A Comprehensive Study on Path Loss Estimation Using Deep Hybrid Learning in 5G Networks

One of the most important factors in radio network design is path loss - a phenomenon that may be measured using a variety of techniques, including deterministic, empirical, machine learning, and deep learning models. Each approach has its own limitations, such as inability to capture non-linear interactions, high computational resource demand, and inability to reflect changes in environmental conditions, among many others. The deep learning model has the capacity to recognize intricate patterns and has been essential in removing those obstacles; therefore, in this study it is used for path loss prediction in 5G communications in the South Asian region. The model makes use of long- and short-term memory (LSTM), gated recurrent unit (GRU), convolutional neural network (CNN), and dense neural network (DNN) approaches to take advantage of all the benefits that each algorithm provides. The performance of the proposed strategy was validated by testing it against multiple state-of-the-art approaches, while relying on the same dataset. An examination of the relevance of characteristics has also been carried out to gain a better understanding of the influence of path loss. A variety of characteristics that are directly related to path loss were evaluated, followed by an examination of how they affect the decision-making process. The results show a possible solution that can help handle this path loss estimation for mmWave communication, especially for 5G networks and beyond.