CAREER:Advances in Universal Data Compression with Applications to Joint Source and Channel Coding

本研究的目的是开发数据压缩的几个未开发领域，以及在其他应用中利用通用数据压缩技术，包括生物建模和联合源信道编码的新方向。研究集中在四个主题上：(a)基于联合信源-信道的通用源代码代码的设计，(b)研究大型和未知源字母的通用压缩，(c)设计先进的通用编码技术，用于非传统的，但更现实的，具有实际实现的数据模型，以及(d)研究随机访问无损压缩。常用的信源信道联合编码技术在恶劣信道条件下存在严重的同步问题，且无法解决信源统计未知时的普适性问题。本研究开发了解决这些问题的技术，甚至可以在冗余信道信息流的信道解码性能中获得“免费”增益。常见的压缩方案假设数据来自一个已知的字母表，它有一个“标准的”固定的或不断变化的统计模型，它由一个长序列组成。然而，(a)存在大量未知字母的压缩应用，例如由单词组成字母的文本压缩，(b)大多数真实数据序列通常既不是平稳的，也不是不断变化的统计数据，(c)在大型压缩数据库中随机访问是必要的。研究者研究了这三个非传统问题。研究工作结合了严格的理论结果的发展，包括冗余和描述长度界限，经验测试，算法设计，重点是实用的低复杂度技术，以及拟议技术的实施。最后，本研究还探讨了通用压缩技术在生物序列分割和建模中的应用。

The purpose of this research is to develop several unexplored areas indata compression, as well as to utilize universal data compressiontechniques in other applications including biological modelling and anovel direction of joint source-channel coding. The research focuses onfour topics: (a) design of joint source-channel universal source codebased codes, (b) study of universal compression for large and unknownsource alphabets, (c) design of advanced universal coding techniques fornon-traditional, yet more realistic, data models with practicalimplementations, and (d) the study of random access lossless compression.Common techniques in joint source-channel coding suffer fromsevere synchronization problems in bad channel conditions and donot address universality issues when the source statistics areunknown. This research develops techniques to combat theseproblems, and even attain "free" gain in channel decodingperformance for redundant channel information streams. Commoncompression schemes assume that the data is from a known alphabet,it has a "standard" stationary or constantly changingstatistical model, and it consists of a long sequence. However,(a) there exist compression applications with large unknownalphabets, such as text compression where the words constitute thealphabet, (b) most real data sequences are usually neither stationarynor of constantly varying statistics, and (c) random access isnecessary in large compressed data bases. The investigator studiesthese three non-traditional problems. The research work combinesthe development of rigorous theoretical results includingredundancy and description length bounds, with empirical testing,algorithm design with focus on practical low-complexitytechniques, and implementation of proposed techniques. Finally,the research also investigates the use of universal compressiontechniques to segmentation and modelling of biological sequences.