A Resource and Capability Virtualization Method for Cloud ...

4 downloads 7220 Views 517KB Size Report
A resource & capability virtualization method for cloud manufacturing systems. Ning Liu, Xiaoping Li, Qian Wang. School of Computer Science and Engineering.
A resource & capability virtualization method for cloud manufacturing systems Ning Liu, Xiaoping Li, Qian Wang School of Computer Science and Engineering Key Laboratory of Computer Network and Information Integration Southeast University Nanjing, China PR Abstract—Resources should be virtualized before they are deployed to cloud manufacturing systems. In this paper, a method is proposed for resource virtualization by transforming manufacturing resources into cloud services through two phases. Manufacturing resource features are comprehensively analyzed. A virtual specification is established for describing heterogeneous manufacturing resources in an isomorphic manner. By extracting characteristics of resources, an algorithm is proposed for resources partitioning according to manufacturing capabilities. Resources are encapsulated as cloud services and deployed to the cloud service platform, where manufacturing resources can be shared and accessed by heterogeneous applications in cloud manufacturing systems. Keywords- virtualization; cloud manufacturing; integration; sharing

I.

INTRODUCTION

Cloud Manufacturing is a new service-oriented networked manufacturing model [1]. By virtualized and deployed on the cloud service platform, heterogeneous manufacturing resources can be shared by different complex collaborative manufacturing demands. Although some achievements have been obtained in networked manufacturing platform[2], application prototype system[3], manufacturing resource modeling[4], resource scheduling[5], and collaborative product design[6], networked manufacturing resources are not as broadly shared and integrated as expected. The critical challenge lies in the agile and interoperable resource virtualization. Though STEP and its enhancements[7] are the standards of manufacturing products, they do not provide standard information models for the corresponding manufacturing resource virtualization. Vichare et.al[8] proposed a unified manufacturing resource model for not only providing element information required by the CNC machining system, but also supporting the process planning decision making. However, the unified manufacturing resource model only provides the static representation of the resources in the CNC machining system. It does not provide the functionality of these resources. Manufacturing enterprises in virtualized manufacturing systems[9] take use of external manufacturing resources[10,11] by integrating and sharing heterogeneous resources [12,13]. Most resource virtualization methods manage to virtualize resources by encapsulating the fundamental descriptions of resources into services. However, resources have the characteristic of diversity and complexity, web services cannot

978-1-4577-0653-0/11/$26.00 ©2011 IEEE

expose the nature of the resources fully. Semantic Web is an important approach to deal with semantic heterogeneity within manufacturing systems[14], it has been widely applied in resource virtualization to describe the semantic functionality of the manufacturing resources[15,16]. However, there are still many challenges on resource semantic annotation, description etc.. At present, existing resource virtualization methods can enhance resource sharing and promote resource interaction and management to some extent. There are still many unsolved problems. For example, there is no common understanding for resource representation since different manufacturing enterprises have different resource management strategies. No standardized resource virtualization description is available. The complexity of the resources cannot be represented explicitly. This paper presents a resource & capability virtualization method with the purpose of realizing large-scale manufacturing resources integration and sharing in cloud manufacturing systems. The method is divided into two phases: virtual description and service encapsulation. Virtual description provides a normalized virtual description specification for integrating heterogeneous manufacturing resources. Service encapsulation encapsulates manufacturing resources as cloud services to present a uniform interface to interoperate with the cloud service platform. The method provides a common understanding for heterogeneous resource information, representing the intrinsic nature of manufacturing resources explicitly. The rest of this paper is organized as follows: manufacturing resource features are analyzed in Section 2. Section 3 makes some preparations for the resource virtualization. In section 4 a virtual description specification is suggested and service encapsulation operation is discussed in detail. Section 5 introduces an example scenario to show the benefits of the proposed method. Future work and conclusions are offered in Section 6. II.

