Research On Angular Parameter Estimation Algorithm Of Coherently Distributed Sources

Array signal processing is an important branch in the field of modern signal processing,which exploits the sensor array to receive the signal.Compared with conventional single directional sensors,sensor array have the advantages of more flexible beam control,higher signal gain,stronger interference suppression ability and higher spatial resolution.A significant problem in the field of array signal processing is the estimation of the direction of arrival.Because of the existence of multipath scattering phenomena in the actual signal transmission process,such sources are known as distributed sources.When the distributed sources are formed by dispersion or refraction of smooth objects,the sources often present the characteristics of coherently distributed(CD)source.Among them,the angular parameter estimation of CD sources has become a research hotspot because of its important applications in radar,wireless communication and other systems.The algorithms of angular parameter estimation of CD source are investigated in this paper,which is of theoretical significance and application value.The main work of this paper is as follows:1)The anglular parameter estimation algorithms of CD source for linear array based on rotation invariance are studied in this paper,including estimating signal parameters via rotational invariance techniques(ESPRIT)algorithm and propagator method(PM).These two algorithms utilize the rotation invariance of the signal subspace and avoid two-dimensional(2D)spectral peak search.Furthermore,they can automatically match the parameters.Owing to avoiding two-dimensional search,the computational complexity of the two algorithms is lower than that of the traditional distributed signal parameter estimation(DSPE)nialgorithm and the PM algorithm based on spectral peak search.The simulation results show that these two algorithms can achieve good angular estimation.2)An angular parameter estimation of CD sources based on a fast parallel factor(PARAFAC)algorithm with multi-invariance array is proposed in this paper.The problem of nominal DOA estimation for CD sources is linked to PARAFAC decomposition technique,which has not been studied in the previous literatures.The proposed fast PARAFAC algorithm can achieve a fast convergence by utilizing PM to initialize the directional matrices,which can dispense with the high complexity caused by the random initialization.And the spectral peak search and covariance eigenvalue decomposition are avoided to further reduce the complexity.Simulation results show that the nominal DOA estimation performance of the propsed algorithm is obviously better than that of ESPRIT and PM algorithms,and is almost the same as that of the traditional PARAFAC algorithm.Moreover,the proposed algorithm can be directly applied to the scenario with multiple sources,where sources have different angular distribution shapes.3)Two angular parameter estimation algorithms of CD source based on spectral peak search with linear array are investigated in this paper,including successive DSPE(S-DSPE)and successive PM(S-PM)algorithms.The successive idea is exploited to replace the 2D spectrum peak search to three one-dimensional(1D)spectrum peak searches,so the computational complexity of these two algorithms is reduced.And the automatic angle matching can be realized.The simulation results show that the proposed two algorithms are superior to the ESPRIT and PM algorithms,respectively.The validity and superiority of the proposed two algorithms are verified in the simulation results from the aspects of complexity and performance.4)Three angular parameter estimation algorithms of noncircular CD sources for uniform linear array are proposed,including reduced dimensional noncircular DSPE(RD-NC-DSPE)algorithm,noncircular ESPRIT(NC-ESPRIT)and noncircular PM(NC-PM)algorithms.Owing to avoiding the 2D search,so the complexity of the RD-NC-DSPE algorithm is far lower than that of the conventional DSPE algorithm.And the estimation performance is close to the traditional DSPE algorithm.By utilizing the property of the noncircular signal to double array aperture,NC-ESPRIT and NC-PM algorithms outperform the traditional ESPRIT and PM algorithms in terms of the nominal DOA estimation accurancy,respectively.Meanwhile,NC-ESPRIT and NC-PM algorithms require no peak search,and are suitable for the case of noncircular coherently distributed sources,where different angular distribution shapes coexist.These three algorithms all avoid estimating the noncircular phase.The simulation results analyze the effectiveness and superiority of these three algorithms in terms of complexity and performance.