|M.Sc Student||Atzmon Yuval|
|Subject||Phase Noise Impairments and Quadratic Detection Effects in|
Differential Phase-Shift Keyed Transmission with
|Department||Department of Electrical and Computer Engineering||Supervisor||PROFESSOR EMERITUS Moshe Nazarathy|
While optical Differential Phase Shift Keying (DPSK) is already being deployed in the field, a new generation of advanced self-homodyne differential formats such as multi-symbol DPSK with decision-feedback based on either optical pre-processing or digital post-processing, has been recently conceived, and initially experimentally demonstrated. Remarkably, such “self-coherent” systems emulate optically coherent operation without necessitating the physical presence of a local oscillator in the receiver.
Coherent and differentially coherent phase-based communication systems are in principle prone to phase-noise impairments. For optical differential systems, the treatment of phase noise is even more complex than its wireless communication counterpart, as it includes in addition to the source (laser) phase noise, the non-linear phase-noise induced by the amplified-spontaneous emission via the Kerr effect (the so-called Gordon-Mollenauer effect).
This work addresses key modeling aspects and limiting impairments of self-coherent differential systems (optical DPSK in particular). It revisits the topics of phase noise and quadratic detection in differentially coherent optical communication, by means of a simplified yet accurate Gaussian statistics approach based on polar-domain filtering of the angular fluctuations, and a Volterra-series-based description.
Our novel analytic Gaussian formulation provides high accuracy (±0.1dB), while accounting for the combined effect of all three (linear, nonlinear and laser) phase noise sources impairing DPSK and self-coherent detection. The laser phase noise emerges as a key impairment limiting the amount of attainable “self-coherent gain”.
To further model the interaction between quadratic optical detection and electrical post-detection filtering and optical pre-processing driven by decision feedback, the Volterra-based technique introduced in a recent master thesis in our department is further improved, making it exponentially more efficient, is then validated by means of previous DPSK modeling results, and is finally applied to account for the interaction between quadratic detection and optical and electrical filtering with decision feedback.