Information processing in soft-output decoding

Recent literature presents methods for the analysis of concatenated coding schemes by solely characterizing the behavior of the component codes [4, 12, 14, 9, 7]. Component codes are analyzed either analytically using unique properties of special component codes, e.g., single–parity–check code or accumulator, or via simulations. The goals of this paper are to find fundamental insights into concatenated codes by analyzing the input–output relation of their components from an information– theoretic point of view. We derive a Information Processing Characteristic (IPC), which completely characterizes the behavior of a coding scheme comprising encoder, code and decoder for the general class of linear codes. For time invariant convolutional codes it is studied how the IPC can be obtained in practice.