A Parallelizable Task Offloading Model with Trajectory-Prediction for Mobile Edge Networks.

Area of Science:

• Edge Computing
• Mobile Networks
• Task Offloading

Background:

• Edge computing enhances server collaboration and resource utilization near users.
• Task offloading is crucial for edge network efficiency, but mobile device mobility poses challenges.
• Existing methods struggle with unpredictable user movements in mobile edge networks.

Purpose of the Study:

• To develop a trajectory prediction model for moving targets in edge networks without relying on historical user paths.
• To propose a mobility-aware, parallelizable task offloading strategy leveraging trajectory prediction.
• To evaluate the proposed model and strategy against existing methods using real-world data.

Main Methods:

• Developed a trajectory prediction model for mobile targets in edge computing environments.
• Designed a parallelizable task offloading strategy incorporating mobility awareness and trajectory prediction.
• Conducted experiments using the EUA dataset to compare prediction hit ratio, network bandwidth, and task execution efficiency.

Main Results:

• The trajectory prediction model achieved high accuracy, with an 80% hit rate for speeds under 12.96 m/s.
• The proposed mobility-aware strategy significantly outperformed random, non-position prediction, and non-parallel strategies.
• Network bandwidth utilization increased over eightfold with the parallel strategy compared to non-parallel methods.

Conclusions:

• The novel trajectory prediction model effectively addresses mobility challenges in edge networks.
• The mobility-aware parallelizable task offloading strategy enhances efficiency and network resource utilization.
• This approach offers a robust solution for optimizing task execution in dynamic mobile edge environments.