This thesis addresses the problem of decoding high rate convolutional codes directly without resorting to puncturing. High rate codes are necessary for applications which require high bandwidth efficiency, like high data rate communication systems and magnet recording systems. Convolutional (rate k/n) codes, used as component codes for turbo codes, are preferred for their regular trellis structure and the resulting ease in decoding. However, the branch complexity of the (primal) code trellis increases exponentially with k for k/(k+1) codes, making decoding on the code trellis quickly impractical with increasing code rate. 'Puncturing' is the method traditionally used for generating high rate codes, which keeps the decoding complexity nearly the same for a wide range of code rates, since the same ?mother? code decoder is used at the receiver, while only the puncturing and depuncturing pattern is altered for changes in code rate. However, 'puncturing' puts a constraint in the search for the best possible high rate code, thereby resulting in a performance penalty, particularly at high SNRs.
| Identifer | oai:union.ndltd.org:ADTP/283995 |
| Date | January 2008 |
| Creators | Srinivasan, Sudharshan |
| Source Sets | Australiasian Digital Theses Program |
| Language | EN-AUS |
| Detected Language | English |
| Rights | Copyright Sudharshan Srinivasan 2008 |
Page generated in 0.1713 seconds