Cooperation has been recently introduced to mitigate
the impairments imposed by fading on a communication
channel. A cooperative transmission, in which the destination
may overhear all transmission attempts, benefits from incremental
redundancy techniques, achieving a higher throughput
and reducing the packet delay and the outage probability. This
paper presents a theoretical framework to assess the limiting performance
of efficient coding techniques in Decode-and-Forward
cooperative scenarios. In this paper it is shown that suitable
designed codes, such as punctured turbo codes, approach the
limiting performance in all the studied scenarios. The paper also
addresses the design of punctured turbo code schemes capable
of performing close to the presented limiting bound. A novel
Genetic Algorithm (GA) is introduced for designing incremental
puncturing schemes suited to cooperative transmissions. The
algorithm uses a recently proposed tri-dimensional (3D) EXtrinsic
Information Transfer (EXIT) chart technique to model the
decoding behavior of parallel concatenations of non-systematic
periodically punctured convolutional codes. It is shown that the
design algorithm allows contriving capacity approaching codes
for the source and the relay. The paper further shows that the
GA aided designed codes closely approach the theoretical results,
outperforming previously proposed systematic incremental and
complementary schemes.
