DSP and Optimization for System-Level and Component-Level Improvements in Coherent Optical Communications

Abstract

Fiber-optic communications are the backbone of the internet and are responsible for carrying an enormous amount of traffic each day. This thesis presents system-level and component-level innovations to unlock the full potential of fiber-optic communications and address existing challenges, including limited tracking performance, static and dynamic distortions in digital-to-analog converter (DAC), and the issue of joint precoding and constellation shaping.

Equalization is an essential part of optical communications systems to mitigate fiber distortions. State-of-polarization (SOP) is a stochastic property of fiber and fast changes can cause the receiver to lose track. In the first part of this thesis, several equalization methods are proposed to enhance the tracking performance of the coherent optical receivers.

DAC and analog-to-digital converter (ADC) are two prime enablers of high-speed optical transmission. However, their performance is largely limited by the distortions caused by component mismatches. In this thesis, I study three sources of DAC distortions, fall/rise asymmetry response, current amplitude mismatches, and random timing errors among DAC switches.

In the second part of this thesis, a novel optimization framework is introduced for deriving optimized DAC structures that achieve significantly lower distortion and can be easily adopted in the industry. Various digital signal processing (DSP) algorithms are also proposed for further improvement of DAC performance.

The third part of this thesis tackles the issue of combined precoding and constellation shaping at the transmitter. A shaping algorithm is introduced based on a stationary Markov process where the transition probability distribution of symbols is optimized. The proposed scheme can achieve shaping gain in combination with the pre-equalization of channel response at the transmitter.