Coding and signal space diversity for fading and impulsive noise channels with side information

A general model for memoryless fading and impulsive noise channels with perfect channel state information (CSI), including e.g. the memoryless Rayleigh fading channel with CSI, is considered. Lower bounds on the pairwise error probability (PEP) and the codewords achieving these bounds are derived. The importance of signal space diversity is pointed out. A large class of unitary rotation matrices is determined for which - if used for encoding $the maximum PEP in the code approaches the lower bound as the codeword length is increased.