Constrained polarization demultiplexing for coherent optical receivers

Blind electronic polarization demultiplexing in coherent optical receivers using the constant modulus algorithm occasionally suffers from singularities. The introduction of constraints in the polarization demultiplexer's transfer function compensates for this deficiency, greatly enhancing convergence speed.

Introduction

Polarization demultiplexing in polarization-diversity coherent optical receivers can be performed electronically, using a two-input/two-output adaptive filter with four complex taps [1]–[7]. The constant modulus algorithm (CMA) [8] is a good candidate for adaptation of the demultiplexer taps due to its low computational complexity and its robustness in the presence of intermediate frequency (IF) offsets and laser phase noise [5]. However, its major shortcoming is its possible erroneous convergence to the same polarization tributary [6]–[7]