Toward Intelligent Fading Channel Prediction: A Comprehensive Survey

Accurate knowledge of future time-varying channel state is a key enabler for adaptive transmission in current and future generation wireless networks. In this paper, we present a structured and comprehensive survey of advances in fading prediction, alongside a comparative analysis of the traditional and modern methodologies utilized in this domain. We place a significant emphasis on the shift towards machine learning techniques, which offer a promising avenue for enhancing prediction accuracy and operational efficiency. The survey delves into various predictive models, including neural networks, parametric models, and their applications across multiple input multiple output (MIMO) systems, highlighting their relative strengths and limitations. Furthermore, we propose an innovative neural network architecture for predicting multipath parameters, which significantly reduces the computational complexity typically associated with conventional methods. This model utilizes transformed channel state information for generating predictions, proving to be more effective in dynamic environments. Through this survey, we aim to provide a foundation for future research in intelligent channel prediction, highlighting the need for more sophisticated and adaptive algorithms to cope with the increasing complexity of wireless communication systems.