Improving Neural Network Channel Estimation Robustness Against Doppler Effects

Neural networks for wireless channel estimation often fail on unseen channels caused by high-speed movement and Doppler shifts. This research proposes new design criteria for synthetic training data to help models generalize to time-variant channels without the high latency of online fine-tuning.