Genetic algorithm applied to beamspace-multiple-input and multiple-output single-radio frequency front-end reconfigurable transceivers

The multiple-input and multiple-output (MIMO) capabilities of electronically steerable passive array radiator (ESPAR) antennas in the beamspace (BS) domain have designated them a strong candidate for BS-MIMO implementations using transceivers with a single-radio frequency front-end. The core functionality of the ESPAR is based on the multiplexing of information symbols in the BS domain, using orthogonal basis patterns. The radiated pattern of the antenna is produced as the linear combination of the basis patterns weighted by the information symbols to be transmitted. The adequate approximation of the desired radiation pattern produced by the antenna is of high importance in order to retain the orthogonality between the basis patterns and the corresponding information streams. The beamforming in ESPAR antennas is achieved by applying the appropriate set of loads at the parasitic elements. Nevertheless, because of the non-linear relationship between the radiation pattern and the loads of antenna elements, the calculation of these loads is not a trivial task. This paper presents a novel genetic algorithm (GA), designed to calculate the loading values of the parasitic elements of an ESPAR used by BS-MIMO systems in realistic channel environments. The performance evaluation of the proposed GA reveals a robust and valuable optimisation tool even for demanding applications.