Energy-Saving System for Wireless Networks: A Time-Series-Based and Neural Network Approach
Gongcun Hong, Yan Lin, Pengju Luo, Ruoyu Wang · 2025
This paper proposes an intelligent energy-saving system that utilizes advanced time-series forecasting and neural network optimization to tackle the high energy consumption and increasing operational costs of 2G, 3G, and 4G wireless base stations. The system first employs an Autoregressive Integrated Moving Average (ARIMA) model to accurately forecast base station load trends. Based on these forecasts, a dynamic energy-saving strategy is developed, incorporating PRB threshold optimization, adaptive resource deactivation, and feedback-driven threshold adjustment. Additionally, a neural network-driven control mechanism is introduced to dynamically refine threshold settings, ensuring energy efficiency without sacrificing network performance. Experimental results from real-world deployments demonstrate that the proposed approach achieves nearly a 9% reduction in power consumption, significantly exceeding traditional energy-saving methods. The system's adaptive iteration further enhances energy efficiency over time, making it a robust solution for the sustainability of future wireless networks.