Machine Learning-Based Channel Estimation for 5G New Radio

In this letter, we present a novel approach to channel estimation in 5G New Radio uplink utilising machine learning. The proposed method offers a continuous adaptation to dynamic channel conditions by performing online training. Periodic training allows for continuous learning and adjustment, effectively capturing and responding to variations in channel characteristics. We examine the proposed method using the normalised mean squared error of the estimated channel coefficients, comparing it to the ideal channel. Furthermore, we evaluate the bit error rate performance of the proposed method for higher-order modulation schemes. The simulation results demonstrate that the proposed channel estimation method achieves a lower normalised mean squared error and bit error rates compared to reference methods even in higher modulation schemes. Further, the proposed slot arrangement has high spectral efficiency.