MS3F: Maximizing Bandwidth and Data Transmission Performance in SUCCESS-HPON Architecture for WDM-PON Using AI Techniques

Recent advances in urban and access communication networks have enabled the use of various technologies, with optical transmission mediums and Wavelength Division Multiplexing (WDM) being the most prominent. The SUCCESS-HPON architecture has a WDMPON (Passive Optical Network) component, which necessitates the development of scheduling algorithms. Sequential scheduling with schedule time framing (S3F) algorithms are designed to facilitate efficient, bidirectional connections between the Optical Network Units (ONUs) and Central Office and make sure that the upstream and downstream traffic is organized and sequential. In S3F, Virtual Output Queues (VOQs) are employed as a granting program that consists of timeslots for both upstream and downstream traffic. This helps to reduce the overhead associated with framing and guard bands, while also preserving memory space between traffic flows. The S3F algorithm outperforms the sequentially scheduling algorithm by providing higher throughput, lower delay, and traffic fairness between upstream and downstream traffic. The primary objective of this study is to improve the bandwidth utilization and data transmission performance, the Modified Sequential scheduling with schedule time framing (MS3F) algorithm is enhanced with Reinforcement Learning (RL) and Deep Neural Network (DNN) techniques. Reinforcement learning algorithms can be used to optimize the scheduling decision-making process and dynamically adjust the allocation of resources to improve network efficiency. Deep neural networks can be used to analyze network behavior and predict potential issues, such as network congestion, to improve network reliability. These Artificial Intelligence techniques can help to enhance the performance of the MS3F algorithm by providing more intelligent and efficient scheduling of the WDM-PON traffic.