AI-Driven Channel Estimation Using Geo-spatial Data With Modified DMRS

Accurate channel estimation is very important in wireless communication systems and particularly with the increasing demand for enhanced connectivity in diversified environment (multiple). Traditional techniques that often rely on complex mathematical models, which struggles to balance for dynamic and noisy conditions, especially in multicarrier signal transmission/propagation. This paper will explore the application of a fully connected deep neural network (DNN) for better channel estimation across various scenarios, include urban areas like Tokyo and Bengaluru, as well as rural and semi-urban regions like Bastar and Patiala. MATLAB 2024b as the primary simulation tool, the proposed AI-based approach eliminates the need of complex and twisted channel modeling and adapt to real-world like conditions. Once the model is trained on specific scenario and environmental parameters, the DNN model will then demonstrates superior performance and accuracy compared to the traditional methods, even in challenging different propagation environments. By leveraging data-driven and insights, the AI model can achieve lower mean squared error (MSE) notably, which showcase its hidden potential to address the limitations of traditional techniques. This study will highlight efficiency and comparison of AI-driven channel estimation with traditional method in ensuring better communication across different environment scenarios. Our findings will later be useful in easily integrating it into 5G and beyond, due to its effectiveness.