Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Cloud-Fog-Edge Collaborative Computing for Sequential MIoT Workflow: A Two-Tier DDPG-Based Scheduling Framework

Yuhao Fu, Yinghao Zhang, Yalin Liu, Bishenghui Tao, Junhong Ruan

TL;DR

The paper tackles the NP-hard problem of scheduling sequential MIoT workflows across a three-layer cloud-fog-edge infrastructure with the objective of minimizing the end-to-end makespan $T_{\text{makespan}}$. It introduces a Two-tier DDPG scheduler that decomposes decisions into a global layer-selection policy and local per-layer node-allocation policies, with actions represented as $a_t=(\ell,n)$ and a shared reward guiding learning. The authors formalize the problem as an MDP with detailed state, action, transition, and reward definitions, and implement a coordinated global-local actor-critic architecture to learn effective policies. Empirical results from a Python simulator demonstrate that the proposed scheduler outperforms baselines such as HEFT, Greedy, FCFS, and Random, with performance advantages growing as workflow complexity increases, and the gap to HEFT narrowing for larger problems, indicating strong adaptability and scalability for large-scale MIoT deployments.

Abstract

The Medical Internet of Things (MIoT) demands stringent end-to-end latency guarantees for sequential healthcare workflows deployed over heterogeneous cloud-fog-edge infrastructures. Scheduling these sequential workflows to minimize makespan is an NP-hard problem. To tackle this challenge, we propose a Two-tier DDPG-based scheduling framework that decomposes the scheduling decision into a hierarchical process: a global controller performs layer selection (edge, fog, or cloud), while specialized local controllers handle node assignment within the chosen layer. The primary optimization objective is the minimization of the workflow makespan. Experiments results validate our approach, demonstrating increasingly superior performance over baselines as workflow complexity rises. This trend highlights the frameworks ability to learn effective long-term strategies, which is critical for complex, large-scale MIoT scheduling scenarios.

Cloud-Fog-Edge Collaborative Computing for Sequential MIoT Workflow: A Two-Tier DDPG-Based Scheduling Framework

TL;DR

The paper tackles the NP-hard problem of scheduling sequential MIoT workflows across a three-layer cloud-fog-edge infrastructure with the objective of minimizing the end-to-end makespan . It introduces a Two-tier DDPG scheduler that decomposes decisions into a global layer-selection policy and local per-layer node-allocation policies, with actions represented as and a shared reward guiding learning. The authors formalize the problem as an MDP with detailed state, action, transition, and reward definitions, and implement a coordinated global-local actor-critic architecture to learn effective policies. Empirical results from a Python simulator demonstrate that the proposed scheduler outperforms baselines such as HEFT, Greedy, FCFS, and Random, with performance advantages growing as workflow complexity increases, and the gap to HEFT narrowing for larger problems, indicating strong adaptability and scalability for large-scale MIoT deployments.

Abstract

The Medical Internet of Things (MIoT) demands stringent end-to-end latency guarantees for sequential healthcare workflows deployed over heterogeneous cloud-fog-edge infrastructures. Scheduling these sequential workflows to minimize makespan is an NP-hard problem. To tackle this challenge, we propose a Two-tier DDPG-based scheduling framework that decomposes the scheduling decision into a hierarchical process: a global controller performs layer selection (edge, fog, or cloud), while specialized local controllers handle node assignment within the chosen layer. The primary optimization objective is the minimization of the workflow makespan. Experiments results validate our approach, demonstrating increasingly superior performance over baselines as workflow complexity rises. This trend highlights the frameworks ability to learn effective long-term strategies, which is critical for complex, large-scale MIoT scheduling scenarios.
Paper Structure (26 sections, 12 equations, 3 figures, 2 tables, 1 algorithm)

This paper contains 26 sections, 12 equations, 3 figures, 2 tables, 1 algorithm.

Table of Contents

  1. Introduction
  2. System Design
  3. System Framework
  4. Medical Computing Task Workflows
  5. Cloud-Fog-Edge Collaborative Computing Framework
  6. The Two-tier DDPG Scheduling Framework
  7. Problem Formulation of Scheduling Strategy
  8. Preliminary modeling
  9. Optimization problem.
  10. Two-tier DDPG-based Scheduling
  11. Reinforcement Learning Formulation
  12. MDP Modeling
  13. State Space $\mathcal{S}$
  14. Action Space $\mathcal{A}$
  15. State Transitions $\mathcal{P}$
  16. ...and 11 more sections

Figures (3)

  • Figure 1: The two-tier DDPG-based scheduling framework for processing sequential MIoT task workflows in cloud-fog-edge infrastructures.
  • Figure 2: Training performance of the Two-tier DDPG scheduler over 700 episodes.
  • Figure 3: Total makespan comparison across four difficulty levels (L1-L4).