An electromagnetic vector sensor consists of a triad of electric dipoles in orthogonal orientation, plus another triad of similarly arranged magnetic loops, all in spatial collocation. This electromagnetic vector sensor has been used in a series of algorithms to estimate the incident sources' directions-of-arrival and polarizations. However, these algorithms have presumed the dipole triads and the loop triads of perfect ideality in their gain/phase responses, their orientations, and locations. Such idealization is rarely (if ever) attained in actual field deployment. Instead, the nonidealities need to be calibrated, often blindly with no training signal impinging from any prior known direction-of-arrival at any prior known polarization. For such a scenario, this work proposes a new algorithm for direction finding, for polarization estimation, and for “blind” calibration (a.k.a. “self-calibration,” “autocalibration,” or “unaided calibration”) of all above nonidealities. This new algorithm is orders-of-magnitude computationally simpler than maximum likelihood estimation. This reduction in complexity is achieved here by exploiting the electromagnetic vector sensor's quintessential array manifold and by a judicious breakdown of the originally high-dimensional problem (of estimating the directions-of-arrival, polarizations, and the antenna nonidealities) into suitably chosen/sequenced low-dimensional subproblems. This proposed algorithm is first in the open literature to exploit the electromagnetic vector-sensor's quintessential array manifold for “blind” calibration of all above mentioned nonidealities simultaneously. Monte Carlo simulations verify this proposed algorithm's effectiveness in “blind” calibration and this algorithm's orders-of-magnitude computational efficiency over the maximum likelihood approach.
- array signal processing
- direction-of-arrival estimation
ASJC Scopus subject areas
- Condensed Matter Physics
- Earth and Planetary Sciences(all)
- Electrical and Electronic Engineering