An effective initialization of interference cancellation algorithms for distributed MIMO systems in wireless datacenters

We discuss a method for interference cancellation in distributed MIMO systems which accommodate K Tx and Rx node pairs (links) and d streams for each link. The method is a geometric iterative optimization of SIR by natural gradients on matrix manifolds. We also apply partial linear zero-forcing to obtain more interference-suppressive initial points that can improve convergence property of the iterative algorithm. The proposed method is applied to a channel model for a typical datacenter and the simulation results show that our method can improve SIR and achieve higher sum rate at high SNR.