A Machine Learning Approach for CQI Feedback Delay in 5G and Beyond 5G Networks

5G and Beyond 5G Networks apply Adaptive Modulation and Coding to adjust the downlink modulation order and coding rate according to the channel condition, reported by the user equipment. However, the delay incurred in this feedback process may make the channel quality indicator (CQI) outdated and cause severe degradation in the user communication. This paper proposes a machine learning-based approach to deal with the outdated CQI problem. It takes into account the UE context, current signal-to-interference-plus-noise ratio (SINR), and the delay length to compute the updated SINR to be translated into a CQI value. Our proposal acts as a multi-variable function and runs at the UE side, neither requiring any modifications in the signalling between the 5G base station (gNB) and the UE nor overcharging the gNB. Results in terms of mean squared error (MSE) by using 5G network simulation data show its high accuracy and feasibility to be adopted in 5G networks.