Embodied intelligence-based hybrid edge computing networks for scalable task execution in Industrial IoT

Abstract

As Industry 5.0 advances, the increasing data volumes generated by end devices and the diverse and strict application requirements present significant challenges to traditional computing and communication infrastructures. There is a strong demand for more scalable and efficient computing and communication infrastructures. In this paper, we investigate a Hybrid Fog Computing Network (HFCN) architecture to enhance computing and data analytics capabilities in Industrial Internet of Things (IIoT) systems. In this architecture, Ad-hoc Fogs (A-Fogs) and Dedicated Fogs (D-Fogs) are formed to utilize the computing power of end devices and are integrated with cloud computing to create a seamless and scalable computing system. We propose a resource management framework and a novel Admission Control and Resource Allocation (ACRA) algorithm, which incorporates iterative optimization and Quality of Service (QoS)-awareness. The algorithm jointly considers computing and communication resources to maximize system utility while satisfying the QoS requirements of IIoT applications. The proposed ACRA algorithm is evaluated and compared with two baseline non-cooperative algorithms via a system-level simulator. Experimental results demonstrate the feasibility and scalability of large-scale task processing with HFCN. The cooperative ACRA algorithm achieves significant improvements in resource utilization, QoS, and processing capacity.