Error-Pattern-Correcting Codes: A New Approach to Error Correction Coding for Interference-Dominant Channels

错误模式校正码：符号间干扰是通信信道中传输错误的主要来源。本研究涉及这种通道的纠错。输入受限的ISI信道是本研究中特别感兴趣的，因为用于此类信道的有效编码方法在很大程度上仍然是一个开放的领域。这项工作的重点是1）开发低复杂度的输入受限的ISI信道的纠错方法，重点是高密度数据存储应用，和2）清除路径，以接近理论极限的纠错，从根本上改善复杂性/性能的权衡这样的信道。设计用于输入受限ISI信道的代码的一种有效方式是以信道输出处固有的主要错误簇模式为目标，因为ISI的结构通常引起相对小的一组不同错误簇模式的主导，而不管信道信噪比如何。这个集合可以很容易地使用距离分析或通过基于模拟或实验室测量的经验观察来找到。仅关注主导集，非常高速率的代码是可能的，具有优异的整体纠错能力。编码和解码必须高度适应少数已知的错误模式。这种类型的码可以作为独立码或作为较大编码系统中的关键构建块来应用，例如基于迭代软判决处理的turbo均衡器。虽然研究解决计算效率的方法，努力也是针对建立一个理论框架，以及了解理论性能限制的信道匹配码。

Error-Pattern-Correcting Codes: A New Approach to Error Correction Coding for Interference-Dominant ChannelsJaekyun Moon, University of MinnesotaAbstractIntersymbol interference (ISI) is a major source of transmission errors in a wide variety of communications channels. This research concerns error correction for such channels. Input-constrained ISI channels are of specific interest in this research, as efficient coding methods for such channels are still largely an open area. This work focuses on 1) developing low-complexity error correction methods for input-constrained ISI channels, with emphasis on high-density data storage applications, and 2) clearing a path to approaching theoretical limits of error correction for such channels with fundamentally improved complexity/performance tradeoffs. One effective way of designing codes for input-constrained ISI channels is to target at dominant error cluster patterns inherent at the channel output, as the structure of the ISI typically give rises to a dominance of a relatively small set of distinct error cluster patterns, regardless of the channel signal-to-noise ratio. This set can easily be found using distance analysis or through empirical observation based on simulation or laboratory measurement. Focusing only on the dominant set, very high rate codes are possible with excellent overall error correction capability. Encoding and decoding are necessarily highly tailored to the few known error patterns. This type of code can be applied either as a stand-alone code or as a critical building block in a larger coding system, such as a turbo equalizer based on iterative soft decision processing. While the research addresses computationally efficient approaches, efforts are also directed to establishing a theoretical framework as well as to understanding theoretical performance limits of the channel-matched codes.