Kurzfassung: |
In this paper, the 3GPP downlink eigenbeamformer scheme is investigated. The eigenbeamformer scheme performs an eigendecomposition (EVD) of the spatial longterm covariance matrix to obtain beamforming weights. Diversity gains are obtained due to switching between socalled eigenbeams. Since the EVD is carried out by the mobile station, computationally efficient algorithms are required. We show that the computational efforts can be drastically reduced by exploiting the quasi-stationarity of the long-term spatial covariance. For a given feedback rate, the performance of closed-loop schemes drop at high velocities due to delay introduced by feedback of side information. To cope with this problem, an efficient spatiotemporal prediction scheme is proposed. Our simulations, in which realistic coding, feedback delay, channel estimation and quantization of eigenbeams are considered, show that for a given target block error rate, the operational range, i.e. the maximally allowed speed of the mobile station, can then be significantly increased. |