Research On Low-complexity DOA Estimation Method Based On Spectral Peak Slope And Variable Step Size Search

DOA(direction-of-arrival)estimation has attracted much attention in military and civil fields as one of the main research topics in array signal processing,including navigation,sonar,weather observation and radar target detection and so on.The classic Capon method obtains the spectral peak information of the target through uniform search,which requires a large amount of computation.As for the famous MUSIC high resolution subspace method,although the estimation accuracy is improved,it still estimates the target direction based on the spectrum peak exhaustive search.Meanwhile,the MUSIC method uses the eigenvalue decomposition,which further increases the computation cost.In order to make the search class method be applied in reality,the primary task of improving the real-time and reliability of the hardware system is reducing computational complexity.This paper proposes a DOA estimation method with low complexity based on variable step search of spectral peak slope,on the foundation of the classic Capon method.The classical Capon super-resolution direction of arrival estimation method uses a fixed step size to search the spectrum peak exhaustively,which has the problems of large computation and storage,long running time and unsatisfactory estimation accuracy under the condition of specific computation cost.To address these problems,we propose a low-complexity variable-step search algorithm,which uses the large-step Capon algorithm to search roughly.According to the spectral peak value,we can determine the rough range of the direction of arrival detection,and calculate the slope value of the spectral peak in this range.The spectrum peak is on its left side P due to the negative slope value of first time,therefore,the left value of the range which corresponded to the slope and the point P-2towards the left are taken.Then we start the variable step search from the point P-2,and select the step length according to the previous step length x=1.The step length function is selected for the iterative approximation of variable step length until the slope is less than zero for the first time.We use the left value of the range which slope is less than zero as the estimated angle of the expected signal.The method we proposed avoids the exhaustive search of spectral peaks and reduces the amount of calculation.In this paper,the computational cost of the proposed method is analyzed theoretically,and it is proved that the proposed algorithm significantly reduces the computational cost.We simulated the RMSE of the Angle estimation,calculated the running time and the resolution success rate of our algorithm.The analysis and experimental results show that the proposed method can not only accurately estimate the Angle information of multiple signals,but also has a higher accuracy than the Capon method.And more importantly,the proposed method has no spectrum peak exhaustive search and eigenvalue decomposition,which greatly reduces the computation and complexity.At the end of this paper,the proposed method is derived theoretically in the aspect of Angle estimation of multiple signal sources.And the experiment of Angle estimation performance under different conditions of is conducted on MTALAB.