| Coherent detection combined with digital signal processing (DSP) is a technique to further increase the transmission capacity in a single fiber. This dissertation focuses on DSP in coherent optical communications. First, the coherent optical fiber communication and some related DSP algorithms are introduced. Next, two topics are introduced on which the author worked during his Ph.D. study. One topic is digital Nyquist spectral shaping, and the other one is chromatic dispersion (CD) estimation algorithms in Coherent optical receivers.;The impairments in coherent optical fiber communications include laser frequency offset and phase noise, amplified spontaneous emission (ASE) noise, fiber CD, fiber polarization mode dispersion (PMD), random polarization rotation, fiber nonlinearity, etc. DSP methods to take care of some of the impairments are introduced.;Digital Nyquist spectral shaping is usually used to increase the fiber capacity. This dissertation investigates the required finite-impulse-response (FIR) filter tap number and required frequency-domain-equalization (FDE) filter overlap length. This dissertation shows the feasibility of digital-to-analog converter (DAC) speed at 1.2 times the symbol rate for such systems. This dissertation also shows integrating the matched filter into the CD equalizer can reduce the blind equalization complexity, without introducing extra complexity to the CD equalizer.;CD estimation is important for dynamically routed optical systems. The author proposed two CD estimation algorithms in coherent receivers. The proposed non-data-aided CD estimation is an upgrade of the proposed data-aided CD estimation. The proposed non-data-aided CD estimation also has a big tolerance of the frequency offset, besides it does not need training sequences. |
