Adaptive Frequency Resolution for Downlink Beamforming in 5G NR

Abstract: As the demand for higher data rates and lower latencies increase, together with the use of higher frequency bands, MIMO and beamforming have become an essential part of the 5G NR technology. Calculating the beamforming weights is a costly process, and reducing the frequency resolution of the weight calculations would save a significant amount of computational power. Which resolution is suitable to use depends on the channel’s frequency selectivity, which can be measured by for instance the RMS Delay Spread. This thesis analyzes two algorithms for adaptive resolution of the beamforming weights. The main focus of the work lies on the first algorithm, one that analyzes the channel estimation from the uplink and selects a suitable resolution for the downlink beamforming weight calculations. First, suitable thresholds for the resolutions and decision parameters were estimated through simulations, after which the performance of the algorithm could be evaluated. The impact of number of antennas and number of UEs was also explored. The second algorithm is simpler in nature, where the resolution is maximized while maintaining a lower limit of errors. The development and evaluation of the algorithms was performed through Matlab simulations.