Performance improvements for sector antennas using feature extraction and spatial interference cancellation

Effective utilization of the spatial domain enhances the capacity of a mobile radio network. A common technique is to use sector antennas, where the sectors are formed by weighting the outputs from the antenna elements. This results in spatial domain selectivity, which significantly improves the signal-to-(noise and interference) ratio in the received signals. However, the operation of the sector antenna will be limited by the sidelobes of the corresponding beam patterns. By introducing a blind spatial interference canceler that combines the fix beamformers in the sector antenna with blind signal separation, a significant improvement in the multiuser interference suppression can be achieved. Thus, it will be able to efficiently handle the near-far problem, where the users are received with different power. The blind signal separation is performed by an independent component analysis algorithm. The convergence rate of the algorithm is significantly improved compared to the standard formulation by taking into account the modulation format. The algorithm is further improved by introducing a forgetting factor on the weight update. The blind spatial interference canceler is evaluated by simulations using the mean square error and the bit error rate as quality measures. The results show that the mean square error obtained from the blind spatial interference canceler is within 0.5 dB from the optimum Wiener solution for signal-to-noise ratios greater than 0 dB.