ANALYSIS OF RESOURCE FEATURES

Manufacturing resources play an important role for production development in cloud manufacturing systems. Attributes, states and capabilities of resources keep on changing throughout the whole resource life cycle. Resource features are closely related to resource description. The more features require the more complex description. Therefore, a tradeoff between resource description and selected features

1003

should be made in terms of two aspects: (1) Robust features should be selected and described. (2)Information for resource management and utilization should meet the fundamental requirements. There are three categories of manufacturing resources: equipment, implements and auxiliaries. The most important viewpoint of existing manufacturing resource models lies in the representation of manufacturing resource capabilities [17]. In this paper, resource features are classified into non-functional and functional features. In terms of resource life cycles, business scopes, and collaborative manufacturing modes, the structure of resource features can be depicted as Fig.1. Manufacturing resources are geologically decentralized located. By deployed to cloud manufacturing systems, their non-functional features can be used to identify/trace resources and efficiently evaluate the resource organization, storage, management, or utilization. Therefore, it is necessary to divide non-functional features into the profile and the management information parts. Functional features provide manufacturing resource capabilities. In the cloud manufacturing environment, there are mainly three kind of information in functional features of manufacturing resources, i.e. the task information, the process information, and the production information. The task information closely depends on the design, geometry and processes of products. The process information describes the processing routine to convert the raw material into final products. The production information shows the manufacturing resource status during producing. Various business activities and collaborative manufacturing modes result in different functional features. For example, in a product life cycle, resources are required throughout all the stages, such as process design, shop scheduling, material demand planning, production planning and product simulation. Meanwhile, collaborative manufacturing modes range from workshops to enterprises, or even to value chains. Different information is focused on in various specific manufacturing scenario. Therefore, functional features should be represented in multi-view. This paper organizes functional features in partial order. A comprehensive functional features repository is constructed to reflect the nature of the manufacturing resources flexibly and systematically. Implement

Equipment

III.

PREPARATIONS

The proposed method contains three aspects: (1)Space Abstraction. There are two spaces in cloud manufacturing systems, the manufacturing resource space and the manufacturing capability space. The two spaces are respectively described by uniform formal feature sets. (2)Space Partition. The manufacturing resource space and the manufacturing capability space are partitioned by different granularities and various perspectives. Manufacturing resources are considered as an assembly of various manufacturing capabilities with specific granularity from specific perspective to facilitate constrained operations and thus can perform the desired productions. Space partition is critical to build a flexible and extensibility resource description specification in cloud manufacturing systems. Manufacturing capabilities can be explicitly manifested from a diversity of aspects. Functional features are loosely-coupled with each other. (3)Cell-View. Cell view is the window to display the space partition with specific constraints, by which manufacturing resources could be discovered in particular context. A. Concepts and notations Definition 1. Let MR be the manufacturing resource space and Ω be the capability space over MR . π : Ω → MR is called the cell mapping. For ∀ ci , c j ∈ Ω , i, j ∈ N and

i ≠ j , the following conditions could be satisfied: (1) , π (c j ) ≠ φ ;(2) π (ci ) ∩ π (c j ) = φ ;(3)

π (ci ) ≠ φ

π (ci ) = Ri , Ri ⊆ MR . Ri is the collection of equipment and accessories with the same capability ci . Definition 2. Let π 1 , π 2 be two cell mappings. If for

∀ ci ∈ Ω , ∃c j ∈ Ω , such that π 1 (ci ) ⊆ π 2 (c j ) ,

π 1 is

finer than π 2 , denoted as π 1 ≤ π 2 . If π 1 (ci ) ⊂ π 2 (c j ) , π 1 is called fully finer than π 2 , denoted as π 1 ≺ π 2 . Definition 3. Let Π be the cell mapping set which is called

partitioned by Π1 , Π 2 ,......Π k . ( Π1 , Π 2 ,......Π k ) is called the total-ordered partition sequence of Π on ≤ when the following conditions are satisfied: (1) Every Π i , 1 ≤ i ≤ k ,

Auxiliary

k

is a total-ordered cell mapping subset; (2) Π = ∪ Π i ; (3) i =1

Manufacturing Resource

∀i, j , 1 ≤ i, j ≤ k and i ≠ j , Π i ∩ Π j = φ ; (4) ∀i

Status Accessory Production

Non-Function

Function Process

Profile Information Task

Geometry Method

Manage Information

Category Part Property Intention

Roughcast

( 1 ≤ i ≤ k ),

Π i were not a total-ordered cell mapping subset any more if ∀π , π ∈ Π − Π i , could be added to Πi . A total-ordered partition sequence for the cell mapping set could be achieved by the algorithm shown in Fig.2. Because Π is a finite set and at least one cell mapping would be deleted in each cycle, the algorithm could be terminated in limited steps .

