Contents I. Introduction
Global IP traffic continues to rise, especially in metro area networks where the expected compound annual growth rate is between 25% and 67% [1], calling for the development of the next generation of optical links. Future datacom and telecom networks should feature increased throughputs that will surpass the current 100 Gb/s standard with the deployment of systems operating at 400 Gb/s and beyond, already underway [2]–[4]. Such growth can only be sustainably accommodated by the implementation of digital coherent optical transmission schemes together with high-order modulation formats, such as quadrature amplitude modulation (m-QAM) which, by encoding information in both the amplitude and phase components of an optical carrier, significantly increase the spectral efficiency, exploiting better the bandwidth of the optical fiber [5], [6]. These modulation formats are usually generated following a transmitter (Tx) scheme as presented in [7]–[9], which relies on high-speed digital-to-analog converters (DACs) [10], [11] and linear driver amplifiers [12]–[15], resulting in very power hungry modules. More customized solutions with lower power dissipation while offering comparable resolutions are desired in order to meet future demands. In this context, digital to analog conversion performed in the optical domain is being intensively explored, as it precludes the need for the external DACs on the Tx, clearing the way to new low-power, low-cost solutions featuring large transmission capacity per channel and increased component density.
