Comparative Analysis of Major Machine-Learning-Based Path Loss Models for Enclosed Indoor Channels

Area of Science:

• Wireless Communications
• Machine Learning
• Signal Propagation

Background:

• Fifth-generation (5G) wireless communication necessitates utilizing super-high frequency (SHF) and millimeter-wave (mmWave) bands for high data rates.
• Accurate path loss prediction models are crucial for planning and optimizing future wireless networks operating in these sensitive frequency bands.

Purpose of the Study:

• To evaluate and compare the performance of various machine learning methods for path loss prediction in mmWave frequency bands.
• To identify the most accurate and stable machine learning-based models for wireless communication network planning.

Main Methods:

• The study evaluated multiple linear regression (MLR), polynomial regression (PR), support vector regression (SVR), decision trees (DT), random forests (RF), K-nearest neighbors (KNN), artificial neural networks (ANN), and artificial recurrent neural networks (RNN), specifically long short-term memory (LSTM).
• Model performance was assessed using measurement data, focusing on root-mean-square error (RMSE), R-squared, and correlation values.

Main Results:

• Artificial neural networks (ANN) and RNN-LSTM demonstrated the best performance with the lowest RMSE values, ranging from 0.0216 to 2.9008 dB.
• Most evaluated machine learning models, excluding MLR, showed excellent fitting capabilities for indoor wireless communication data, with R-squared and correlation values exceeding 0.91 and 0.96, respectively.

Conclusions:

• Machine learning techniques, particularly ANN and RNN-LSTM, are highly effective and stable for predicting path loss in the mmWave frequency regime.
• These models offer accurate predictions essential for the development and optimization of future wireless communication systems.