Comparing DMT Variants in Medium-Reach 100G Optically Amplified Systems

Several research works are currently addressing 100+ Gbps per wavelength high-capacity dense wavelength division multiplexing (DWDM) medium reach systems (80 km on SMF fibers) to cover the metro market segment. For the purposes of cost reduction, it is interesting to consider the use of direct detection receivers rather than the more expensive coherent receivers. However, DWDM transmission in the C-Band imposes severe limitations in terms of tolerances to chromatic dispersion, electrical low-pass filtering, analog-to- digital converter and digital-to-analog converter quantization effects, and receiver optical signal-to-noise ratio (OSNR). In this paper, we analyze the ultimate performance of several variances of discrete multitone (DMT) modulation (namely dual-sideband DMT, single-sideband, and vestigial-sideband), comparing them in terms of required OSNR as a function of several system parameters in a medium reach scenario. We found that for the 80 km target link only single-sideband DMT seems a viable option, while double-sideband DMT has exceedingly high OSNR requirements.

I. Introduction

Optical transmission systems are steadily evolving to 100+ Gbps per wavelength solutions in several market segments, from the inter-data center to the long-haul systems. While in long-haul the clear winner is the combination of advanced modulation formats, digital signal processing (DSP) and coherent detection, the situation is less clear for shorter links, so that several recent research papers have addressed solutions that retain less-expensive direct detection receivers coupling them with advanced modulation formats and DSP [1]–[4]. Our research focuses on 100+ Gbps on a “medium-reach” scenario characterized by distances up to 80 km, using optical pre-amplification in front of direct detection receiver and operating in the C-Band in order to be compliant with dense wavelength division multiplexing (DWDM). This scenario is typically applicable to next-generation ultra-high capacity metro network. If one wants to target 100 Gbps (or more) per over these links, traditional on-off keying is completely out of question, due to electrical bandwidth limitations and the exceedingly strong impact of chromatic dispersion, which cannot be reasonably counteracted by any form of electronic equalization at the receiver. Additionally, PAM-4 solutions currently under standardization for 100 Gbps solutions over short link (i.e., up to 2 km) seems also very sensitive in the previously described medium-reach scenario, particularly due to chromatic dispersion (unless optical chromatic dispersion compensation is used, an option that is intentionally neglected in this paper).