Implementation of low-complexity suboptimal turbo decoders using greedy algorithms

Turbo coding is one of the best channel coding procedures presented to the coding community in the recent history of channel coding. The efficient error correcting capability of turbo codes is attributed to its decoding method. The purpose of this thesis is to design and implement a novel decoding method which is simpler than the existing decoding methods thereby reducing the decoding complexity to a reasonable level. The novel method of decoding is based on a greedy algorithm which involves the search of proper error patterns that aims at reducing the overall global metric. This novel method is implemented for a convolutional encoder and then for the turbo encoder with both recursive and the non-recursive component structures. It is found that this new method of decoding is simpler than older methods and also gives encouraging performance --Abstract, page iii