Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Multi-Agent RL-Based Industrial AIGC Service Offloading over Wireless Edge Networks

Siyuan Li, Xi Lin, Hansong Xu, Kun Hua, Xiaomin Jin, Gaolei Li, Jianhua Li

TL;DR

This work addresses the challenge of deploying generative AIGC at the network edge for industrial IoT by introducing GMEL, a diffusion-model–driven edge learning framework, and AMARL, an attention-enhanced multi-agent reinforcement learning algorithm for offloading AIGC tasks. The approach jointly optimizes computation and communication resources for heterogeneous AIGC workloads, via a centralized training–decentralized execution MARL paradigm with a multi-head attention critic to handle large joint state spaces. The authors demonstrate that AMARL achieves faster convergence and higher task completion rates than baseline DRL methods, reducing overall system latency and improving scalability as the number of IEDs grows. These results suggest that diffusion-based AIGC at the network edge, coupled with attention-driven MARL, can enable practical, privacy-preserving, low-latency industrial AIGC services with few-shot learning capabilities.

Abstract

Currently, the generative model has garnered considerable attention due to its application in addressing the challenge of scarcity of abnormal samples in the industrial Internet of Things (IoT). However, challenges persist regarding the edge deployment of generative models and the optimization of joint edge AI-generated content (AIGC) tasks. In this paper, we focus on the edge optimization of AIGC task execution and propose GMEL, a generative model-driven industrial AIGC collaborative edge learning framework. This framework aims to facilitate efficient few-shot learning by leveraging realistic sample synthesis and edge-based optimization capabilities. First, a multi-task AIGC computational offloading model is presented to ensure the efficient execution of heterogeneous AIGC tasks on edge servers. Then, we propose an attention-enhanced multi-agent reinforcement learning (AMARL) algorithm aimed at refining offloading policies within the IoT system, thereby supporting generative model-driven edge learning. Finally, our experimental results demonstrate the effectiveness of the proposed algorithm in optimizing the total system latency of the edge-based AIGC task completion.

Multi-Agent RL-Based Industrial AIGC Service Offloading over Wireless Edge Networks

TL;DR

This work addresses the challenge of deploying generative AIGC at the network edge for industrial IoT by introducing GMEL, a diffusion-model–driven edge learning framework, and AMARL, an attention-enhanced multi-agent reinforcement learning algorithm for offloading AIGC tasks. The approach jointly optimizes computation and communication resources for heterogeneous AIGC workloads, via a centralized training–decentralized execution MARL paradigm with a multi-head attention critic to handle large joint state spaces. The authors demonstrate that AMARL achieves faster convergence and higher task completion rates than baseline DRL methods, reducing overall system latency and improving scalability as the number of IEDs grows. These results suggest that diffusion-based AIGC at the network edge, coupled with attention-driven MARL, can enable practical, privacy-preserving, low-latency industrial AIGC services with few-shot learning capabilities.

Abstract

Currently, the generative model has garnered considerable attention due to its application in addressing the challenge of scarcity of abnormal samples in the industrial Internet of Things (IoT). However, challenges persist regarding the edge deployment of generative models and the optimization of joint edge AI-generated content (AIGC) tasks. In this paper, we focus on the edge optimization of AIGC task execution and propose GMEL, a generative model-driven industrial AIGC collaborative edge learning framework. This framework aims to facilitate efficient few-shot learning by leveraging realistic sample synthesis and edge-based optimization capabilities. First, a multi-task AIGC computational offloading model is presented to ensure the efficient execution of heterogeneous AIGC tasks on edge servers. Then, we propose an attention-enhanced multi-agent reinforcement learning (AMARL) algorithm aimed at refining offloading policies within the IoT system, thereby supporting generative model-driven edge learning. Finally, our experimental results demonstrate the effectiveness of the proposed algorithm in optimizing the total system latency of the edge-based AIGC task completion.
Paper Structure (17 sections, 9 equations, 6 figures)

This paper contains 17 sections, 9 equations, 6 figures.

Table of Contents

  1. Introduction
  2. Related Works
  3. AIGC Tasks Edge Computing in Industrial IoT
  4. Edge-based Industrial AIGC Tasks for Simulation
  5. Computing Latency in IoT
  6. Communication Latency in IoT
  7. Problem Formulation
  8. Attention-based MARL for AIGC Computing Tasks Offloading
  9. Formulating the Multi-agent Model
  10. State
  11. Action
  12. Reward
  13. AMARL-based computing offloading algorithm
  14. Experiment
  15. Experimental Setups
  16. ...and 2 more sections

Figures (6)

  • Figure 1: Overview of the proposed generative model-driven industrial AIGC collaborative edge learning framework.
  • Figure 2: Structure of the AIGC-assisted industrial offloading decision and simulation.
  • Figure 3: Convergence and cumulative rewards of different algorithms vs. episodes
  • Figure 4: Task completion rate vs. task generation probability
  • Figure 5: Task completion rate vs. maximum latency before deadline
  • ...and 1 more figures