Improved deep residual network for chirp signal recovery in microwave photonic receiving systems

We demonstrate an improved deep residual network model that can recover the distorted chirp signals of the microwave photonic receiving systems. The method can achieve high-precision recovery of distorted signals by learning the time-frequency characteristics of broadband chirp signals. The simulation results show that the peak-to-floor ratio (PFR) of signal correlation function is improved by about 20.3dB compared with that without deep learning processing. The residual network model can improve the performance of the microwave photonic receiving systems and is expected to reduce the hardware requirements of the receiving systems.