Performance of punctured pragmatic codes

The paper studies punctured pragmatic (PPR) codes which are constructed by using the convolutional code with rate r₁=1/2 from the original pragmatic coding approach and puncturing in order to get rates r>1/2 resulting in total transmission rates of 2<R<3 bit/symbol. Considered are PPR codes as inner codes of concatenated coding schemes. In order to use the original trellis diagram for decoding a new metric increment has to be defined which approximates the maximum-likelihood metric increment. The loss in performance is very small. By comparing the decoding complexity and bit error rates of PPR codes with competing multi-dimensional TCM codes of the same rate R it turns out that multi-dimensional TCM codes are somewhat more efficient for R>2 bit/symbol