Coding, Shaping, and Modulation for High-Throughput Fiber-Optic Communication

点击翻译按钮获取中文摘要

This project applies the tools of coding theory and information theory to improve the information-carrying capacity of fiber-optic communication networks operating at two different distance scales: in long-haul inter-urban and international optical transport networks, and in dense short-haul networks as found in data centres. The challenge is to develop coded data transmission schemes that (a) have high efficiency, approaching fundamental information-theoretic limits; and (b) can be encoded and decoded with practical algorithms operating with reasonable energy efficiency. Fiber-optic communication networks form the backbone of Canada's digital economy. The economic activity conducted through digital technologies such as the Internet, represents a large-and rapidly growing-part of Canada's total economic activity. According to a 2019 paper of Amanda Sinclair, senior analyst of the National Economic Accounts Division of Statistics Canada, technology and the Internet have "fundamentally changed" how Canadians and Canadian businesses interact. Sinclair reported that the gross domestic product connected to the digital economy was nearly $110 billion, representing 5.5 per cent of Canada's total economic activity in 2017. The COVID pandemic has only underscored the importance of digital connectivity as many workers have transitioned to working from home with socially-distanced modes of interaction with co-workers. Three important characteristics create technical challenges that distinguish fiber-optic communications from other communications technologies: (a) per-channel data transmission rates are very high (often exceeding 400 Gb/s per wavelength)-this severely constrains the allowable complexity of any proposed signal processing and decoding algorithms; (b) the target bit error probability is very low (one error per petabit)-this means that communication strategies allowable in other communication scenarios (such as wireless) may not be applicable to fiber-optic communications; (c) the long-haul fiber-optic channel is nonlinear (pulse propagation in one polarization is governed by the notorious nonlinear Schrödinger equation)-the nonlinearity has so far defied exact information-theoretic analysis, yet it is the dominant impairment limiting the performance of today's systems. Keeping these characteristics in mind, the overall aim of this project is to design new high-performance low-power forward error-correction (FEC) coding, shaping, and modulation schemes specifically tuned for nonlinear fiber-optic channels. The FEC subsystem is one of the most power-consuming modules in optical transceivers. While soft-decision decoding algorithms have superior performance compared with hard-decision algorithms, they consume a factor of 5 to 7 times more power, even with careful design and optimization. It is of great interest to find schemes that provide the best trade offs between decoding performance and power consumption.