A comparison of quadrature and single-channel receiver processing in adaptive beamforming

A performance comparison between two different types of receivers in an HF adaptive array application is presented. The data used for the study were recorded using a bistatic HF radar located in the central valley of California. Eastword-looking transmissions were generated using a continuous sweep-frequency (chirp) format with a sweep repetition rate of 60 Hz. Backscatter signals, produced by a one-hop forward ionospheric path followed by a one-hop ionospheric return path were received using eight adjacent 32-element linear subarrays spanning a total linear aperture of 2.5 km. Single-channel receivers at each of the eight subarrays were employed for deramping and conversion to baseband. The overall system bandwidth was 960 Hz, and a common local oscillator was used for the receivers. All adaptive processing was carried out off-line using digitized data recorded simultaneously at the eight receiver outputs. The objective of the study described in this paper was to compare the results obtained using the receiving system described above with those which would be obtained using quadrature receivers at the subarray outputs. In the latter case, dual common local oscillator injection, with a
90\deg
phase difference, is employed to obtain two receiver outputs, (in-phase and quadrature components) for each receiver. The component outputs then each have a bandwidth equal to one-half that of the single-channel receiver or 480 Hz for both the in-phase and quadrature signals. One advantage of quadrature reception is that a single real multiplying weight can be used at the two outputs to obtain a frequency-independent gain and phase-shift across the received bandwidth, a property which is not easily obtained for all azimuths in the single-channel receiver structure. The study described in this paper was undertaken to quantitatively determine the importance of this difference in an adaptive beamforming system. The approach taken was to synthesize quadrature receiver signals from the recorded single-channel data using Hilbert transform methods. Comparisons between the two receiving systems were then carried out using identical data sets. The results obtained illustrate that minor differences (less than 5 dB) exist in the processed outputs provided that the same number of adaptive degrees of freedom are used in the two processors. As a result, decisions as to the specific form of a receiver to be used in an HF radar should be based on considerations relating to ease of implementation rather than the overall performance of the adaptive processor.