Near-optimal pilot assignment in cell-free massive MIMO

The main source of performance degradation in cell-free massive MIMO is pilot contamination, which causes interference during uplink training and affects channel estimation negatively. Contamination occurs when the same pilot sequence is assigned to more than one user. This is in general inevitable, as the number of mutually orthogonal pilot sequences corresponds to only a fraction of the coherence interval. We introduce an algorithm for pilot assignment that has an approximation ratio close to 1 for a plausibly large number of orthogonal pilot sequences. It also has low computational complexity under massive parallelism.