Quasi-deterministic space-time second-order blind sequence estimation

A fast, robust, second-order, blind space-time algorithm is presented for identifying the symbols from a transmitting mobile source, received at the base station of a wireless uplink channel, by estimating the effective inverse channel via a quasi-deterministic approach. This technique achieves a trade-off between speed and robustness in the presence of increasing additive channel noise by using estimates of the additive-noise variance and short-term temporally averaged estimates of an analogue of the ensemble averaged autocorrelation of the received sequence. Using these psuedo-autocorrelations provides a faster sequence-estimation procedure than existing deterministic techniques which merely seek to use the properties of the signal subspace, ignoring the correlation between successive received symbols-an approach that has been confined to statistical techniques. Likewise, inclusion of the additive noise variance (which needs to be computed less frequently than the block-processing rate) lends robustness to the sequence estimation procedure. Since no other statistical averages are computed, the proposed technique is sufficiently fast in both time- and frequency-selective fading environments. Further, no assumptions are made regarding the distribution of the received signal, the dimensionality of the signal subspace, or other properties of the symbol alphabet. Compared with existing blind identification techniques, this approach has mild identifiability conditions, low computational requirements and superior performance characteristics