Resource requirements of an edge-based digital twin service:an experimental study  

作　　者：Federico MUNGARI Milan GROSHEV Carla Fabiana CHIASSERINI 

机构地区：[1]Department of Electronics and Telecommunications,Politecnico di Torino,Turin 10129,Italy [2]Departamento de Ingenieria Telematica,Universidad Carlos III de Madrid,Madrid 28911,Spain

出　　处：《Virtual Reality & Intelligent Hardware》2022年第6期506-520,共15页虚拟现实与智能硬件（中英文）

基　　金：Supported by the EU through the 5Growth project (856709);the UNICO 5G I+D 6G-EDGEDT-01/02;the Spanish Ministry of Economic Affairs and Digital Transformation;NPRP-S 13th Cycle (NPRP13S-0205-200265) from the Qatar National Research Fund (a member of Qatar Foundation)。

摘　　要：Digital Twin(DT)is a pivotal application under the industrial digital transformation envisaged by the fourth industrial revolution(Industry 4.0).DT defines intelligent and real-time faithful reflections of physical entities such as industrial robots,thus allowing their remote control.Relying on the latest advances in Information and Communication Technologies(ICT),namely Network Function Virtualization(NFV)and Edge-computing,DT can be deployed as an on-demand service in the factories close proximity and offered leveraging radio access technologies.However,with the purpose of achieving the well-known scalability,flexibility,availability and performance guarantees benefits foreseen by the latest ICT,it is steadily required to experimentally profile and assess DT as a Service(DTaaS)solutions.Moreover,the dependencies between the resources claimed by the service and the relative demand and work loads require to be investigated.In this work,an Edge-based Digital Twin solution for remote control of robotic arms is deployed in an experimental testbed where,in compliance with the NFV paradigm,the service has been segmented in virtual network functions.Our research has primarily the objective to evaluate the entanglement among overall service performance and VNFs resource requirements,and the number of robots consuming the service varies.Experimental profiles show the most critical DT features to be the inverse kinematics and trajectory computations.Moreover,the same analysis has been carried out as a function of the industrial processes,namely based on the commands imposed on the robots,and particularly of their ion-level,resulting in a novel trade-off between computing and time resources requirements and trajectory guarantees.The derived results provide crucial insights for the design of network service scaling and resource orchestration fra-meworks dealing with DTaaS applications.Finally,we empirically prove LTE shortage to accommodate the minimum DT latency requirements.

关 键 词：Digital twin EDGE NFV TESTBED Performance evaluation 

分 类 号：TP399[自动化与计算机技术—计算机应用技术]