Designing convolutional codes used in multi-binary turbo codes

Multi-binary turbo codes (MBTCs) have a global performance superior to the classical turbo codes. They also offer some advantages to the uni-binary turbo codes as better convergence and lower error floor. The selection criterion proposed here was the convergence of the MBTCs. We used this criterion in two steps. The first step we made was an exhaustive search over memory-3 component convolutional codes set in conjunction with a random S type interleaver. In the second step we investigated a set of interleavers used in conjunction with the best codes selected from the first step. The searching results were materialized in two MBTCs which outperform with more than 0.2 dB, for Frame Error Rate, FER=10^-4, the multi-binary turbo-code with same characteristics, proposed in literature.