Emergent Communication Protocol Learning for Task Offloading in Industrial Internet of Things

TL;DR

The paper addresses joint task offloading and multichannel access in Industrial IoT MEC under strict deadlines. It introduces a MARL framework with emergent communication where base station and IIoT devices operate as cooperative agents within a Dec-POMDP augmented with signaling and are trained with MAPPO to maximize deadline-compliant task completion, using rewards $r_n(t)$ that incentivize timely execution. Empirical results show that the emergent communication protocol significantly improves channel access success, task completion within deadlines, and goodput compared with contention-based, contention-free, and no-communication baselines, while reducing signaling overhead. This approach offers scalable, coordinated edge computing solutions for dynamic IIoT workloads in industrial settings.

Abstract

In this paper, we leverage a multi-agent reinforcement learning (MARL) framework to jointly learn a computation offloading decision and multichannel access policy with corresponding signaling. Specifically, the base station and industrial Internet of Things mobile devices are reinforcement learning agents that need to cooperate to execute their computation tasks within a deadline constraint. We adopt an emergent communication protocol learning framework to solve this problem. The numerical results illustrate the effectiveness of emergent communication in improving the channel access success rate and the number of successfully computed tasks compared to contention-based, contention-free, and no-communication approaches. Moreover, the proposed task offloading policy outperforms remote and local computation baselines.

Figures (6)

• Figure 1: System Model.
• Figure 2: IIoT MDs and BS are cooperative MARL.
• Figure 3: Number of tasks successfully computed versus training episodes.
• Figure 4: Channel access success rate versus training episodes.
• Figure 5: Collision rate for different schemes.