Wide-linewidth coherent systems with parallelized extended Kalman filter for data center inter-connection

Abstract

Adopting wide-linewidth lasers is viewed as a promising way to reduce cost in coherent systems for data center inter-connection. However, in this application, low-complexity and effective carrier phase recovery algorithm is indispensable for compensating phase noise of lasers. In this paper, a data-assisted based parallelized extended Kalman filter (EKF) algorithm is firstly proposed for carrier phase recovery in coherent systems with low-cost and wide-linewidth lasers, as the proposed algorithm has large linewidth tolerance and realizes a parallelized processing architecture. We verify the feasibility of our proposed algorithm in a wide-linewidth coherent system with 448 Gb/s polarization division multiplexing 16-ary quadrature-amplitude-modulation (PDM-16QAM). The results show that the proposed algorithm has an enhanced linewidth tolerance than the blind phase search (BPS) and QPSK partition. For lasers linewidth up to 20 MHz, at 20% forward error correction (FEC) threshold, the OSNR sensitivity of the proposed algorithm is 0.39 dB and 0.68 dB better than BPS and QPSK partition, respectively. At 7% FEC threshold, the OSNR sensitivity of the proposed algorithm is 0.68 dB and 1.38 dB better than BPS and QPSK partition, respectively. Finally, the computational complexity is analyzed, where the proposed algorithm reduces the number of multiplication by 72% compared to BPS.