Aspects of Power Consumption in Coherent Fiber-Optical Communication Systems

The power consumption of coherent fiber-optical communication systems is becoming increasingly important, for both environmental and economical reasons. The data traffic on the Internet is increasing at a faster pace than that at which optical network equipment is becoming more energy efficient, which means that the overall power consumption of the Internet is increasing. In addition, wasted energy leads to higher costs for network operators, through increased electricity expenses but also because the heat generated in the equipment limits how closely it can be packed. This thesis includes power consumption modelling, trade-off studies and investigations of novel schemes that may lead to an improved energy efficiency in future systems. In particular, the power consumption of optical amplifiers is modelled and connected to a performance model based on the Gaussian-noise model. Using these models, the trade-offs between amplifier power consumption and the choice of modulation format and forward-error-correction (FEC) scheme is studied. We find that 16-ary quadrature-amplitude modulation (16QAM) has a lower energy consumption per bit than quadrature phase-shift keying (QPSK) due to its higher spectral efficiency, and that using a shorter amplifier spacing to improve signal quality may be more energy efficient than using a powerful FEC. Furthermore, the power consumption for a coherent link with minimal digital signal processing (DSP) is studied. We find that when the DSP is minimized, it accounts for below 10% of the power consumption, which is dominated by the optical components. Finally, we demonstrate that the phase-coherence of optical frequency comb lines can be utilized to simplify the carrier-recovery by joint processing of multiple wavelength channels. We use the phase-drift from one wavelength channel to compensate the phase-drift of another one spaced up to±275GHz away with negligible penalty.