Abstract
Digital signal processing (DSP)-enabled dual-parallel Mach-Zehnder (DP-MZM)-based spectral converter can overcome traditional converter insufficient to realize high transparency, dynamic re-configurable, and low-cost spectral conversion. However, in practice, the converter's driving RF signals have variables caused by physical device deviations. To explore the converter robustness, this paper, through numinous numerically simulations in the intensity-modulation and direct-detection (IMDD)-based network nodes, analyses the impact of variables on the converter performance in terms of conversion efficiency, the bit error rate (BER) performance and conversion-associated power penalty in the hybrid gigabit passive optical network (GPON) and the fifth-generation (5G) system. Simulation results demonstrate the converter robustness on driving radio frequency (RF) signals' amplitudes, phases, and frequency acceptable variables range, meanwhile, the frequency variables restrict the user signal bandwidth and can limit the converter practical application. In addition, based on variable analysis, this paper suggests a driving approach to eliminate the frequency variables for practical application and demonstrate the simultaneous impact of amplitudes and phases variables on the converter performance.