Material

Figure 1. The structure of manufacturing resource features

1004

Algorithm 1. Π _ Partition Input: A cell mapping set Π Output: One of the total-ordered partition sequences ℑ for Π Step1. Π ← {π 1 , π 2 ,..., π n } , ℑ ← () , k ← 0 .

Cell View1

Step2. Repeat 2.1.Take an element π i from Π , Π ← Π − {π i } . 2.2. k ← k + 1 , ∂ k ← {π i } .

Cell View2

Cell View3

Cell View4

Ω

2.3.For every cell mapping π j in Π For each element μ ∈ ∂ k If μ ≤ π j or π j ≤ μ , then ∂ k ← ∂ k ∪ π j and

{ }

{ }

MR

Π←Π− πj .

2.4. ∂ k is appended to ℑ . Step3. Until Π == φ .

MR

MR

Figure 3. The illustration of cell view

Step4. Return the final solution ℑ = (∂ 1 , ∂ 2 ,

,∂k ) .

Figure 2. Algorithm of total-ordered partition

Definition 4. Let ℑ = (∂ 1 , ∂ 2 ,

MR

, ∂ k ) be a total-ordered

partition sequence for Π . ε i is called the F-mapping of ∂ i

if ∀π j ∈ ∂ i (1 ≤ i ≤ k ) , ε i ≤ π j .

Definition 5. Let M = {ε 1 , ε 2 , , ε k } be the F-mapping set of ℑ . CVi =({< ci , Ri >}, ε i ) is called a cell view.

A cell view makes manufacturing resources to be used with pluralism, which could improve utilization and sharing of the resources by partitioning the manufacturing capabilities with the finest granularity in each specific perspective. B. The illustration of cell view Fig.3 illustrates the mechanism of the cell view. There are three levels: the bottom, the middle, and the top levels. The bottom level represents the manufacturing resource space. The middle level denotes the manufacturing capability space. And the top level shows cell views. The four yellow blocks at the bottom represent respective results of partition from different perspectives of the same manufacturing resources space. Each block corresponds to an abstraction perspective. The size of elements within each block represents the different granularities of the same abstraction perspective. The smaller is the size, the finer the granularity. With dashed lines, the manufacturing capability space is divided into many sub-spaces from different perspectives, which illustrates the influences on resource capabilities with different roles of employees, divergence in collaborative modes, and variations in business activities. At the cell view level, manufacturing resources related to manufacturing capabilities with the finest granularity (the core of manufacturing capabilities) in different perspectives are presented to the cloud service platform. In cloud manufacturing systems, manufacturing resources exhibits the distributed, autonomous, heterogeneous and diversity characteristics. It is difficult to represent resources because of the growing categories and quantities of manufacturing resources. Cell views facilitate to discover manufacturing resources.

The suggested mechanism provides flexibility and extensibility for resource description specification. Traditional resource virtualization methods abstract manufacturing resources information according to their usability in the manufacturing organization, the abstraction layers are just like station, cell, shop and factory. Manufacturing capabilities are restricted by the hierarchical structure of the abstraction layers, which make the resource information tightly-coupled with each other. When the manufacturing business environment changes, information cannot be integrated and shared as expected. With the mechanism proposed in this paper, manufacturing resource information is abstracted with a high level from multigranularity and multi-perspective which depends on the inherent features of resources. Some issues of traditional methods can be solved by this mechanism. IV.

