Error Control for Terabit Links: Spatially-Coupled Staircase Codes

在过去的二十年里，基于光纤的传输网络实现的数据传输速率呈指数级增长，但随着传输系统接近基本的信息论极限，这种增长可能很快就会达到所谓的“容量紧缩”。该项目致力于解决一项关键技术-前向纠错，或“FEC”-该技术将使未来的低成本互联网能够实现TB/S/波长光通信链路，并可能在更长一段时间内推迟容量紧张。 由于未编码传输的效率太低，因此，以接近基本的“香农极限”的吞吐量操作信道的唯一方式是实现先进的FEC。这使得FEC成为所有现代通信系统的关键组件。现代光通信系统实现的高数据速率(目前正在向400 Gb/S/波长甚至更高的方向发展)带来了低速率系统设计者不会面临的挑战。因此，需要进行新的研究来找到纠错方案，这些方案一方面从信息论的观点来看是高效的，另一方面又是便于实施的。阶梯代码就是这样一个家族：它们在编码性能和实现复杂性之间进行极好的折衷。 在这个项目中，我们计划研究阶梯码和相关的空间耦合数据通信方案。这项建议有两个基本目标和一个实际目标：(A)增加我们对这些代码的理论理解；(B)增加这些代码的潜在应用范围(例如，将它们与高阶调制相结合，或允许它们利用软信息)；以及(C)在真实的可编程硬件平台上开发实用算法。该项目还有一个重要目标，那就是培养新一代专家，能够为未来光通信系统的设计带来严格的信息理论上合理的方法。这项工作的成果将使支持未来互联网所需的TB/S/波长光通信链路成为可能。

For the past two decades, the data transmission rates achieved by optical-fiber-based transport networks have increased exponentially, but this increase may soon reach a so-called "capacity crunch" as transmission systems approach fundamental information-theoretic limits. This project addresses a key technology---forward error correction, or "FEC"---that will enable tomorrow's low-cost Internet-enabling Tb/s/wavelength optical communication links and may hold off the capacity crunch for a while longer. The only way to operate a channel at a throughput near the fundamental "Shannon limit" is by implementing advanced FEC, as uncoded transmission is simply too inefficient. This makes FEC a key component of all modern communication systems. The high data rates achieved by modern optical communications systems (today pushing towards 400 Gb/s/wavelength and beyond) creates challenges not faced by designers of lower-rate systems. Thus, new research is needed to find to find error-correction schemes that, on the one hand, are highly efficient from an information-theoretic viewpoint and yet, on the other hand, are implementation friendly. Staircase codes are one such family: they operate at an excellent tradeoff point between coding performance and implementation complexity. In this project, we plan to study staircase codes and related spatially-coupled data communication schemes. This proposal has two fundamental aims and one practical one: (a) to increase our theoretical understanding of these codes; and (b) to increase the range of potential applications for these codes (for example, to combine them with higher-order modulation, or allow them to take advantage of soft information); and (c) to develop practical algorithms on real programmable hardware platforms. This project also has the important objective of training a new generation of specialists able to bring a rigorous information-theoretically sound approach to the design of future optical communication systems. The results of this work will enable the Tb/s/wavelength optical communication links required to support the Internet of the future.