Real-time implementation of non-integer oversampling timing recovery algorithm incorporating IQ compensation for coherent optical communication systems

In coherent optical communication systems, digital signal processing (DSP) plays a critical role in achieving efficient signal recovery and demodulation. The traditional analog-to-digital converter (ADC) must operate at integer oversampling (>= 2 samples per symbol, SPS) to support accurate timing recovery algorithms. However, the computational complexity involved in processing individual symbols presents significant challenges in meeting the low-power consumption demands of high-speed optical modules. In this paper, we propose a timing recovery algorithm based on the Modified-Godard algorithm for coherent optical communication systems that can work under non-integer baudrate oversampling, which greatly reduces the complexity of DSP and effectively alleviates the demand for ADC with high oversampling rates. Our algorithm enables simultaneous timing recovery and IQ skew compensation in the frequency domain (FD). In order to ensure the reliability of our DSP algorithms under non-integer baudrate oversampling, we analyze and simulate the impact of the laser frequency offset and IQ skew on the timing recovery algorithm. Finally, the DSP algorithms are further implemented on a field-programmable gate array (FPGA) chip. The experimental results demonstrate that the real-time coherent receiver achieves a minimum received optical power of -35-dBm at the KP4-FEC BER threshold, while reducing the ADC sampling rate by 42.9 % compared to conventional timing recovery schemes.