Research On DOA Estimation Algorithm Based On Sparse Representation

Recently, the direction of arrival (DOA) estimation algorithms based on sparse representation and compressive sensing theory have gained rapid development and their advantages make them become hot topics in the domain of DOA estimation. Compared to traditional DOA estimation algorithms, the DOA estimation by using sparse representation and compressive sensing theory can achieve better performance with low signal-to-noise ratio(SNR) and few snapshots. And these algorithms can handle coherent signals directly without decorrelation. This paper mainly studies the sparse signal DOA estimation with multiple snapshots and off-grid DOA estimation.For the high computational complexity of MMV model, this paper transforms the MMV model to a SMV one by using the linear combination of eigenvectors and proposes SRBWEV algorithm. On this basis, Rife algorithm is modified for off-grid DOA estimation and MRife algorithm is proposed for sparse signal DOA estimation. Moreover, this paper deduces and analyzes the realization condition of MRife algorithm. Finally, the simulation experiments are made to verify that MRife algorithm can solve the off-grid problem effectively and achieve accurate off-grid DOA estimation.For the existing shortcomings of L1-SRACV algorithm,this paper modifies L1-SRACV algorithm and proposes L1-RVSKR algorithm. This algorithm first transforms complex operation to real number operation via a unitary transformation. Then Khatri-Rao product is utilized for vectorization operation which transforms the MMV model to a SMV one.Meanwhile, the generated virtual array is considered as a new overcomplete dictionary. The calculated amount is decreased and the estimation performance is improved.As the estimation performance degradation resulted by the strong correlation of the neighbouring atoms in the overcomplete dictionary, IHT algorithm can not distinguish each signal exactly. For this problem, this paper improves IHT algorithm. A threshold detection is added by using the correlation of the atoms in the overcomplete dictionary and the modified IHT algorithm can distinguish each signal effectively.In this paper, the sparse signal DOA estimation models based on array covariance matrix and signal subspace are improved and the sparse signal DOA estimation model based on eigenvalues is achieved. The modified model reduces the dimension of the overcomplete dictionary so that the amount of computation is reduced and the computational efficiency is improved.Finally, simulated verification and contrast analysis are made by using the received data of the actual direction finding system. Firstly, the interchannel amplitude phase inconsistency is corrected by using calibration algorithm. Then the radiation signal's DOA is estimated by SRBWEV algorithm and MUSIC algorithm. And the performance of these two algorithms in the actual direction finding are compared and analyzed.