Digital signal processing for laser phase noise compensation in high-capacity digital coherent transmission systems

Abstract

Capacities in optical transmission networks must evolve to catch up with the recent customers' demand for high bandwidth services such as streaming video, video-on-demand, and cloud-based storage/computing. The emerging demands for bandwidth consuming services promote the development of digital signal processing in coherent optical communication systems. In the receiver side, recent advances in high order modulation detection requires future carrier phase estimation (CPE) algorithms to be more robust, efficient and even adaptive. This thesis focuses on developing three advanced carrier phase recovery techniques that are suitable for various applications such as high efficient transmissions, elastic optical networks and advanced data detection technologies with software-defined forward error correction.In the second chapter, an advanced CPE technique suitable for hardware efficient implementation in dual-polarization (DP)-16QAM system is proposed. The proposed CPE use quadrature phase shift keying (QPSK) partitioning and maximum likelihood detection to offer similar laser linewidth tolerance with much reduced calculation complexity comparing to other CPE techniques reported in the literatures. The impact of average length to the laser linewidth tolerance is numerically studied, and the computational complexity is discussed in detail. We also experimentally verified its tolerance to laser linewidth in a 200 Gb/s DP-16-QAM system. Due to its feed-forward structure and reduced computation complexity, the algorithm is suitable for future real-time recovery applications for 16-QAM signals.For higher order modulation formats and elastic optical networks, a blind and universal digital signal processing technique for the recovery of any square-shaped QAM or time-domain hybrid QAM (TDHQ) signals is proposed. Without the aid of training symbols, the algorithm provides fast and robust signal recovery that greatly simplifies the receiver implementation for future dynamic or elastic optical networks. The platform is consisted of several newly proposed techniques such as a modulation-format-independent CPE (MFI-CPE) and a modulation independent joint timing phase and frequency offset estimation module. The transmission performance is fully studied both numerically and experimentally in 28Gbaud transmission systems with various modulation formats. Comparing with traditional training symbol aided technologies, the proposed MFI-CPE achieves similar performances.Finally, a blind and universal cycle-slip detection and correction (CS-DC) technique is proposed. We analytically derive the probability density function (pdf) of cycle slip (CS) after CS-DC and analyzed the impact of detetion threshold to the correction accuracy. The analytical model agrees well with the Mont-Carlo simulation results and gives more accurate estimation when CS probability is extremely low. Extensive simulation results suggest that the proposed CS-DC technique can successfully reduce the CS probability by over one order of magnitude. In addition, we also proposed to cascaded two CS-DC stages to drive the CS probability further down by two order of magnitudes to 2×10⁻⁷ in 28 GBaud PM-QPSK and PM-16-QAM transmission systems even with an extreme amount of fiber nonlinearities. In the end, simulations results demonstrate that the proposed algorithm is robust against residue frequency offset and inter-channel nonlinear distortions.