DOA Estimation Using Multiple Kernel Learning SVM Considering Mutual Coupling

Abstract

The Knowledge of Direction of Arrival (DOA) of the signal impinging on a smart antenna enables us to reduce the effect of interference. In this paper, we investigate efficiency of multiple kernel learning SVR-based direction of arrival estimation in a smart antenna with mutual coupling effect. Mutual coupling effect can degrade performance of traditional DOA estimation methods such as MUSIC severely especially when the distance between elements is very small, but SVR-based methods such as MKL support vector regression method can deal with this problem very well. In this work, receiving mutual impedance method is used to calculate mutual coupling matrix. This method can deal with mutual coupling effect better than conventional mutual impedance method when the array is in receiving mode and the distance between elements is very small.