Dual-modality channelized-microwave frequency measurement with machine learning calibration
摘要
Photonic-assisted microwave frequency identification systems have been extensively investigated in radio frequency (RF) applications, offering key benefits such as wide bandwidth and immunity to electromagnetic interference. In practical scenarios, simultaneous measurement of multiple microwave signals with high precision is particularly valuable. However, existing methods often struggle to balance wide-bandwidth multi-frequency detection with high measurement precision. In this work, we demonstrate a novel dual-modality silicon-based microwave frequency measurement (MFM) system using channelization techniques. The system integrates a tunable high-Q Mach–Zehnder interferometer (MZI)-coupled micro-ring resonator (MRR), a flat-passband dual-cascaded MRR, and a wavelength division demultiplexer. By deliberately introducing a frequency spacing mismatch between the dual-cascaded MRR and the input multi-wavelength lasers, the modulated RF signals are effectively divided into multiple narrow optical channels, enabling independent frequency measurement within each channel via the MZI-coupled MRR frequency identifier. The MZI-coupled MRR operates as both a frequency-to-time mapper (FTTM) and a frequency-to-power mapper (FTPM). This dual-modality scheme allows the chip to identify different types of microwave signals, including single-frequency, multi-frequency, and multi-band signals. By incorporating machine-learning-calibrated algorithms, the system achieves high-precision measurements with an error of 14.13 MHz across a 1–40 GHz range. This dual-modality approach combines wide bandwidth and high precision, offering programmable reconfiguration capability and reducing the footprint for advanced microwave signal applications.