Limited feedback vector perturbation precoding by MinMax optimization

A vector perturbation (VP) scheme is proposed for the downlink of multi-user multiple input multiple output (MU-MIMO) systems with limited feedback. Instead of a computationally expensive sphere search used in conventional VP, the proposed scheme uses a MinMax Optimization to select the perturbation quantities. In addition, the proposed VP circumvents the need for receive-scaling by constraining the search of perturbing vectors to the area in the symbol constellation which is constrictive to the information symbols, i.e. the area where the distances from the decision thresholds are increased with respect to a distance threshold. Consequently, while conventional VP requires the feed-forwarding of the scaling factor to the receiver for correct detection, the proposed scheme does not require the scaling of the received symbols. This advantage is particularly pronounced in limited feedback scenarios where the scaling factors forwarded to the receiver are prone to quantization errors. As illustrated by the results, the error floor encountered in conventional VP in limited feedback scenarios is avoided in the proposed scheme.