RFSoC-based radar frequency-measurement hardware for aerospace avionic monitoring and signal analysis systems

Abstract

Avionics systems often operate in dense electromagnetic environments, requiring both wideband signal interception and accurate radar frequency measurement. However, achieving high resolution typically demands expensive hardware and introduces long processing delays. To address this challenge, we present a complete radar frequency measurement solution that combines an improved monobit Fast Fourier Transform (FFT), a two-stage Zoom-FFT refinement method, and an RFSoC-based hardware implementation. On the theoretical side, we analyze the characteristics of monobit quantized signals, focusing on the nonlinear distortion introduced by the sign function and rotation factor quantization. To reduce the extraneous nonlinear components commonly generated by traditional square approximation kernels, we adopt a high-precision alternative. At the algorithm level, we simplify the spectral calculation process using a segmented FFT method based on symmetry decomposition. We also introduce a rotation factor windowing method, which reduces spectral leakage without requiring expensive point-by-point multiplication in the sampling domain. For multi-bit sampled data, we employ a bitwise splitting and weighted accumulation method. This approach preserves the low-complexity advantage of monobit processing while retaining substantially more useful information from the ADC output. These techniques form the basis of a two-stage frequency measurement architecture. The first stage handles rapid coarse estimation, while the second stage achieves fine estimation through digital down-conversion, filtering, resampling, zeropadding, and energy centrobaric correction. Tested on a Xilinx Zynq UltraScale+ RFSoC platform, the system achieves a refined frequency error below 1MHz, an instantaneous dynamic range of 37dB, and a measurement cycle of about 500ns at 4Gsps.