TABLE OF CONTENTS Double-binary elementary codes provide better error-correcting performance than binary codes for equivalent implementation complexity [89]. And also, a parallel concatenation of Circular Recursive Systematic Convolutional codes (CRSC) [90] makes convolutional turbo codes efficient for coding of data cells in blocks. The double-binary CRSC codes were adopted in the DVB-RCS standard for their excellent performance as an alternative to the conventional scheme consisting of the concatenation of a convolutional code and a RS code.
The codes investigated in this chapter are constructed via parallel concatenation of double-binary CRSC codes by a non-uniform interleaver. Circular coding is a kind of "tail-biting" technique that avoids reducing the code rate and increasing the transmission bandwidth. The influence of puncturing and suboptimal decoding algorithm, Max-Log-MAP algorithm, are less significant with double-binary turbo codes than with binary turbo codes. Using double-binary codes, the latency of the decoder is halved. Double-binary CRSC code could be easily adopted for many applications, for various block sizes and code rates, with retaining excellent coding gains.
For efficient convolutional turbo coding, the number of memory elements is a key consideration since the component codes with small constraint lengths ensure convergence at very low signal to noise ratios and the correlation effects are minimized [88]. Moreover, reasonable constraint lengths make hardware implementation on a single integrated circuit possible since the material complexity of the decoder grows exponentially with the code memory. The solution chosen uses component codes with memory ? =...
