EXPECTED ERROR LENGTH AND PROBABILITY OF LOSS OF SYNCHRONISM OF CERTAIN ERGODIC CODES,
Abstract
Several ergodic codes are compared with respect to both expected error length and probability of loss of synchronism due to channel errors as a function of compressibility. For these compari sons, assume the existence of a binary source without memory which emits 'ones' with a proba bility of k 0.01 and ''zeroes'' with a probability of 1 - k 0.99. The source bits are then passed into a coder which, by use of a neat input code, creates words from the source bits and, by use of output code, decomposes the words into channel bits. Both input and output codes are unitary (i.e., no word begins another word) but the output code need not be neat. The com bination of an input and an output code is called a coding scheme. Two schemes using the same input and output codes but differing in the number of code words ''n'' are said to belong to the same code scheme family. This paper deals with a comparison of different code familes. A listing of the code families considered and the equations for their compression appear. The coding schemes are compared on the basis of loss of synchronism and they are compared on the basis of expected error length. Hamming output codes are discussed and compared to the codes listed here with respect to expected error length. (Author)
Document Details
- Document Type
- Technical Report
- Publication Date
- Apr 10, 1963
- Accession Number
- AD0418419
Entities
People
- Stephen P. Morse
Organizations
- New York University Tandon School of Engineering