Research Of Time Delay Estimation Based On High Order Cumulants

Time delay estimation is an active research field in signal processing. It has various and important applications in the fields of radar, sonar, mobile communication and so on. It has got broad attention in recent years. With the advantage of blindness to Gaussian noise, high order cumulants has played a more and more important role in time delay estimation. Research on time delay estimation of the two key problems based on high order cumulants for mathematical tool, time delay estimation and the direction of arrival are put forward based on this idea.Previous work studies the principle of least mean square time delay estimation based on fourth order cumulants, due to the fact that the signal to noise (SNR) is low, the performance of the algorithm decreases, fourth-second order normalized cumulants is used in adaptive time delay estimation method under the conditions of noises are correlated and uncorrelated, simulation results show that the new method has the better performance in low SNR condition. On this basis, change fix step to variable step in the adaptive processing, the improved algorithm not only gets accurately estimation in low SNR, but also has the better convergence and stability.Among smart antenna, in order to solve algorithms of the direction of arrival estimation, as the super-resolution spatial spectrum estimation technique, Multiple Signal Classification (MUSIC) algorithm is one of the most classical algorithms. In the given precondition and meeting adaptive conditions, MUSIC algorithm can accurately estimate signal's the direction of arrival, and has very high resolution ability and estimation precision. However, when the system sample number and the number of antenna array elements are finite or SNR is low, the performance of the algorithm sharply decreases and the algorithm even becomes invalid.Due to the limitation of MUSIC algorithm, MUSIC algorithm based on normalized cumulants improves signal to noise ratio in certain range, calculates the fourth-second order normalized cumulants of received signal covariance matrix. The noise subspace matrix is more accurate after singular value decomposition. Through simulations, the proposed method is more accurate than MUSIC method to estimate signal's the direction of arrival and has the better ability to distinguish between adjacent angle. The improved algorithm obtains better estimation accuracy and resolution.