Frequency domain equalizer in few-mode fiber space-division-multiplexing systems

Few-mode fiber (FMF) communication system has been emerging as a promising space-division-multiplexing (SDM) technology to overcome the next-generation capacity crunch. The key challenges of FMF system are inter-modal crosstalk due to random mode coupling and large differential mode group delay (DMGD). Adaptive frequency domain least mean square (FD-LMS) algorithm has been proposed and demonstrated as the most hardware efficient method in compensating large DMGD and random mode coupling. In this paper, we propose a noise power directed adaptive FD-LMS algorithm, which adopts variable step size to render the posterior error of each frequency bin converge to the background noise level in the additive white Gaussian noise (AWGN) channel. Our simulation result shows that, in a 3000 km two-mode transmission system with 35 ps/km DMGD, noise power directed algorithm can improve the convergence speed by 34% and 54% compared to signal power spectrum density (PSD) dependent adaptive FD-LMS method and conventional fixed step-size adaptive FD-LMS method, with the hardware complexity (number of complex multiplication) increased by only 5.7% and 8.1% respectively. We also propose to use a single-stage adaptive equalizer for compensating both chromatic dispersion (CD) and DMGD simultaneously for further decreasing the overall hardware complexity of digital signal processor (DSP) in coherent receivers. We show that such single-stage equalizer may have a slower convergence speed due to a larger mean square error (MSE) induced by uncompensated CD in equalizer's initial condition. We extend the proposed noise power directed algorithm to increase the convergence speed of the single-stage equalizer; and the simulation results show that the noise power directed algorithm can achieve 51% faster convergence speed than conventional algorithm in a 3000 km transmission system with DMGD of 35 ps/km and CD of 20 ps/nm/km.