ADAPTIVE LINEAR PREDICTION BASED DOWNLINK BEAMFORMING FOR WIRELESS COMMUNICATION SYSTEMS
In Time Division Duplex (TDD) wireless communications, downlink beamforming performance of a smart antenna system can be degraded due to the variation of spatial signatures in the fast fading environment. To mitigate this, prediction based downlink beamforming can be applied, which relies on using updated weight vectors via adaptive linear prediction of spatial signatures in the downlink interval based on their autoregressive (AR) modeling in the uplink interval. In this study, the effectiveness of employing predicted spatial signatures as downlink weight vectors is demonstrated under varying mobile speed (V), prediction filter order (p) and the number of multipath (L) conditions. It is observed that in the event that Doppler shifts in the multipaths are integer multiple of Doppler shift in the fist path (fixed Doppler shift), prediction based beamforming achieves better SNR improvements in the received signal at the mobile terminal with increasing V, p, and L.