Knowledge-Assisted Actor Critic Proximal Policy Optimization-Based Service Function Chain Reconfiguration Algorithm...

Area of Science:

• Telecommunications Engineering
• Computer Networking
• Artificial Intelligence

Background:

• Future 6G networks will leverage Network Function Virtualization (NFV) for dynamic service delivery.
• Service Function Chains (SFCs) composed of Virtual Network Functions (VNFs) are crucial for meeting diverse service demands.
• Optimizing SFC reconfiguration is essential for the highly dynamic 6G Internet of Things (IoT) environment.

Purpose of the Study:

• To develop a flexible and efficient SFC reconfiguration algorithm for 6G networks.
• To enhance the performance of traditional Actor Critic (AC) algorithms in dynamic optimization problems.
• To ensure the Quality of Service (QoS) for end-to-end services in 6G IoT scenarios.

Main Methods:

• Utilized Proximal Policy Optimization (PPO) with a clip function to stabilize policy updates.
• Integrated PPO with AC, incorporating historical decision data as network knowledge for improved initial policies.
• Proposed the Knowledge-Assisted Actor Critic Proximal Policy Optimization (KA-ACPPO) algorithm for SFC reconfiguration.

Main Results:

• The KA-ACPPO algorithm demonstrated effective SFC reconfiguration in simulations.
• Achieved significant reductions in computing cost and power consumption.
• Ensured the Quality of Service (QoS) for end-to-end services.

Conclusions:

• The KA-ACPPO algorithm offers a robust solution for dynamic SFC reconfiguration in 6G networks.
• Deep learning-based approaches, like KA-ACPPO, are well-suited for optimizing complex network functions.
• The proposed method balances efficiency, cost, and service quality in next-generation networks.