The Research Of Direction-of-Arrival Estimation For FM Signals

Direction-of-Arrival(DOA)estimation technique is a key issue in the field of array signal processing and plays an important role in many fields,such as radar,sonar,medicine,and communications.DOA estimates the research object from narrow-band signals to ultra wide-band signals,researching targets from stationary signals to non-stationary signals,noise models include additive white Gaussian noise and impulse noise,and antenna array structures ranging from uniform arrays to non-uniform arrays.In practice,many common signals do not satisfy the assumption of smooth Gaussian signals in theoretical research,and they have obvious non-stationary characteristic,such as voice signals,heart sound signals,and radar signals.Frequency Modulation(FM)signal is a typical non-stationary signal.The second-order statistical characteristic changes with time and has obvious time-frequency aggregation.It is widely used in academic research.It is important to study the DOA estimation of FM signals.With the rapid growth of wireless transmission and the increasingly complex communication environment,it becomes more and more difficulty to achieve high-resolution DOA estimation.Spatial Time-Frequency Distribution(STFD)theory provides a new research idea for the DOA estimation of FM signals.This theory makes full use of the time-frequency information of FM signals and combines time-frequency domain with spatial domain by using STFD theory,the DOA estimation of FM signals has become the current research point.However,in DOA estimation of the FM signals,the second time-frequency transform of the array receive signals will produce unavoidable cross terms,which leading to reducing the estimation performance.In addition,under low signal-to-noise ratio(SNR)and small number of snapshots,some existing FM signals DOA estimation algorithms are invalid.In view of the above problems,this paper takes FM signals as the research object and conducts in-depth research.The specific contents are as follow:1.Firstly,this paper briefly summarizes the research background of DOA estimation and the current research progress at home and abroad,analyzes the value of DOA estimation research of FM signals,and then makes a detailed analysis of the classical MUSIC algorithm and the principle of spatial smoothing algorithm,and uses MATLAB simulation software to evaluate the estimation performance of the two algorithms;2.Under the assumption of the incident sources are far field narrowband sources,the paper researches the DOA estimation method of the FM signals.The STFD theory is applied to the problems applied in the DOA estimation and the key research is carried out.The solutions to suppress the cross terms and select the time-frequency points are given.This paper proposes a time-frequency clustering algorithm for DOA estimation of FM signals.The algorithm first constructs a whitening matrix to modify the received signals of the arrays,and obtains a modified STFD matrixs.This operation can suppress the cross terms effectively.The K means clustering(KMC)algorithm is used to classify the time-frequency points,so that the DOA estimation can be performed separately for each source,so the angular resolution can be improved,the performance advantage is also obvious in the case of small angle intervals.3.Considering the antenna arrangement as Nested Array(NA)structure,studying the time-frequency sparse reconstruction problem under the NA structure,the paper achieves DOA estimation of under-determined conditions and coherent sources,and proposes a nested array time-frequency sparse reconstruction(NA-TFSR)DOA estimation algorithm.The algorithm firstly builds the array receiving signal model of the nested array which can improves the freedom of the arrays.Combined with the STFD theory and the sparse reconstruction theory,the DOA estimation problem of the FM signals is transformed into a sparse reconstruction problem,and finally the DOA problem can be solved by using the Second Order Cone Program(SOCP).Comparing with three similar algorithms,this algorithm can effectively deal with coherent sources and improve the estimated number of sources,and has obvious advantage under low SNR and small number of snapshots.