Optimizing Radiation Patterns in Electrical Communication Networks Using LSTM Model

In this paper, we present a model to optimize electrical communication network radiation patterns using a Long Short-Term Memory (LSTM) neural network. The objective is to increase the signal-to-noise ratio (SNR) and received power for users. We improve the selection between omnidirectional and directional radiation patterns of access points. In the simulation, we set the environment with user distributions and obstacles. The LSTM model achieved a training RMSE of 8.62 dB and a test RMSE of 25.82 dB for the omnidirectional radiation pattern, and a training RMSE of 7.3 dB and a test RMSE of 28.37 dB for the directional radiation pattern. We integrated LSTM predictions into a proposed decision-making method. We discovered that the omnidirectional radiation pattern is often appropriate in environments with uniformly distributed users, while the directional radiation pattern is suitable for environments with a small number of users. This study demonstrates the potential of machine learning for optimizing radiation patterns to improve electrical communication networks.