A Superresolution Technique for Antenna Pattern Measurements
Yasutaka Ogawa, Teruaki Nakajima, Hiroyoshi Yamada, Kiyohiko Itoh · Institutional Repositories DataBase (IRDB) · 1993
In measurements of gain and radiation pattern for a large aperture antenna, a measurement system must be constructed outdoors to achieve a far-field range. In this ease, reflected signals from the ground and other objects often impinge on the antenna, and the measured data. are disturbed by them. A recently developed vector network analyzer provides time-domain processing based on the fast Fourier transform (FFT). We can mathematically remove the unwanted responses that appear as ripples in the frequency domain by gating them in the time-domain presentation [I ). However , desired (direct path) and unwanted (e.g. ground path) responses must be clearly separated in the time-domain presenta.tion. Thus, the difficulty often arises in measuring na.rrow passbandwidth antennas. Furthermore, outdoor wideband antenna measurements are not preferable because t he wideband radiation may interfere with another radio system, and because the measurement system may be interfered with by another radio system. The authors have proposed antenna gain measurements [2] [3] using a MUSIC algorithm(4], and have shown t hat we can measure the antenna gain using much narrower frequency bandwidth data in comparison with the conventional FFT based time-domain processing. However, the MUS1C algorithm usually needs many sets of frequency-domain data (snapshots) to estimate a correlation matrix.. We used 50 snapshots to obtain the antenna gain[3]. However, it is not preferable to need many snapshots for the antenna pattern measurements because it lakes a very long measurement time. It should be noted that we must obtain the response of antenna at each angle of rotation for the antenna pattern measurements. In this paper, we propose a new superresolution me t hod which employs the MUSIC algorithm accompanied with the FFT and gating techniques. The new method needs only a few snapshots. In this paper, we show examples of the method using a single snapshot.