Performing Volterra Series Based Indirect Learning Digital Pre-Distorter in Self-Homodyne Detection

Nonlinear data distortion effect caused by nonlinear transfer function of communication component devices is one critical limitation factor of using high order modulation formats in short-reach optical fiber communications. In order to mitigate the nonlinear distortion effect with memory, indirect learning architecture based digital pre-distorter (ILA-DPD) algorithm has been widely applied for computing a Volterra filter based DPD (Volterra-DPD), which is used as a pre-inverse of nonlinear distortion effect. Given the potential prospect of self-homodyne detection (SHD) for shorter reach optical communications, in this paper, we study the effectiveness of ILA-DPD for a SHD system based on high order quadrature amplitude modulation (QAM) format with different input signal powers of modulator by varying output voltage amplitudes of arbitrary waveform generator (AWG). The principle of ILA-DPD algorithm is shown from Volterra series theory. In our work, an arcsin function based DPD (arcsin-DPD) which is proposed for addressing nonlinear transfer function of Mach-Zehnder modulator is considered in ILA-DPD algorithm and performed together with computed Volterra-DPD. The performance of considered DPD scheme is evaluated in a proof-of-concept 25 Gbaud dual polarization 64QAM SHD system with back-to-back (BtB) and 10 km transmissions. Experiment results show that the considered DPD scheme could significantly improve the transmission performance compared with using arcsin-DPD and achieve a better performance than by using a Volterra-DPD alone for relatively higher output voltage amplitude of AWG.