Intelligent collaborative computation offloading for dependent tasks in edge networks

Key Points

• The aim is to optimize computation offloading for dependent tasks in resource-constrained edge networks.
• Developed a distributed intelligent cooperative computation offloading method.
• Established a model considering edge node service limits and task dependencies.
• Applied graph attention multi-agent deep reinforcement learning for optimization.
• The proposed algorithm shows superior performance compared to existing methods across varied network parameters.
• Achieved maximum long-term average task processing utility under latency constraints.