Channel Coding for Massive IoT Satellite Systems

Concatenation of linear block codes enhances the performance with respect to the use of single linear block codes, while increasing the decoding complexity. In particular, the serial concatenation of linear block codes has proven to be an important tool for the design of powerful binary linear short codes. This was decisive for the discovery of new codes capable of approaching the bounds of the best possible codes at finite length. The class of convolutional codes, upon termination, are known to provide very powerful codes for large memory values at the encoder. However, the decoding complexity of these codes grows exponentially with their encoding memory, preventing the practical use of very large memories. The serial concatenation of convolutional codes with outer (cyclic redundancy check) linear block codes can reduce the needed memory of the convolutional code to achieve very good performances, at the expense of using a list decoding scheme. In this Master thesis we study the performance of such concatenation via weight enumerator and the performances of the serial-list and the parallel-list Viterbi algorithm as decoding algorithms with bounded list sizes. Lastly, we propose a suitable sorting network capable of reducing the time complexity of the parallel-list Viterbi algorithm when implemented in hardware.