3D Urban Wireless Network Planning and Analysis Using Ray-Tracing and Machine Learning

ReliableWi-Fi coverage is essential for the functioning of IoT devices in smart cities. This study utilizes advanced Ray-Tracing techniques and Machine Learning to enhance Wi-Fi signal coverage across a university campus. We analyze a comprehensive dataset comprising measurements from three Wi-Fi frequencies (2.4 GHz, 5 GHz, and 6 GHz) and five ray-tracing configurations (0/0, 0/1, 1/0, 1/1, and 2/1), along with two power levels (100mW and 200mW) and a 5-meter resolution grid featuring 3,135 receivers, resulting in a total of 94,050 rows and 28 columns, amounting to 2,633,400 data points. The Random Forest Regressor Model demonstrates strong predictive capabilities, as evidenced by low error metrics (Mean Absolute Error of 1.44 dBm, Mean Squared Error of 6.46, and R-squared of 0.99), validating its effectiveness for accurate signal strength estimations. Furthermore, feature importance analysis highlights the significance of link status and signal coverage in prediction accuracy, underscoring the necessity for effective feature selection. This research provides valuable insights for optimizing Wi-Fi network planning in urban environments, establishing a solid foundation for future investigations.