Location-Aided User Selection and Sum-Rate Analysis for mmWave NOMA

Abstract

In this paper, we propose a user selection schemebased on location-aided interference prediction to reduce thetraining overhead in a non-orthogonal multiple access (NOMA)system. First, we cluster the users based on their locationinformation, enabling the use of non-orthogonal pilot sequencewithin a cluster and orthogonal pilot sequence between differentclusters to reduce the uplink pilot training length. Secondly,we exploit the location information in the computation of thecovariance matrices, enabling the prediction of the interferencebetween users. The predicted interference is employed to selectthe set of users with minimum interference for uplink channelestimation and downlink NOMA data transmission. Finally,the achievable sum-rate of the massive multiple-input multiple-output millimeter wave NOMA system is analyzed. The analyticaland numerical results reveal that the location information canbe exploited for user selection to reduce the effect of pilotcontamination, enhancing the uplink channel estimation anddownlink achievable sum-rate.