AI-Assisted Dynamic Port and Waveform Switching for Enhancing UL Coverage in 5G NR

Abstract

The uplink of 5G networks allows selecting the transmit waveform between cyclic prefix orthogonal frequency division multiplexing (CP-OFDM) and discrete Fourier transform spread OFDM (DFT-S-OFDM) to cope with the diverse operational conditions of the power amplifiers (PAs) in different user equipment (UEs). CP-OFDM leads to higher throughput when the PAs are operating in their linear region, which is mostly the case for cell-interior users, whereas DFT-S-OFDM is more appealing when PAs are exhibiting non-linear behavior, which is associated with cell-edge users. Therefore, existing waveform selection solutions rely on predefined signal-to-noise ratio (SNR) thresholds that are computed offline. However, the varying user and channel dynamics, as well as their interactions with power control, require an adaptable threshold selection mechanism. In this paper, we propose an intelligent waveform-switching mechanism based on deep reinforcement learning (DRL) that learns optimal switching thresholds for the current operational conditions. In this proposal, a learning agent aims at maximizing a function built using available throughput percentiles in real networks. Said percentiles are weighted so as to improve the cell-edge users’ service without dramatically reducing the cell average. Aggregated measurements of signal-to-noise ratio (SNR) and timing advance (TA), available in real networks, are used in the procedure. In addition, the solution accounts for the switching cost, which is related to the interruption of the communication after every switch due to implementation issues, which has not been considered in existing solutions. Results show that our proposed scheme achieves remarkable gains in terms of throughput for cell-edge users without degrading the average throughput.