A QoS Prediction Approach Based on Truncated Nuclear Norm Low-Rank Tensor Completion

Area of Science:

• Computer Science
• Mobile Computing
• Artificial Intelligence

Background:

• Mobile Edge Computing (MEC) enables numerous similar mobile services.
• Quality of Service (QoS) is crucial for evaluating service quality.
• Limited and unstable QoS data hinders accurate service assessment in MEC.

Purpose of the Study:

• To develop an effective method for predicting Quality of Service (QoS) values.
• To address the challenge of incomplete QoS data in mobile edge computing environments.
• To enhance the quality of mobile services through accurate QoS prediction.

Main Methods:

• Constructing tensors to represent complex multivariate QoS data.
• Applying truncated nuclear norm for low-rank tensor completion.
• Utilizing the Alternating Direction Multiplier Method (ADMM) for iterative optimization.

Main Results:

• The proposed method demonstrates superior QoS prediction accuracy compared to existing approaches.
• Effective mining of correlations within QoS data improves prediction performance.
• The general rate parameter allows control over tensor mode truncation.

Conclusions:

• The truncated nuclear norm low-rank tensor completion method is effective for QoS prediction in MEC.
• This approach offers a robust solution for handling missing QoS data.
• Accurate QoS prediction facilitates the provision of high-quality mobile services.