The Reinforcement Learning (RL) Based SFC Request Scheduling in Computer Networks

Abstract

This study investigates the use of Reinforcement Learning (RL) to minimize the latency between the source and destination of SFC requests in Neural Networks. This problem gains relevancy owing to Service Function Chaining (SFC) becoming a fundamental concept in modern computer networks to efficiently route and process network traffic through a sequence of specialized network functions. The approach utilizes Deep-Q-Network (DQN) reinforcement learning to determine the shortest path between two nodes using the Greedy-Simulated Annealing (GSA) Dijkstra's Algorithm. The containers within the SFC chain help train the model based on bandwidth restrictions (fiber networks) to optimize the different pathways in terms of action space. Through rigorous evaluation of varying action spaces in models, we assessed the predictive accuracy of each model based on reward-request relationships graphed. A scheduling agent can then manipulate this algorithm to handle maximum scheduling pathways. The findings offer practical implications for reinforcement learning (RL) that can be applied to request scheduling in computer networks to optimize resource allocation, improve the quality of service, and enhance network performance.