DRL-Driven Intelligent SFC Deployment in MEC Workload for Dynamic IoT Networks

Area of Science:

• Computer Science
• Network Engineering
• Artificial Intelligence

Background:

• The proliferation of Internet of Things (IoT) sensor networks generates vast amounts of data, necessitating efficient resource management.
• Maintaining Quality of Service (QoS) in heterogeneous IoT networks is challenged by limited multi-access edge computing (MEC) resources and increasing task offloading demands.
• Network congestion, service latency, and inefficient resource utilization degrade IoT-MEC system performance.

Purpose of the Study:

• To propose an intelligent task offloading and resource orchestration framework for dynamic IoT-MEC environments.
• To optimize energy consumption, computational cost, network congestion, and service latency.
• To enhance overall system efficiency and achieve an optimal policy in edge computing.

Main Methods:

• Developed a framework integrating task offloading and dynamic resource orchestration.
• Employed Service Function Chaining (SFC) for Virtual Network Functions (VNFs) placement and routing path determination.
• Utilized Deep Reinforcement Learning (DRL), specifically Deep Q-network (DQN), for adaptive resource allocation and task offloading decisions.

Main Results:

• The DRL-based scheme significantly outperformed reference methods.
• Demonstrated substantial reductions in service latency and energy consumption.
• Showcased improvements in delivery, throughput, and cumulative reward.

Conclusions:

• The proposed intelligent framework effectively addresses challenges in IoT-MEC environments.
• DRL-driven dynamic resource orchestration and task offloading optimize system performance.
• The approach offers a flexible and adaptive solution for managing resources in edge computing.