Use of the method of parametric estimation in the problem of determining antenna beam patterns from weak radio sources

Abstract

UDC 621.317.743.7 A means of treating measurement data consisting in the direct estimation of the numerical parameters characterizing the main lobe of the pattern by the method of least squares is analyzed in application to the radio-astronomical method of antenna measurements. The errors in determining the gain, the direction of maximum reception, etc. are found from the information matrix. The results of a test of the method on a 7-m reflecting antenna at a wavelength of 6 cm are presented. One of the limitations of the radio-astronomical method of determining antenna beam patterns (BP) [i] is the smallness of the increment AT a of antenna temperature from the source in comparison with the fluctuation threshold Tfl of the measurement receiver. In determining the parameters characterizing the main lobe of a BP it is possible, however, to lower the demands on the ratio ATa/Tfl through optimum methods of treatment of the measurement results. For this purpose one usually uses linear filtration (convolution of the measured BP cross section with a patternlike function: see [2], for example), which increases the signal/noise ratioat each point of the "smoothed" BP cross section, keeping its shape unchanged under certain conditions. At the same time, in practice one is often required not to reconstruct the shape of the main lobe of the BP, which is usually known, but to determine numerical parameters such as the gain, proportional to ATa, the direction 9 m of maximum reception, and the width Ag0. s of the main lobe at the 0.5 level, and with an error not exceeding an assigned value. These parameters are usually found graphically from the "smoothed" BP, but in this case one cannot always give precisely the error of their determination; in addition, such a procedure is not optimal. In this paper we propose to find the required values of the parameters directly from the measurement data using suitable optimal procedures of parameteric estimation; an advantage of such an approach is, in particular, the possibility of finding the errors of determination of the unknown parameters in explicit form (see below).