A Data-Driven ONU Assignment Strategy for Optimizing GPON Resource Management

With the explosive growth in high-bandwidth application demands, existing GPON networks face challenges in balancing service quality, user capacity, and operational costs. This study presents a Data-Driven ONU Assignment Strategy (DOAS) aimed at improving resource utilization of PON ports while ensuring service quality and reducing operational costs. DOAS adopts a modular four-layer architecture that dynamically adjusts resource allocation based on traffic data. Analysis of real-world operational data shows that using the seven-day peak hour traffic as an indicator, the DOAS strategy ensures service quality while increasing the number of PON ports exceeding traditional capacity limits by $124 \%$, enhancing system assignment efficiency by $90 \%$, and significantly reducing weekly reassignment operations due to bandwidth insufficiency by $75.6 \%$. DOAS not only breaks through traditional static allocation limitations, providing a data-driven decision framework, but also effectively reduces operational costs, offering a practical solution for telecom operators. Future work will integrate traffic prediction models, extend the strategy to next-generation PON technologies, and develop differentiated service strategies to optimize network resources.