A Neural Network Convolutional Decoder (NNCD) - Towards software defined radio

This paper outlines the concepts of a Neural Network Convolutional Decoder (NNCD), having the ability to implement multiple decoding functions/arrangements without the need to change the Decoder architecture itself or to re-write software code. In application to a Software Radio environment, the technique would not require that large amounts of executable code be downloaded over-the-air, simply that the Neural Network Weights are. This would result in a saving on bandwidth requirements and on the time needed to download new functionality (thus ultimately resulting in a cost saving).