Blind MLSE equalization using subspace methods

In this paper it is proposed to design a matched filter for a single-input multiple-output (SIMO) channel directly from the outputs without knowledge of any transmitted sequence. It Is shown that such a design can be completed directly from the second-order statistics of the received signal by examining properties of the -spaces of a set of autocorrelation matrices. It is also shown that this can directly be used to simply derive an equalizer for the system as either a decision-feedback structure or Viterbi decoder. Performance is demonstrated through simulation.