A Comparison of Direction Finding Results From an FFT Peak Identification Technique With Those From the Music Algorithm

A peak identification technique which uses the FFT algorithm is presented for unambiguously identifying up to three sources in signals received by the sampled aperture receiving array SARA of the Communications Research Center. The technique involves removing phase rotations resulting from the FFT and the data configuration and interpreting this result as the direction cosine distribution of the received signal. The locations and amplitudes of all peaks for one array arm are matched with those in a master list for a single source in order to identify actual sources. The identification of actual sources was found to be subject to the limitations of the FFT in that there was an inherent bias for the secondary and tertiary sources to appear at the side-lobe positions of the strongest source. There appears to be a limit in the ratio of the magnitude of a weaker source to that of the strongest source, below which it becomes too difficult to reliably identify true sources. For the SARA array this ratio is near-10 dB. Some of the data were also analyzed using the more complex MUSIC algorithm which yields a narrower directional peak for the sources than the FFT. For the SARA array, using ungroomed data, the largest side and grating lobes that the MUSIC algorithm produces are some 10 dB below the largest side and grating lobes that are produced using the FFT algorithm. Consequently the source-separation problem is less than that encountered using the FFT algorithm, but is not eliminated.