Correlation Matrix Bordering for Optimal Massive MIMO Power Allocation Algorithms

The dramatic increase in the amount of communicating antenna elements in Massive and Ultra-Massive MIMO systems on the one hand necessitates the application of the optimized (in case of multipath fading) power allocation strategies and on the other exploiting fast and efficient algorithms of channel correlation matrix estimation. The research addresses both problems by proposing a bordering procedure applied to the mercury/water-filling power allocation algorithm. The performed analysis demonstrated its sufficient dominance (in terms of overall system capacity) over the equal gain algorithm and insignificant loss, compared to the classic water-filling method. It was shown that bordering of the correlation matrix of Massive MIMO system up to pentadiagonal structures introduces insignificant capacity deviation, hence can be successfully exploited in practice for its estimation via fast recursive procedures.