Space-time filters are very useful to recuperate a desired user signal in a MIMO environment with selective frequency channels. One of the main obstacles to implement space-time filters in real time systems is due to the numerical complexity. Indeed, the large number of coefficients to be adapted requires a great computational cost that may be prohibitive for real systems. In this work we propose the use of orthonormal bases in the filter structure in manner to reduce the number of coefficients and consequently the overall numerical complexity. Actually, we show an equalizer structure formed by a set of generalized orthonormal functions. In order to achieve a noticeable numerical complexity reduction, we also discuss a method for the orthonormal function poles parameterizing based on the channel characteristics. The proposed structure performances are compared with conventional FIR space-times filters. Trained algorithms are employed for filter weight adapting. Our simulations show that the proposed structure leads to enhanced performances offering an alternative reduced complexity solution for MIMO communication channel equalization problem.
展开▼