IEEE TRANSACTIONS ON NEURAL NETWORKS, VOL. 13, NO. 5, SEPTEMBER 2002 1149 Adaptive Acquisition and Tracking for Deep Space Array Feed Antennas Ryan Mukai, Victor A. Vilnrotter, Senior Member, IEEE, Payman Arabshahi, Member, IEEE, and Vahraz Jamnejad, Senior Member, IEEE Abstract—The use of radial basis function (RBF) networks and least squares algorithms for acquisition and fine tracking of NASA’s 70-m-deep space network antennas is described and evaluated. We demonstrate that such a network, trained using the computationally efficient orthogonal least squares algorithm and working in conjunction with an array feed compensation system, can point a 70-m-deep space antenna with root mean square (rms) errors of 0.1–0.5 millidegrees (mdeg) under a wide range of signal-to-noise ratios and antenna elevations. This pointing accuracy is significantly better than the 0.8 mdeg benchmark for communications at Ka-band frequencies (32 GHz). Continuous adaptation strategies for the RBF network were also implemented to compensate for antenna aging, thermal gradients, and other factors leading to time-varying changes in the antenna structure, resulting in dramatic improvements in system performance. The systems described here are currently in testing phases at NASA’s Goldstone Deep Space Network (DSN) and were evaluated using Ka-band telemetry from the Cassini spacecraft. Index Terms—Adaptive, antennas, array feed, deep space network, NASA, neural networks, orthogonal least-squares, radial basis function (RBF) networks. I. INTRODUCTION T HE NASA Deep Space Network (DSN) is an international network of steerable high-gain reflector antennas, which supports interplanetary spacecraft missions, radio and radar as- tronomy observations for the exploration of the solar system, and select Earth-orbiting missions. The DSN currently consists of three deep-space communication facilities, placed approx- imately 120 apart around the world; at Goldstone, in Cali- fornia’s Mojave Desert; near Madrid, Spain; and near Canberra, Australia. This strategic placement permits constant observation of spacecraft as the Earth rotates, and helps make the DSN the largest and most sensitive radio science and telecommunications system in the world. Over the past years, there has been increasing interest in the use of shorter carrier wavelengths to enhance the DSNs telecommunications and radio science capabilities. Shorter carrier wavelengths, or equivalently higher carrier frequencies, yield greater antenna gains and increased useful bandwidth, with reduced sensitivity to deep-space plasma effects, that tend to degrade the quality of the received signal. However, there are also new problems associated with the use of higher carrier frequencies, namely greater losses due to Manuscript received December 4, 2000; revised July 16, 2001. The authors are with the Jet Propulsion Laboratory, California Institute of Technology, Pasadena, CA 91109 USA (e-mail: payman@jpl.nasa.gov). Publisher Item Identifier S 1045-9227(02)01801-5. gravity-induced antenna deformations and wind, greater sen- sitivity to misalignments of the radio-frequency (RF) compo- nents, and more stringent pointing requirements—all of which are further complicated by time-varying distortions imposed on the antenna structure. Even in the absence of external distur- bances, such as wind, time- and elevation-dependent loss com- ponents are introduced by gravity as the antenna tracks the target (whether it is a spacecraft or a radio-source). The combination of these factors can lead to unacceptably large pointing errors and signal-to-noise-ratio (SNR) losses if left uncorrected. Recovery of SNR losses due to gravitational deformation has been addressed in [1]–[3]. Here, we consider the problem of acquiring and tracking spacecraft with sufficient accuracy to maintain acceptably small pointing losses (nominally 0.1 dB) on large DSN antennas. A. Array Feed Compensation System A recently developed approach for recovering losses due to gravitational deformations, thermal distortion and wind consists of a real-time compensation system employing a seven-element array of feeds in the focal plane of the antenna’s subreflector [1]. The array feed compensation system (AFCS) has been evaluated at the DSN’s Goldstone complex, and has successfully demon- strated real-time gravity-compensation and closed-loop tracking of spacecraft and radio-source signals at Ka-band frequencies (nominally 32 GHz). Its application to recovering losses due to mechanical antenna distortions at high frequencies (32 GHz or higher) is described in [2] and [3]. A conceptual block diagram of the Ka-band AFCS designed for the DSN’s 70-m antennas is shown in Fig. 1. Its main compo- nents are an array of seven 22 dBi horns with a separate Ka-band low-noise amplifier (LNA) connected to each horn; a seven- channel downconverter assembly that converts the 32 GHz RF signal to 300 MHz IF (intermediate frequency), followed by a seven-channel baseband downconverter assembly that generates 14 real (seven complex) baseband signals. A digital signal pro- cessing assembly then extracts parameters from the digital sam- ples in real-time to obtain the optimum combining weights and determine the antenna pointing updates needed to maximize the combined SNR. In the absence of antenna distortions, a single properly de- signed receiving horn collects virtually all of the focused signal power. Distortions generally lead to a shift in the peak of the signal distribution, as well as a redistribution of the signal power in the focal plane. This leads to loss of power in the central channel, which can be recovered by the outer horns of an array placed in the focal plane. When the horn signals are multiplied 1045–9227/02$17.00 © 2002 IEEE