Robust Parameter Estimation of Multichannel Damped Sinusoids in Impulsive Noise

In this article, the robust parameter estimation of damped sinusoids measured by multiple channels in the presence of impulsive noise is addressed. To enhance the resistance to the outliers, the signal subspace is extracted from each channel by minimizing the $\ell _{p}$ -norm of the projection error between its basis and the observation. As a computationally fast scheme, the majorization-minimization (MM) approach is employed to solve it in an iterative way. Based on this, the model order selection rule is first developed for determining the number of multichannel sinusoids from the multichannel signal subspace, and it is named as the robust multichannel model order estimator (RMC-MOE). Next, an estimation technique based on the signal subspace extraction is tailored to calculate the parameter estimates of these sinusoids from the impulsive noise corrupted observation. The resulting method is termed as robust multichannel sinusoidal parameter estimator (RMC-SPE). Finally, the estimated sinusoids are matched to their corresponding channels according to the orthogonality errors (OEs) between these sinusoids and the noise subspace of each channel. The performance advantages over the state-of-the-art methods are verified through the extensive simulation results.