Symbol-wise processing implementation of semi-iterative turbo principle in multi-hop relay networks

The latency presents one of the major problem in communication systems. The seriousness increases in more complex physical scenarios, where the latency is accumulated in sequent nodes. We propose a novel concept of a latency reduced distributed iterative decoder design in a multi-node wireless network. The concept is based on the traditional turbo decoder, where we exploit the topology of the network to decompose the iterative decoder among more physically separated nodes. In this concept, the relay-part of the decoder computes symbol-wisely a soft metric, which is then transmitted through a wireless link into the destination-part of the decoder. The iterative process in the destination-part of the decoder has therefore only limited capabilities for iterations via the relay-part of the decoder. We formulate an optimization task for the proposed concept, design an initial (add-hoc) decoder decomposition and evaluate its properties. Since fundamental limits of the proposed optimization task are not available so far, we compare the proposed decomposition with the conventional turbo decoder.