proposed method.In chapter 4, the problem of parameter estimation of multicomponent chirp signals using non-uniformly sampled data is investigated. Multicomponent chirp signals can also be considered as one form of sparse multiband signal. Chirp signal needs to be described with parameters such as carrier frequency and frequency modulation slope. In order to resolve signal parameter estimation problem better under the condition of low SNR and multi-component chirp, the parameter estimation algorithm based on modified discrete Chirp-Fourier transform(MDCFT) and adaptive importance sampling(AIS) is proposed. The proposed method cannot only use the selected importance function(IF) to generate the samples fitting target function in the parameter region of interest, but can also utilize samples and the corresponding importance weights to update the IF adaptively. Compared with the two existing methods based on Monte Carlo methodology, the proposed method exhibits improved performance.In chapter 5, the problem of parameter estimation using non-uniformly samples in complex noise environment is investigated in two cases. In the first case, the non-uniform sampling technique is composed of multi-channel structure. For the situation that noise power is non-uniform among multiple channels, the frequency estimation algorithm based on SBL and inverse iteration methods is proposed. The proposed algorithm can be applied to the case with multiple noise parameters, which is a further expansion to the frequency estimation algorithm based on the SBL. In the second case, on the situation in colored noise model with an autoregressive process, joint estimation of multi-component chirp signal modulation parameters and model order using the parameter estimation of multicomponent chirp signals based on MDCFT and AIS is carried out. This algorithm solves the parameter estimation problem of multi-component chirp signals in complex noise environment.