Multi strategy fusion enhanced channel estimation algorithm based on deep learning

The increasing frequency of maritime activities has fueled a growing demand for advanced wireless communication systems, making accurate channel estimation a crucial technology. Traditional channel estimation algorithms often face limitations when dealing with noise factors. To address this issue, we propose an enhanced channel estimation algorithm based on deep learning, which integrates multiple strategies and is named the IMBP algorithm. This method simulates the insertion of pilot signals at the receiving end and combines the efficiency of mean filter. Additionally, it utilizes random forests to optimize end-to-end information transmission and adjusts strategies through dynamic thresholds. Simultaneously, by incorporating the powerful feature learning capability of deep learning in channel estimation, it upgrades traditional linear mapping to nonlinear mapping. The simulation results demonstrate that the IMBP algorithm proposed in this paper significantly reduces BER in communication, demonstrating superior performance.