Reduced dimension equalizer and interference canceller for MIMO-OFDM

In this paper, an interference cancellation and equalization method is proposed for a multiple-input, multiple-output orthogonal frequency division multiplexing (MIMO-OFDM) system. Standard equalization techniques such as zero-forcing (ZF), and adaptive array processing techniques such as per-tone or time-domain equalization, may require large amounts of training and memory for storage of the weight coefficients. In a MIMO-OFDM system with N carriers, ZF considerations help dictate required equalizer lengths, which are characterized by the solution space associated with an underdetermined system of linear equations. In the practical case where the cyclic prefix length, and hence the channel length, is a fraction of the OFDM frame duration, it is shown that the respective equalizers to all tones (across antennas) are characterized by the basis of a data-independent, low-dimensional subspace. A structured per-tone technique is proposed that utilizes the extra degrees of freedom afforded by having multiple antennas at the receiver in order to cancel interfering signals.