A scalable performance-complexity tradeoff for vector precoding by partial perturbation

We explore a performance-complexity tradeoff for vector precoding (VP) for the downlink of multi-user multiple input multiple output (MU-MIMO) systems. We introduce this tradeoff by employing a partial perturbation in the vector precoding, where the perturbation optimization is applied to a subset of the transmitted symbols. This greatly reduces complexity compared to conventional VP where the optimization involves all transmit symbols, at the expense of a reduced performance. We show that this tradeoff is scalable by means of the number of users to be perturbed and explore how it can be optimized by changing the size of this subset of users. The presented results show that partially perturbing the data provides a favorable performance-complexity tradeoff compared to conventional VP and increases the power efficiency of VP.