PROPOSED METHOD Resource virtualization analyzes, designs, simulates and represents various facets of manufacturing resources. It is foundation for cooperative manufacturing. In cloud manufacturing systems, resource virtualization is the interface between the cloud service platform and heterogeneous manufacturing resources. It transforms the heterogeneous manufacturing resources to homogeneous cloud services with uniform interfaces for manufacturing resources integration and sharing. Different local manufacturing resource models focus on particular manufacturing businesses. Resource virtualization provides a unified and adaptable resources description specification which covers all necessary manufacturing resource features. The specification can interoperate with different heterogeneous resource models. Then the application scope of the models can be extended from local to global networked manufacturing environment. In this paper, resource virtualization which transforms manufacturing resources into cloud services can be implemented by two phases: virtual description and service encapsulation. A. Virtual description The virtual description aims to build a resource description specification which represents the manufacturing resources information comprehensively in a unified virtual resource data

1005

model. It supports various activities related to the manufacturing business. All kinds of manufacturing business information are derived from the virtual resource data model. Fig.4 shows the conceptual structure of the data model, in which three main components are included: the manufacturing resource dictionary, the manufacturing capability dictionary and the cell mapping pool. The manufacturing resource dictionary describes the non-functional features of manufacturing resources. The manufacturing capability dictionary illustrates functional features of manufacturing resources. The cell mapping pool is an intelligent application which facilitates and improves flexibility and extensibility of the data model. As well, the manufacturing business describes both the business and the efficient collaboration strategies. The collaborative mode represents the manufacturing business granularity. The business activity illustrates the manufacturing business perspective. According to the knowledge stored in cell mapping pool, cell views can be generated and updated when a manufacturing resource enters or quits the cloud manufacturing system, or when the new manufacturing business command is coming. Definition 6. The virtual resource data model is CVDM::= (MRD, MCD, CMP), in which: (1) MRD describes the static manufacturing resources, such as physical equipments, the application software and documents; (2) MCD represents the dynamic and invisible manufacturing capabilities, which can only be explicitly exhibited by manufacturing activities; and (3) CMP is a knowledge repository, which connects manufacturing resources with manufacturing capabilities according to specific manufacturing activities or enterprise business goals. Definition 7. MRD::=(ProfileInfo, ManageInfo), in which : (1) ProfileInfo is the basic information of manufacturing resources (such as id, name, type etc.); (2) ManageInfo records the management information (such as provider, operator, location etc.). It ensures manufacturing resources to be tracked throughout the life cycles. Definition 8. MCD::=(CType, CDescription), in which (1) CType is the categories of the manufacturing capabilities (such as design capability, process capability, production capability, simulation capability etc.); (2) CDescription expresses the functional features of some type of manufacturing capabilities. Collaboration mode

