RTWP-interference forecasting in RF signals using LSTM method

The rapid advancement of communication technologies today has amplified the societal and economic impact of wireless communication systems. The primary goal of GSM (Global System for Mobile Communications) operators is to provide uninterrupted communication to users. However, with the rapid growth of mobile communication infrastructure, fluctuations in signal strength and signal quality have increased their impact on mobile users. Since RTWP(Received Total Wideband Power) value is an important indicator of network performance, it is known that high RTWP values cause a decrease in communication quality. Predicting RTWP values in the uplink direction aims to facilitate the implementation of necessary measures to ensure uninterrupted communication between the base station and mobile terminals, thus reducing interference effects in the communication network for more reliable communication. This study aims to address these issues by focusing on predicting RTWP (Received Total Wideband Power) values in the uplink direction of UMTS (Universal Mobile Telecommunications System) technology-based base stations. The Long Short-Term Memory (LSTM) method, a type of Recurrent Neural Network (RNN), is employed for this prediction process. it used to model time series forecasting problems. We can use RTWP predictions to minimize network problems. These predictions greatly help in making decisions about network optimization. The training data comprises RTWP values measured in the uplink direction of UMTS technology-based base station in Silopi-Şırnak , which are then used for training the LSTM-based deep neural network. By experimenting with different hyperparameter values and layer configurations of the LSTM-based model, the study investigates the conditions under which the best performance is achieved. According to the test results, the lowest Root Mean Square Error (RMSE) value is obtained batch size 64, epochs 100 in the two-layer and one-layer LSTM models. These results underscore the effectiveness of the predictive model developed to enhance communication quality.