Rain Attenuation Prediction Modeling for Microwave and Millimeter Wave Band Using LSTM

This study introduces a Long-Short Term Memory (LSTM) network prediction model that can accurately forecast rain attenuation in tropical regions, where weather-related interference is substantial because of long periods of wet seasons. The International Telecommunication Union-Radiocommunication (ITU-R) rain attenuation prediction models and Crane model were used to acquire historical attenuation data in dB, which reveals rain attenuation up to 31.61 dB and 28.65 dB, respectively. This data was obtained by utilizing local rainfall data and an 11GHz microwave link profile from Addis Ababa, Ethiopia. The proposed model uses this data as input and achieves great accuracy in predicting rain-induced signal degradation, with an MSE of 3.3644 * 10(-4) and RMSE of 0.018. This indicates the model's usefulness in anticipating rain-induced signal degradation, which is critical for frequency bands above 7 GHz.