HF radio direction finding operating on an heterogeneous array of antennas

When a channel generates multipaths, radio direction finding must resort to high resolution algorithms such as MUSIC (multi-signals classification), ESPRIT (estimation of signals parameters via rotational invariance techniques), and maximum of likelihood. An example of application is the estimation of the directions of arrival (DOA) of HF radiowaves (3-30 MHz) after a propagation through the ionosphere: several ionospheric layers may act as reflectors for waves propagating in one or more hops. Classically, these algorithms operate on arrays of isotropic or identical sensors because the information they use is the geometrical phase that the space diversity induces on the received signals. We apply the MUSIC algorithm in an original way in which the polarization (function of the directions of arrival) is used as a factor of the separation for the impinging modes, in the same way as the directions of arrival, Doppler frequency shift or group delay. This technique operates on an heterogeneous array of antennas and assumes that the different complex antenna responses are well known. After a description of our method, we give some experimental results