Deep Learning-based Channel Estimation in High-Speed Wireless Systems With Imperfect Frame Synchronization
This paper considers an application of deep learning for channel estimation with imperfect frame synchronization in mobile communication systems. Without prior knowledge of the channel model and its characteristics, the proposed method can dynamically estimate and track channel transfer function variations based on received pilot symbols. Furthermore, this method is applicable in practical scenarios, as it considers imperfect frame synchronization and channel estimation for high-speed wireless communication scenarios. The performance and practical feasibility of the deep learning (DL)-based models are assessed by taking into account realistic frequency-selective fading scenarios. Numerical results demonstrate that the proposed method performs better for practical signal-to-noise ratios than the state-of-the-art approaches. In addition, the fine frame offsets are estimated and compensated in the synchronization block with a DL-based algorithm, which outperforms the traditional fine frame synchronization algorithms.