Duality-Based Joint Clustering and Precoding for Cell-Free Distributed MIMO

We address the problem of user-centric cluster formation for the downlink of a distributed multiuser MIMO system where each spatial layer is associated with only a subset of transmission-points (TP). All clusters are jointly optimized with MIMO precoders and transmission powers. The resulting mixed continuous-discrete optimization problem is intractable even for moderately sized systems because of its large combinatorial search space. Our new approach avoids this problem by using an equivalent dual problem formulation instead. The advantage of the dual problem solution is that one user's choice of cluster is independent of other users' choices. This enables algorithms that scale well with the number of users. As an example, we propose an algorithm with a rapidly converging fixed point iteration as inner optimization block, and a subgradient-based outer iteration for controlling the per-TP powers. The performance gain is investigated by means of simulations.