Asymptotic distance and convergence analysis of braided protograph convolutional codes
Marcos B. S. Tavares, Michael Lentmaier, Gerhard Paul Fettweis, Kamil Sh. Zigangirov · 2008
We analyze a class of LDPC convolutional codes that are constructed from tightly braided convolutional base codes by a lifting procedure. For these braided protograph convolutional codes, we show that the distances grow linearly with the constraint length, and we present lower bounds on their asymptotic segment distance and free distance as well as on the asymptotic minimum distance of their tail-biting versions. With some constraints imposed on the lifting permutations, braided protograph convolutional codes can also be decoded as turbo-like codes by iterative application of the BCJR algorithm. For this case, we derive an explicit upper-bound on the asymptotic decoding error probability as a function of the number of iterations.