A normalized spatial spectrum for DOA estimation with uniform linear arrays in the presence of unknown mutual coupling

Eigenstructure based self-calibration methods usually employ MUSIC algorithm or its variations to estimate DOAs. But ambiguous DOA estimates may be obtained since the spatial spectra of these methods might be seriously disturbed by unknown mutual coupling. In this work, we try to mitigate the perturbations caused by unknown mutual coupling in eigen-structure based self-calibration. We analyse the property of mutual coupling matrix and find that part of the mutual coupling inducing false peaks in the spatial spectrum can be predicted. Thereby, a normalized spatial spectrum is proposed to automatically remove these false peaks. The proposed spectrum is applicable to some types of existing algorithms with a low cost of computation. Since most of the false peaks in the original spatial spectrum are removed, a more robust DOA estimate can be expected. The effectiveness of the normalized spatial spectrum is validated via numerical experiments.