Abstract :
This contribution proposes a new approach to channel coding, which is designed for combined source-channel coding on channels with memory. The simulations are based on a digital model of the HF channel, which takes into account the dependences between errors. Contrary to approaches assuming memoryless channels like additive white Gaussian noise (AWGN) channels, slowly changing channels, or channel coding methods which maps channels with memory onto memoryless channels using fully bit-interleaving, our approach uses solely a symbol-interleaving (4-bit symbols) with a depth of 30 to break up the structures of bursty errors. After a short review of the channel model used for our investigations, a description of the source-channel coding framework follows. Then, the algorithms of the proposed system are explained with numerical results showing its effectiveness
