Blind estimation of output labels of SIMO channels based on a novel clustering algorithm

This paper addresses the problem of data detection for digital communications employing space diversity reception, where the system model contains a single-input multiple-output (SIMO) vector channel. The received vector corrupted by additive white Gaussian noise (AWGN) is modeled as the noisy output of a finite state vector Markov source. Subsequently, a discretely valued basis of minimal dimensionality is identified for the input space. Estimation of the output labels associated with this basis allows labeling of the state transition diagram. For this purpose, certain identifiable characteristics of the output sequences of the Markov source are used to classify its states and generate an initial codebook for a vector quantizer used to restore the output levels.