Tracking channel alignment in blind polarization de-multiplexing

• Published in: Optics communications Vol. 394; pp. 114 - 121
• Main Authors: Nabavi, Neda, Samadian, Parya, Hall, Trevor J.
• Format: Journal Article
• Language: English
• Published: Elsevier B.V 01.07.2017
• Subjects: Coherent communications; Fourier optics and signal processing; Polarization; Polarization; Fourier optics and signal processing; Coherent communications
• ISSN: 0030-4018; 1873-0310
• DOI: 10.1016/j.optcom.2017.03.006

Blind de-multiplexing methods such as the constant modulus algorithm (CMA) or independent component analysis (ICA) are an effective means of de-multiplexing dual polarization modulated optical signals mixed by the evolution of the polarization state on propagation through the transmission medium. However, the recovery matrix estimated by these methods has an inherent permutation ambiguity in the assignment of polarization channels to ports. As a consequence, frequent instances of channel alignment permutation can occur resulting in loss of data during system recovery. This paper describes a method of maintaining the initial alignment by setting the diversity combiner weight matrix to a member of the equivalence class of the estimated recovery matrix that is the nearest to the previous weight matrix. The method is verified by computer simulation of de-multiplexing by ICA of dual polarization differential quadrature phase shift keyed coherent optical transmission subject to aggressive changes in polarization state as well as experimental investigation of de-multiplexed QPSK modulation by CMA. Perfect maintenance of channel alignment and accurate recovery of the constellations is achieved. •This manuscript presents a proposed modification of the ICA/CMA algorithm.•Due to the random birefringence of optical fibre, input polarizations will be mixed.•This technique enables accurate tracking of polarization channel alignment.•Experimental results show an improvement in channel alignment using the algorithm.