AI-Driven Dynamic Bandwidth Allocation in Fiber-to-the-Home (FTTH) Networks

Abstract

These new services that require enhanced capabilities like a high speed Internet like ultra-high-def video streaming, online gaming and cloud-based services have necessitated the exponential growth in the Internet demand of high-speed access to the Internet, which has made Fiber-to-the-Home (FTTH) networks a crucial infrastructure to any modern communication system. Dynamic Bandwidth Allocation (DBA) has been crucial in dynamic optimal utilization of resources, but the classical DBA techniques tend to fail on real time fluctuations of traffic and various Quality of service (QoS) needs. The paper introduces a distributed bandwidth allocation scheme that uses AI techniques and machine learning methods to adjust the bandwidth of FTTH-based networks to achieve the maximum bandwidth, lower delay, less loss of packets and more fairness in the network. The given strategy is to utilize the predictive analytics and machine-based decision-making to achieve better performance in the periods of low and high network loads. Results of the experiment show that the AI-based DBA is much better than both traditional DBA and Fixed Bandwidth Allocation (FBA) algorithms in several important measures. In emulated situations, the AI-based model realized maximum throughput of 25 percent, less mean latency of 35 percent, packet loss less than 50 percent, and fairness index values that were nearer to the ideal mark. These results provide a vivid image of the potential of artificial intelligence when it comes to changing FTTH resource administration, as well as an opportunity to create more versatile, successful, and user-friendly optical access networks. Not only the study demonstrates the validity of the application of AI to optimization bandwidth but also it presents a scalable solution to the future changes and developments of a network, such as 5G and beyond.