SFC active reconfiguration based on user mobility and resource demand prediction in dynamic IoT-MEC networks

Area of Science:

• Computer Science
• Networking
• Mobile Computing

Background:

• Mobile Internet of Things (IoT) networks with Multi-access Edge Computing (MEC) rely on Service Function Chains (SFCs) for secure and scalable traffic delivery.
• User mobility and dynamic network traffic in IoT-MEC environments cause performance mismatches between SFCs and allocated resources.
• Proactively reconfiguring SFCs to adapt to network changes is a significant challenge.

Purpose of the Study:

• To develop a Service Function Chain Reconfiguration Strategy (SFC-RS) for IoT-MEC networks.
• To minimize end-to-end delay and reconfiguration costs by predicting user mobility and resource demands.
• To ensure high quality of service despite network state fluctuations.

Main Methods:

• Modeling SFC-RS as an Integer Linear Programming (ILP) problem.
• Developing a user trajectory prediction model using an attention-based codec movement approach.
• Designing a VNF resource demand prediction model utilizing a Long Short-Term Memory (LSTM) network.
• Proposing the Prediction-based SFV Active Reconfiguration (PSAR) algorithm for seamless SFC migration and routing updates.

Main Results:

• The PSAR algorithm significantly reduces end-to-end delay compared to existing algorithms (TSRFCM, DDQ, OSA, DPSM).
• PSAR demonstrates substantial performance improvements in reducing reconfiguration costs.
• Simulation results validate the effectiveness of PSAR in maintaining network consistency and service quality.

Conclusions:

• The proposed PSAR algorithm effectively addresses the challenges of SFC reconfiguration in dynamic IoT-MEC networks.
• Predictive modeling of user mobility and resource demand is crucial for optimizing SFC performance.
• PSAR offers a robust solution for ensuring high-quality service delivery in mobile edge computing environments.