Amplitude based spatial signal processing using high-order statistics

This paper introduces and analyzes a near-optimal solution to a spatial signal processing problem. While it has other potential applications, it was initially designed for very low frequency (VLF) signals where only signal amplitude-based information is available from the antenna array. A collection of linear filters is determined which attempts to decompose the antenna outputs into independent signal components, one of which should be the signal of interest. The remaining components could be additional communication signals or directional atmospheric noise due to thunderstorm activity. The linear filters are determined using standard and high-order correlation functions between the outputs of the antennas. There are no restrictions on the interferer modulation type or, when used properly, on the signal to interference ratio.<>