Model Order Selection in DoA Scenarios via Cross-Entropy based Machine Learning Techniques
This work addresses model order selection in signal processing for direction of arrival estimation, offering an incremental improvement over existing methods.
The paper tackles the problem of estimating the number of incident wavefronts in direction of arrival scenarios by proposing a multilayer neural network with a cross-entropy objective and an online training procedure for adapting to antenna array imperfections. The method outperforms classical model order selection schemes in simulations, showing higher accuracy for small snapshots and low signal-to-noise ratios, with the online training requiring only a few samples after offline initialization.
In this paper, we present a machine learning approach for estimating the number of incident wavefronts in a direction of arrival scenario. In contrast to previous works, a multilayer neural network with a cross-entropy objective is trained. Furthermore, we investigate an online training procedure that allows an adaption of the neural network to imperfections of an antenna array without explicitly calibrating the array manifold. We show via simulations that the proposed method outperforms classical model order selection schemes based on information criteria in terms of accuracy, especially for a small number of snapshots and at low signal-to-noise-ratios. Also, the online training procedure enables the neural network to adapt with only a few online training samples, if initialized by offline training on artificial data.