A Low Complexity Coherent 16×400 Gbit/s 4SC-16QAM DSCM system with Precise Transceiver IQ Skew Compensation and Simplified Equalization

Abstract

1. Introduction Digital subcarrier multiplexing (DSCM) technique has been regarded as an attractive solution in power consumption sensitive datacenter interconnect (DCI), which exhibits the potential in the simplification of the dispersion compensation [1, 2]. However, compared to the conventional single carrier system, the DSCM signal is more sensitive to the transceiver IQ shew. The subcarriers will suffer severe interference from symmetrical frequency subcarriers in the presence of transceiver IQ skew, especially for the subcarriers located at high frequency, causing seriously performance degradation [3]. Multiple-in multiple-out (MIMO) equalizer is an effective method to address IQ skew impairments [4]. Different from the single-carrier system, an 8×8 real-value MIMO equalizer is required for a symmetric subcarrier pair in DSCM systems, resulting in high computational complexity [5, 6]. In this paper, we propose a novel low-complexity transceiver IQ skew estimation method based on specially designed training signal to compensate both the Tx-skew and Rx-skew for DSCM systems to avoid complicated 8×8 real-value MIMO equalizer. Besides, to further reduce the computational complexity of the Rx-DSP, a simplified equalizer structure embedded a 1-tap phase factor is also proposed, which cascades a 1-tap 2×2 MIMO and two Ntap single-in single-out (SISO) equalizers. The proposed method is proved in 50Gbaud 4SC-16QAM DSCM WDM system with 16 channels. The results show that about 6dB OSNR gain can be obtained at the HD-FEC threshold after transceiver IQ skew compensation. And there is almost no performance penalty of the proposed simplified equalizer compared to the conventional CMMA combined blind phase search (BPS) method. According to the results, the proposed low complexity coherent DSCM scheme is a promising solution for DCI applications. Fig. 1. Experimental setup and DSP flow. (a) Training signal structure.