A mobile device - or user equipment (UE) - using 5G is configured to operate with large bandwidths and a large number of MIMO layers to enable high-throughput applications. This, however, comes at the cost of high power consumption. A variety of applications run at the UE with widely varying throughput requirements. For applications with low throughput requirement, configuring the UE with large bandwidth and MIMO layers results in unnecessary power consumption. We propose to classify the traffic at the UE, and then configure the bandwidth and MIMO layers accordingly to meet the requirement of the current application without unnecessary power consumption. We evaluate our proposed method using data collected in commercial networks with commercial smartphones. The evaluation shows that the UE power consumption can be reduced by up to 60\% in the cell-center and 50\% in the cell-edge without degrading the quality-of-service (QoS).

https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=10336804