Business Activity

Cell View1

Perspective

Granularity

Cell View2

Cell Mapping Pool

…… Cell Viewn

Manufacturing Business ID1

Name

Type

Location



Location



…… IDn Manufacturing Capability Space

TaskInfo

Specify 1 ……



Name

Type

Type 3 Type 2

ProduceInfo



Type 1



Manufacturing Resource Dictionary

… Specify n Manufacturing Capability Dictionary

ProcessInfo

Manufacturing Resource Space

Figure 4. Conceptual structure of virtual resource data model

MCD is an important component of the virtual resource data model. It is the foundation for encapsulating manufacturing resources as cloud services. The diversity of manufacturing resources can be entirely exhibited according to different types of manufacturing capabilities. Definition 9. CMP::=(VP, VG, CapInfo, ResSet). CMP represents the relationship between manufacturing resources and manufacturing capabilities dynamically, in which: (1)VP conveys manufacturing activities. (2)VG delivers collaborative manufacturing modes. (3)CapInfo describes manufacturing capabilities. (4) ResSet presents the manufacturing resources information corresponding to the CapInfo. VP and VG give the constraints to a cell mapping. CapInfo and ResSet are the result of the cell mapping with the constraints. Cell mapping pool enhances the flexibility and extensibility of the virtual resource data model. The dynamic relationship between manufacturing resources and manufacturing capabilities represents the variable status of manufacturing resources, which ensures cooperation and coordination in the cloud manufacturing systems. B. Service encapsulation Service encapsulation extracts functional features of manufacturing resources from the virtual resource data model, and encapsulates them as cloud services to interoperate with the cloud platform in a uniform interface. The resource descriptions are encapsulated by a standard encapsulation operation using a cloud service template. Definition 10. CST::=(SID, CSemInfo, RegInfo, BindInfo, StatusInfo). CST is the cloud service template to simplify the encapsulation operation, in which: (1)SID is the cloud service ID; (2)CSemInfo is the semantic description for the cloud service, which consists of semantic items; (3)RegInfo provides the registration information of the cloud service; (4)BindInfo enables the cloud service to act as an agent; and (5) StatusInfo displays the state of manufacturing resources. The cloud service template provides the necessary items of cloud services to communicate with external environment. It shows semantic functional features of manufacturing resources to facilitate resource discovery. As well, it supports how to deploy, publish and manage the cloud services. Definition 11. EncapOpr::=( Φ , Σ , Ι , Γ ). EncapOpr is the cloud service encapsulation operation, in which: (1) Φ applies for the cloud service ID from the cloud manufacturing system; (2) Σ gives semantic annotations to functional features of manufacturing resources; (3) Ι creates indexes for cloud services when they are registered; and (4) Γ binds the cloud service with an agent. Fig.5 illustrates the encapsulation operation procedure, in which the resource description is gradually adapted with the template in four steps. Adaptor is an encapsulation engine which is responsible for controlling the whole encapsulation operation. As soon as a virtual description document is submitted, the adaptor applies for a cloud service ID. Based on the manufacturing resource ontology, the functional features of the manufacturing resources are semantically annotated. Every

1006

Resource virtual description document

Engine for encapsulating

Adaptor

Identification CSID

Bind with the agent

O1

CS11

Oi

CSi1

CSip

On

CSn1

CSnq

CS1m

……

Semantic annotation

Create indexes

Figure 5. Encapsulation operation

keyword of functional features is translated into a semantic item, by which capabilities of the manufacturing resources can be fully exhibited. Cloud service indexes are created according to semantic items. The cloud service has an index item corresponding to each semantic item. By bound with an agent, a cloud service becomes either an active role or a passive role. As a passive role, it acts as a provider waiting for requirements. While as an active role, it acts as a requester to search for requirements it can provide. V.

promising strategy for large-scale distributed manufacturing systems cooperation and integration. By surveying on the work in networked manufacturing, we find that the existing tools, platforms, prototype application systems can only be applied in small-scale collaborative manufacturing environments. They are ineffective to the large-scale settings in cloud manufacturing systems. In this paper, resource & capability virtualization was considered. A two-phase method was presented to transform manufacturing resources into cloud services. Heterogeneous manufacturing resources are normalized with a flexible and extensible resource description specification. By service encapsulation operation, functional features are further encapsulated as cloud services according to the resource description specification. By this way, resources could be shared by cloud manufacturing systems. However, there are still some critical issues to be solved for practical applications in the future.

AN EXAMPLE

In this section, an example is introduced to illustrate the proposed method. The example considers the physical manufacturing equipment description in cloud manufacturing systems. The processing and production capabilities are fully exhibited. The description for the presented method is implemented with XML, which is popular for data exchange and interoperable web applications. Fig.6 shows the XML schema of the manufacturing equipments. It is a representation for non-functional features of manufacturing equipment. The profile part is used to identify manufacturing equipment throughout the life cycle. The management part is the foundation for utilizing and evaluating manufacturing equipment. The schema can be extended or customized when it is applied to other kind of manufacturing resources. Fig.7 depicts the XML schema of manufacturing capabilities, such as design, process, production, simulation and maintain. The manufacturing resource is attached with a manufacturing capability set. When the manufacturing resource is encapsulated as a cloud service, the manufacturing capability set is the port to communicate with the external environment. Manufacturing capabilities in the set are optional according to manufacturing activities. Fig 8 shows the XML document of a machine. The content of the document is generated by a cell view. The cell view reflects both the manufacturing and the production capabilities of the machine. VI.

CONCLUSIONS AND FUTURE WORK

Cloud manufacturing is a new networked manufacturing model which includes two emerging computing paradigms: Cloud computing and the Internet of Things(IoT). It provides a

1007



Figure 6. The schema of the manufacturing equipment

REFERENCES



[1]

[2]

[3] [4]

Figure 7. The schema of manufacturing capabilities [5]

EqMac001 Machine Normal 500 300 100 Average B P1 Employee317 Corp1 circle raw soft turing crane 0.6 500 busy T3 T6 T8

[6]

[7]

[8]

[9]

[10]

[11]

[12]

[13]

Figure 8. XML document of the equipment

[14]

ACKNOWLEDGMENT

[15]

This work was supported by National Natural Science Foundation of China under Grants Nos. 60873236 and 60973073.

