Blind separation of OSTBC signals using ICA neural networks

Space-time block coding (STBC) is a widely studied transmitter diversity scheme able to increase the capacity of the scatter wireless channel. Most of the existing detection techniques at the receiver assume perfect channel knowledge, requiring the use of training sequences. Independent component analysis (ICA) neural networks are a class of non-trained signal processing methods able to separate linearly mixed independent signals. In this paper, we consider the possibility of applying ICA to blindly detect orthogonal STBC without channel knowledge. The recovered signals may also be used to estimate the unknown channel parameters, complementing other classical receiver schemes.