DIRECTION OF ARRIVAL ESTIMATION USING SINGLE TRIPOLE RADIO ANTENNA L. K. S. Daldorff Division of Astronomy and Space Physics, Department of Physics and Astronomy, Uppsala University, Sweden. Email: {lars.daldorff}@rymdfysik.uu.se D. S. Turaga, O. Verscheure and A. Biem IBM T. J. Watson Research Center, Hawthorne, NY, USA. Email: {turaga, ov1, biem}@us.ibm.com ABSTRACT We consider the problem of estimating the Direction of Ar- rival (DOA) of multiple waves incident on a single tripole sensor. Using the physical properties of the electric and mag- netic fields, we show that we can disambiguate the DOA of multiple simultaneously incident waves using a set of time sampled 3D measurements from the single sensor. This is different from the traditional approach that uses arrays of an- tennas to estimate DOA. We use the Unitary Matrix Pencil method to estimate the frequencies of the waves, and use a least squares solver to estimate the amplitude and phase coef- ficients. We combine these to compute the DOA and evaluate the approach using simulations. We show that the method is very effective at estimating DOA for different numbers of in- cident waves, and different noise levels. Index Terms— Direction of Arrival, Radio Astronomy, Matrix Pencil Method 1. INTRODUCTION Radio astronomy has led to several key discoveries over the years, as waves penetrate much of the gas and dust in space as well as the clouds of planetary atmospheres and pass through the terrestrial atmosphere with little distortion. There is an overwhelming interest in the scientific community [1] to sig- nificantly expand (by several orders of magnitude) the explo- ration of the radio signal spectrum to image and understand the transient sky, to probe accretion onto black holes; to iden- tify orphan gamma-ray burst afterglows; and finally to dis- cover new and unknown transient phenomena from currently undiscovered celestial objects. A key problem in this field involves identifying the di- rection of arrival of an electromagnetic signal incident on the radio antenna sensor - especially in the presence of multiple such signals, interference and noise. When multiple signals are incident on the antenna sensor, the result is a superposed combination, making it hard to identify the individual signals of interest. The problem of DOA estimation is related to Blind Source Separation, that has been studied in the astronomy and radar communities. Current approaches for DOA estimation use a spatially distributed array of multiple sensors (anten- nas) to disambiguate the multiple signals of interest 1 . With the use a multiple sensors, various techniques are available for DOA estimation. Prior work includes work by [2] [3] and can be categorized into: Phase-based inteferometry methods, Eigen decomposition methods, and machine learning tech- niques. There is also a large body of related work on vector sensor [4], and 2D frequency estimation [5]. Phase interferometry (PI) based methods [6] use a mea- sured phase differences across an array of sensors to estimate the DOA. These approaches have been successful for simpler radar signals, and have had limited appeal in radio astron- omy applications. Decomposition techniques [7] exploit cor- relations inherent in time-dependent signals to estimate the components and directions of the incoming signals. These approaches include the Multiple Signal Classification (MU- SIC) algorithm [8], Maximum Likelihood Methods, and the ESPRIT algorithm for narrow-band planar signals [9]. These approaches offer asymptotically unbiased estimates of the di- rection of the irradiating sources, but are computationally ex- pensive and not easily implemented in a real-time environ- ment. The success of machine learning techniques is contin- gent on the availability of a sufficiently large training data set, especially for large-scale radio astronomy observations. With the new interest in low frequency radio wave in as- tronomy and physic, as exemplified by the Low Frequency Array (LOFAR) with its centre in the Netherlands and the up- coming Square Kilometre Array (SKA), we can not expect a linear wave front through the array for all objects of interest. Unlike these prior approaches that rely on an antenna array, we focus on DOA estimation using a single tripole antenna. Our work is motivated by recent advances in antenna design that have led to the development of practical tripole anten- nas [10] that receive all 3 components (dimensions) of the electromagnetic signals in terms of the resulting electric field or generated current. It has been shown [11] that for a sin- gle incident wave, the DOA may be directly computed from a measurement of these three components. However, when 1 The number of sensors determines the accuracy of the estimate as well as the number of individual waves that may be disambiguated. 2149 978-1-4244-2354-5/09/$25.00 ©2009 IEEE ICASSP 2009