Construction of coset-based low rate convolutional codes and their application to rate turbo-like code design

In this paper, we propose a novel method to construct low rate recursive convolutional codes. The 'overall' convolutional code has a block code and a simple recursive convolutional code as building blocks. The novelty of this type of convolutional code is that it uses coset leaders in order to distinguish the signals that originate from different states. Several of these codes are then used to construct low rate turbo-like codes. Apart from being bandwidth efficient, simulation results show that these codes outperform the best known low rate turbo-like codes, the turbo-Hadamard codes (THC), in additive white Gaussian noise (AWGN) channel for interleaver sizes of practical interest.