We present a novel sequential hypothesis test for multi user beam selection when knowledge about the SNR operating point of each user is not available. The proposed multi user sequential competition test reliably selects the strongest beam for each user while adaptively changing the average test length (the number of observations) according to the SNR operating point of each user. The performance of the proposed algorithm in terms of overall SINR is almost identical to the performance of the ideal (genie) beam selector when the packing ratio or the number of users over number of beams is small.
展开▼