Improved diversity-multiplexing tradeoff for underwater acoustic channels based on distributed channel coding

Multiuser cooperative schemes usually rely on relay selection or channel selection to avoid deep fading and achieve diversity for all source nodes while maintaining acceptable spectral efficiency. This requires estimating the channel state information (CSI) and reporting it back to the resource allocator. Then the resource allocator needs to inform each node about its task. In some applications, it is impractical to collect the required CSI. For example, in underwater acoustic channels, the low speed of the acoustic wave (500 m/s) leads to a very long delay between the measured CSI and the reported CSI, making the provided CSI severely 'outdated'. To remedy this, we propose a new distributed coding scheme to avoid deeply faded channels and achieve cooperative diversity for multiuser applications where CSI is not available. We consider a network consisting of multiple sources, multiple relays, and a single destination. The proposed scheme is based on a distributed implementation of linear block codes and achieves a better diversity-multiplexing tradeoff (DMT) compared to the competitive schemes. In this scheme, each relay node implements a column of the generator matrix of the code. Each relay receives the symbols from one or more source nodes and performs modulo-q addition on the decoded symbols and retransmits the result to the destination. Finally, soft decoding is employed at the destination to retrieve the transmitted information. We prove that the proposed scheme achieves diversity d_min for the end-to-end (E2E) performance, where d_min is the minimum distance of the underlying code.