M.Sc Thesis

M.Sc StudentLevy Etgar
SubjectLayer Peeling Algorithm for Reconstructing the Birefringence
in Optical Emulators
DepartmentDepartment of Electrical and Computer Engineering
Supervisor PROF. Moshe Horowitz


In this work we theoretically demonstrate a new method that enables, for the first time, to extract the spatial distribution of all of the birefringence parameters in an optical fiber system.

Polarization mode dispersion (PMD), caused by differential randomization of the birefringence along a long optical haul, may limit the performance of high rates optical communication systems. As a result, PMD emulators are used to mimic the PMD effect in a circulating loop at the laboratory. The PMD emulators are built from a cascade of polarization maintaining fibers connecting to each other usually by rotatable connectors.

None of the current techniques available today can measure the distribution of the birefringence parameters along the system, i.e. the differential index of refraction and the orientation angle of its principle axis. Extracting the birefringence parameters is important for studying optical systems using emulators. Such a technique is also very important to develop novel fiber sensors which measure bending strength and orientation.

Our method for extracting the birefringence parameters is based on the differential evolution of the state of polarization for different wavelengths. Since the reflections from connections between polarization maintaining fibers in emulators are much stronger then Rayleigh backscattering we chose to develop our method based on discrete back-reflections. However, our method can be directly extended for studying continuous reflections due to Rayleigh backscattering. The extraction of the birefringence is performed by a layer peeling method which we have developed. Usually, layer peeling algorithms tend to diverge due to error accumulated along the system and, thus, it becomes very sensitive to noise. We have developed a layer peeling method that minimizes the accumulated error. Hence, our layer-peeling method is stable even in the presence of a high noise level. We theoretically demonstrated our algorithm in a practical emulator and obtained a very good reconstruction even when a significant noise was added to the input data.

We suggest a new optical system that can measure the parameters required for our algorithm. A theoretical study indicates that the parameters of the system such as the frequency resolution and the bandwidth required for the measurement can be easily obtained.