A quadratic complexity subspace algorithm for blind channel shortening

The newly introduced blind processing technique, known as channel compaction, has made it possible to design blind subspace techniques with an unprecedented quadratic complexity, as compared to existing subspace techniques who have a cubic complexity. The principle has been applied successfully to design blind identification/equalization methods for SIMO/MIMO channels. This paper presents the first application of blind compaction to blind channel shortening, of utility to OFDM systems where it reduces the channel memory, and as a consequence, the bandwidth loss due to the guard period. Simulation results show compaction to be more efficient for shortening than for identification or equalization.