New Perspectives on Convolutional Coding/Decoding for 6G

In recent years, significant efforts were invested in (de-)coding of convolutional codes and their concatenations by leveraging alternative representations which aim to lower the decoding complexity. One example is the Local-SOVA algorithm that avoids redundant computations by reformulating the BCJR algorithm towards the notion of providing partial extrinsic information. This latter is now provided along path metric computations in the code trellis leading to a significant reduction in complexity. For high coding rates where most redundant computations are made, the trellis compression technique represents an appealing solution to reduce the complexity of warm-up calculations for BCJR-based turbo decoding. In addition, BP based decoding solutions have come a long way to bridge the gap between initially disappointing results and BCJR performance. In this work, we propose to extend and combine several new representations of the code targeting reduced decoding complexity for short convolutional codes and their concatenations. Based on our new perspective, we give an outlook on convolutional based FEC for 6G.