Error correcting least-squares Subspace algorithm for blind identificationand equalization

A subspace based blind channel identification algorithm using only the fact that the received signal can be oversampled is proposed. No direct use is made in this algorithm of either the statistics of the input sequence or ev en of the fact that the symbols are from a finite set and therefore this al gorithm can be used to identify even channels in which arbitrary symbols ar e sent. Using this algorithm as a base and using the extra information whic h becomes available when the transmitted symbols are from a known finite se t, the EC-LS-Subspace algorithm is derived. The EC-LS-Subspace algorithm op erates directly on the data domain and therefore avoids the problems associ ated with other algorithms which use the statistical information contained in the received signal directly. In the noiseless case, if some conditions are met, it is possible for the proposed Basic Subspace algorithm to identi fy the channel exactly using an observation interval of just (J + 2)T, if t he length of the impulse response of a channel is JT, T being the symbol in terval. In the noisy case, simulations have shown that the channel can be i dentified accurately by using a very small observation interval (comparable to (J + 2)T). (C) 2001 Elsevier Science B.V. All rights reserved.