Loading...

Analysis of Universal Coding Algorithms for Some Special Classes of Information Sources

Nangir, Mahdi | 2012

763 Viewed
  1. Type of Document: M.Sc. Thesis
  2. Language: Farsi
  3. Document No: 43413 (05)
  4. University: Sharif University of Technology
  5. Department: Electrical Engineering
  6. Advisor(s): Aref, Mohammad Reza; Behroozi, Hamid
  7. Abstract:
  8. The coding and compression of information sources with special probability distribution is the center of concentration of researchers in this field in the last decades. In this dissertation, after a concise introduction to source coding, we focus on the universal source coding and data compression from an algorithmic and information theoretic point of view. First we introduce the fundamental concepts of source coding and data compression. Then, we analysis some algorithms on integer coding and verify their universality and asymptotically optimality. Usually statistics of information sources is not available in communication systems and storage devices, so that applying universal source coding algorithms is necessary. After introducing some universal algorithms, we provide a new algorithm that has shorter codeword length than Omega Elias coding which is one of the best universal source coding methods. Then we verify problem associated with this algorithm and provide solutions. In the next part then we introduce the fundamentals of text compression algorithms and its basis design methods and see the problems when we extend these algorithms to the sources with infinite alphabet. Similar to channel coding algorithms, source coding methods may have memory as well. We apply this idea to universal source coding by considering block source coding idea and "super codeword" concept. We obtain variety of freedoms to improve compression quality. In the last section of this thesis we also introduce the important concept of "information partitioning" of the output sequences of the discrete sources. With some examples we show that how compression of some characteristics of the output sequence instead of original output sequence, can provide a better compression ratio.
  9. Keywords:
  10. Source Coding ; Data Compression ; Universal Compression Algorithm ; Universal Sources Coding ; Text Compression ; Code Redundancy

 Digital Object List

  • محتواي پايان نامه
  •   view

 Bookmark

No TOC