Cloud Manufacturing is a service-oriented, high efficiency and low consumption, knowledge-based new mode of networked manufacturing. ‡ It aims at facilitating manufacturers to minimize their product life-cycle expenses, maximize production efficiency, and provide agile accommodation of available manufacturing assets to the variable demand of customers. ‡ Cloud Manufacturing assumes that manufacturers use crowdsourcing and outsourcing models for manufacturing operations and supporting services ‡‡. Cloud Manufacturing principles allow flexible scaling of manufacturing assets ‡ and instant access to the most efficient, innovative business technology solutions on a pay-as-you-go basis ‡. Cloud Manufacturing applies well-known basic concepts from the field of Cloud Computing and the Internet of Things to the manufacturing domain. ‡
The purpose of Cloud Manufacturing is to move from production-oriented manufacturing processes to service-oriented manufacturing process networks by modelling single manufacturing assets as services in a similar vein as Software-as-a-Service or Platform-as-a-Service solutions are already provided by Cloud providers. By mapping well-known concepts from the field of Cloud computing to real-world manufacturing processes, Cloud Manufacturing takes service-oriented manufacturing processes to the next level and supports these processes by cloud-based software and IT infrastructure. ‡ To achieve the formulated purpose, the following principles are applied:
Cloud Manufacturing Scenario. A manufacturing process is a set of process steps to be executed in order to create a certain valued product. In reality, these process steps are single services representing specific manufacturing assets and activities on the shop floor. The means that are offered by Cloud Manufacturing are intended to integrate single services of the manufacturing processes from distributed locations as if the complete manufacturing was carried out on the same shop floor (see Fig. 1). To achieve this, manufacturers need to virtualise their manufacturing assets into single software services‡. An integration of those services is possible via a Cloud Manufacturing platform, where these services are presented, advertised, leased, and sold as a part of manufacturing processes maintained in the platform.
As an example manufacturing process, we consider the manufacturing of a generic product, which is assembled by a Manufacturing Company in four steps: (1) some needed composite parts are produced by Supplier A, (2) some auxiliary parts are produced by Supplier B, (3) the product is assembled on the shop floor of the Manufacturing Company's own plant, and (4) the finished product is verified according to quality aspects. Each of these single steps of the considered manufacturing process can be encapsulated as a service. The process model of this manufacturing process is created by means of Business Process Model Notation (BPMN) by adding services as process steps into the model. When a process model is foreseen for enactment, a process instance is created. When a process task (i.e., step) must be enacted, a corresponding service is deployed onto appropriate Cloud-based computational resources and executed accordingly.
The challenge here is that a large amount of interdependent processes with different Quality of Service (QoS) and Service Level Agreements (SLA) aspects may be requested at any point of time. Therefore, elastic processes are a promising approach in Cloud Manufacturing ‡.
Figure 1: High-level Cloud Manufacturing Scenario
Manufacturing Assets. Manufacturing assets consist of diversified and distributed manufacturing resources (equipment, materials, software, knowledge, and skills) and manufacturing capabilities (design, production, experimentation, management, and communication). Manufacturing assets are assigned to the users on demand ‡. For this, three principal actions to be applied:
These actions embrace a provision of a service description (using ontologies and semantic descriptions) including a clear identifier of the service, the definition of related data sources (for monitoring purposes), and the upload of the software service to the Marketplace of the Cloud Manufacturing platform, which is a central place where users are able to market their manufacturing assets and services in a Cloud-based way. Management of the Cloud services (dynamic location, monitoring, allocation, and reconfiguration) has to be provided and a pool of logical resources which are then used by the users has to be established in the Cloud Manufacturing platform by means of interfaces.
Manufacturing Asset Virtualisation. A key characteristic of virtualisation is a quality of a manufacturing service and the efficiency of service encapsulation. First, an identification of manufacturing resources has to be performed. Manufacturing resource information then has to be virtualised and monitored in real-time ‡. The main problems to be addressed by a Cloud Manufacturing platform are the heterogeneity of Cloud Manufacturing assets: multi-domain, multi-level and multi-granularity ‡; manifoldness of QoS information; and selection of sharing strategies for manufacturing assets ‡. These problems are solved by application of service virtualisation methods and tools that combine manufacturing resource virtualisation and manufacturing capability servitization. The process of servitization implies incorporation by manufacturers of the services into their product proposals or even substitution of the products by services‡. Comparing to virtualisation in Cloud computing, Cloud Manufacturing virtualisation addresses a problem of establishing a comprehensive data model for representing manufacturing assets. Notably the meaning of manufacturing services differs from the one of traditional Web services, since Web service standards have no means to represent heterogeneous characteristics of manufacturing resources ‡.
CREMA is a Cloud Manufacturing platform that aims to establish manufacturing collaboration and facilitate stakeholfer interaction, to provide means to virtualise manufacturing services and create manufacturing processes, and to enact processes effectively by performing design time and runtime optimisation.
Recently, the world has seen emerging CPS modeling frameworks addressing various design aspects such as control, security, verification and validation. However, there have been no considerations for reliability and automated debug aspects of verification. The main aim is to fill this gap by introducing reliable design and automated system debug into CPS modeling. To reach this aim, the project will develop a cross-layer CPS model spanning device (analogue and digital), circuit, network architecture, firmware and software layers. In addition, a holistic fault model for fundamentally different error sources in CPSs (design bugs, wear-out and environmental effects) in a uniform manner will be proposed. Moreover, IMMORTAL plans to develop fault management infrastructure on top of the reliable design framework that would allow ultrafast fault detection, isolation and recovery in the emerging many-core based CPS networked architectures that are expected to be increasingly adopted in the coming years.
As a result, the project will enable development of dependable CPSs with improved reliability and extended effective lifetime, aging and process variations. In line with the expected impacts of the Call, the project will have a significant impact in development time as well as maintenance costs of dependable cyber-physical systems.The tool framework to be developed will be evaluated on a clearly specified real-world use-case of a satellite on-board computer. However, since the results are more general and applicable to many application domains, including avionics, automotive and telecommunication, demonstration of the framework tools will be applied to CPS examples from other domains as well.