[16]

[17]

1008

B.H.Li, L.Zhang, S.L.Wang, F.Tao, J.W.Cao, X.D.Jiang, X.Song, X.D.Chai, “Cloud manufacturing: a new service-oriented networked manufacturing model,” Computer Integrated Manufacturing Systems, vol 16, pp.1-7, Jan 2010. Y.C. Song, Z.Wang, Q.Lei, “A kind of integrated modeling method for networked manufacturing platform,” Advanced Measurement and Test, Parts 1 and 2, vol 439-440, pp.1012-1017, 2010. H.B.Lan, “Web-based rapid prototyping and manufacturing systems: A review,” Computers in Industry, vol 60, pp.643-656, June 2009. J.Steele, Y.J.Son, R.A. Wysk, “Resource Modeling for the Integration of the Manufacturing Enterprise,” Journal of Manufacturing Systems, vol 19, pp. 407-427, 2001. F.Tao, Y.F.Hu, D.M.Zhao, Z.D.Zhou, H.J.Zhang, Z.Z.Lei. “Study on manufacturing grid resource service QoS modeling and evaluation,” International Journal Advanced Manufacturing Technology, vol 41, pp.1034-1042, May 2009. B.L.Dong, G.N.Qi, X.J.Gu, X.T.Wei, “Web service-oriented manufacturing resource applications for networked product development,” Advanced Engineering Informatics, vol 22, pp. 282-295, 2008. W.Yang, X.Xu, “Modelling machine tool data in support of STEP-NC based manufacturing. International Journal of Computer Integrated Manufacturing,” vol 21, pp. 745-763, 2008. P.Vichare, A.Nassehi, S.Kumar, S.T. Newman, “A unified manufacturing resource model for representing CNC machining systems,” Robotics and Computer-Integrated Manufacturing, vol 25, pp.999-1007, 2009. A.A.Kadir, X.Xu, E.Hammerle, “Virtual machine tools and virtual machining a technological review,” Robotics and Computer-Integrated Manufacturing, vol 27, pp.494-508, 2011. C.R.Monroy, J.R.V.Arto, “Analysis of global manufacturing virtual networks in the aeronautical industry,” Internation Journal of Production Economics, vol 126, pp.314-323, April 2010. L.Monostori, B.Cs.Csaji, B.Kadar, A.Pfeiffer, E.I.Zudor, Zs.Kemeny, M.Szathmari, “Toward adaptive and digital manufacturing,” vol 34, pp.118-128, April 2010. E.Oztemel, E.K.Tekez, “A general framework of a reference model for intelligent integrated manufacturing systems,” Engineering Applications of Artificial Intelligence, vol 22, pp.855-864, Jan 2009. K.Y.Chen, C.J.Chen, “Applying multi-agent technique in multi-section flexible manufacturing system,” Expert Systems with Applications, vol 37, pp.7310-7318, 2010. H.Panetto, A.Molina, “Enterprise integration and interoperability in manufacturing systems: trends and issues”, Computers in Industry, vol 59, pp.641-646, May 2008. M.Salvadores, P.Herrero, J.L.Bosque, M.S.Perez, “A semantic collaborative awareness model to deal with resource sharing in grids,” Future Generation Computer Systems, pp. 276-280, 2010. G.A.Vouros, A.Papasalouros, K.Tzonas, A.Valarakos, K. Kotis, J.A. Q.Ruiz, P.Lamarre, P.Valduriez, “A semantic information system for services and traded resources in Grid e-markets,” Future Generation Computer Systems, vol 26, pp.916-933, April 2010. A.G.Gunendran, R.I.M.Young, “An information and knowledge framework for multi-perspective design and manufacture,” International Journal of Computer Integrated Manufacturing, vol 26, pp.326-338, 2006.

Suggest Documents