MLGO: A machine learning-based mountain gazelle optimization algorithm for efficient resource management and load balancing in fiber wireless access networks

Addressing the challenges posed by escalating data demands, connected devices, and bandwidth-hungry technologies, Fiber Wireless (FiWi) networks offer a holistic solution. Nevertheless, among the most significant challenges faced by FiWi networks are the component deployment, its connectivity, and performance under varying loads. In the FiWi network the ONUs plays a pivotal role in collecting and forwarding wireless-generated traffic thereby emphasizing the need for efficient resource management to ensure network reliability. Overloading of ONU often results in network congestion, as ONU serves as an intermediatory node between optical backend and wireless frontend. Thus, traffic offloading is a great solution by identifying underloaded ONUs and redirecting a portion of the excess traffic from overloaded ONUs to underloaded ONUs to maintain balanced resource allocation. However, identification of wireless routers within the wireless frontend is a crucial step, enabling strategic decision-making in rerouting traffic and promoting load balancing. Thus, the proposed work suggests a new hybrid two step method termed as MLGO (Machine learning based Mountain Gazelle Optimization Algorithm) which first uses machine learning based k-means clustering algorithm for nodes (Wireless routers and ONU) placement and connectivity and the second step employs the Mountain Gazelle Optimization algorithm (MGO) and GA for identifying optimum wireless routers for traffic offloading which enhances the overall FiWi network performance. The paper contributes to the evolution of FiWi networks, ensuring optimal connectivity, efficient resource utilization, and enhanced packet delivery ratio for end-users. Simulation results validate the effectiveness of this two-step proposed approach.