Estimasi Sudut Kedatangan yang Ditingkatkan dengan CNN pada Array Antena MIMO Menggunakan Data Sinyal IoT Dunia Nyata

Abstract

− This study proposes the application of a Convolutional Neural Network (CNN)–based approach to analyze signals in Internet of Things (IoT)–based MIMO antenna systems, with the aim of enhancing the understanding of system performance characteristics, particularly in predicting latency parameters. The CNN model is trained using real-world IoT signal data that have undergone comprehensive preprocessing stages, including data normalization, missing value handling, and feature engineering to ensure compatibility with the model input format. Experimental results on previously unseen test data demonstrate that the proposed model achieves a test loss of 1.4410, represented by the Mean Squared Error (MSE), and a Mean Absolute Error (MAE) of 0.9395. These results indicate that the model attains a relatively low prediction error and effectively captures the nonlinear relationships between signal features and system responses.

Visualization of the testing results reveals a strong correlation between actual and predicted latency values, although some dispersion remains due to channel complexity and the inherent variability of IoT signals. The distribution of prediction errors is centered around zero, indicating the absence of significant systematic bias in the model. Overall, the findings confirm the potential of CNN as a reliable approach for modeling and performance analysis of IoT-based MIMO antenna systems, while also highlighting opportunities for further development in spatial parameter estimation and intelligent wireless communication system optimization.