Performance Model and Design Rules for Optical Systems Employing Low-Resolution DAC/ADC
Sylvain Almonacil, Fabien Boitier, Patricia Layec
Abstract
We present an additive noise model for the signal-to-noise ratio prediction in optical systems employing low-resolution (4 bits and below) digital-to-analog and analog-to-digital converters. Firstly, the expected ASIC power consumption saving by employing low-resolution digital-to-analog and analog-to-digital converters rather than their high-resolution version (8 bits) is assessed and found to be up to 20% for a 4-bit physical resolution. Secondly, to assess the achievable data rate with low-resolution, we present a prediction model, which relies on two steps. First, the mean square error at the low-resolution quantizer output due to quantization and clipping is computed. Second, the mean square error is converted into a signal-to-noise ratio, carefully accounting for the operating parameters (sampling frequency, symbol rate, clipping factor). The domain of validity of this model is assessed for both quantization and clipping noises separately. The predictions are confronted with numerical simulations and a set of experiments in a back-to-back optical set-up. The results show that the signal-to-noise ratio can be predicted with an error of less than 0.5 dB within the domain of validity of the model. Finally, we use this model to establish the resolution requirements for the next generation of metropolitan optical systems (>400 Gbit/s) and compare them with most recent experiments using low-resolution digital-to-analog converters.