An improved hybrid coding system for very noisy channels
Agarwal, Ramesh Chandra
Huband, Frank L.
Master of Science
Sequential-Algebraic hybrid coding schemes seem to provide efficient error reduction for very noisy channels. One such scheme was proposed by Huband and Jelinek. This hybrid scheme requires large amount, of shift- register storage at the transmitting end. Also, this hybrid scheme does not utilize the available sequential decoding time efficiently. In this thesis, modifications, are proposed to the encoding and decoding schemes of the scheme of Huband and Jelinek. The proposed modification to the encoder eliminates the long shift register requirement, at the cost of additional convolutional encoders. The proposed modification to the decoder utilizes the available sequential decoder time more efficiently, and thus significantly reduces the block erasure probability for the hybrid scheme. The proposed scheme optimally allocates the available sequential decoder cycles to sequential -tracks within a block, so as to minimize the block erasure probability. Upper and lower bounds on the erasure probability for a given allocation are obtained and shown to agree closely with one another. It is shown that the later tracks have available almost all the "equivalent decoder cycles" of the previous unerased tracks.