An agent-based service composition framework

Document Sample
An agent-based service composition framework Powered By Docstoc
					                            An agent-based service composition framework


                          L Zhang 1, YQ Guan1, F Tao1, YL Luo1, AR Hu1, Ralph C. Huntsinger2
                               1. School of Automation Science and Electrical Engineering,
                                       Beihang University, Beijing 100191, P. R. China
                                         2. California State University, Chico, USA
          zhanglin@buaa.edu.cn; guan-yq@asee.buaa.edu.cn; ftao@buaa.edu.cn; yongliang2002@gmail.com;
                                       thiefkeeper@163.com; drralph@huntsinger.net



Key words: Service composition, agent, composition                 for a system that provides tools for adaptive service
framework.                                                         composition and provisioning. In their following studies,
                                                                   they [14] propose a method to enable progressive
Abstract:                                                          composition of non-Web-service-based components, such
      In order to make service to be smart, and enable it to       as native widgets, portlets, legacy systems, web
be delivered, deployed, invoked, and used in a more                applications, and Java Beans. A novel application of
intelligent way, agent technology has been introduced in           semantic annotation with a matching algorithm to find
the service composition. Based on the authors’ previous            functionally equivalent components is introduced by them.
related work, the entire life cycle of service composition         Segev and Toch [15] proposed a two-step and
was investigated. A four layered agent-based service               context-based semantic approach to the problem of
composition framework (ASCF) is proposed. The key                  matching and ranking Web services for a possible service
functions and services provided by each layer in the               composition. Obrenovic and Gasevic [16] investigated the
proposed ASCF are described in detail.                             requirements for spreadsheet-based service composition
                                                                   and presented a framework that implements these
1   INTRODUCTION                                                   requirements, the framework can enable spreadsheets to
     Service composition has been widely studied both in           send requests and retrieve results from various local and
the industrial community and in academia. In the                   remote services. Estevez-Ayres et al. [17] investigated a
industrial community, the description languages for                model for QoS-aware service composition in distributed
service composition are emphasized, and several web                systems with real-time and fault-tolerance requirements,
service oriented flow descriptive languages have been              and Guo et al. [18] investigated three kinds of
proposed, such as WS-BPEL [1], OWL-S [2], WSCI [3],                correlations      in     service     composition    and    a
and WS-CDL [4]. In academia, existing work on service              correlation-aware QoS model was studied, as well as the
composition is concentrating on service composition                impact of each kind of correlation on the whole QoS of
framework [5-6], service composition validation [7-8],             service composition was investigated. In 2010, a special
and the service composition method, especially on the              issue on transactional web services has been published by
service     composition      method.      Existing     service     IEEE Transactions on Service Computing, and some
composition methods can be primarily summarized into               problems related to service composition from the aspect
the following five categories: (a) Business flow-based             of transactional mode are researched [19-21].
service     composition     [9-10],     (b)   AI     (artificial        The       authors   have      done   much     work   on
intelligence)-based service composition [10-11], (c)               multi-objective MGrid resource service composition and
Graph-based service composition [11-12], and (d) QoS               optimal-selection (MO-MRSCOS) problems [22-23]. In
(quality of service)-based service composition.                    the authors’ previous work, the formulation for the
     Furthermore, Sheng et al. [13] described the design           MO-MRSCOS problem to minimize execution time and
                                                                                            Ta
                                                                                           req s k o
                                                                                              uir r us
                                                                                                  em er
                                                                                                     en
                                                                                                        ts
                            rv po ct
                                e te
                          se om tra
                              ic si
                            c bs
                               A




                                                                            D
                                                                             es
                                                                                 -p
                                                                                  ha
                                                                                      se
                                               Dep-phase
             Abst osite
                 ract
             compce
              servi




                                                           Co oor pos r tion




                                                                                                                   ase
                                                             c m to osi
                                                             m din itio




                                                                                                                 ph
                                                               po at n
                                                               Co on mp ol te
                                                                                   as e

                                                                 sit ion
                                                                 m Co ntr osi


                                                                                 ph




                                                                                                              M-
                                                                    io
                                                                             M

                                                                       n
                                                                       i




                                                                                                             E&
                                                                          E&
                                                                         co mp ce n
                                                                           Co rvi tio
                                                                             se elec
                                                                              management
                                                                              ship
                                                                              Correlation
                          D




                                                                               Proc
                                                                               s e ss
                                                                               Aggr
                          ep




                                                                                QoS ation

                                                                                bou
                                                                                 evalu
                              -p
                                 h




                                                                                     ndin
                                as




                                                                                      essin




                                                                                                  e
                                  e




                                                                                                        s
                                                                                               vic ion
                                                                                          g




                                                                                            ser olut vice
                                                                                            g




                                                                                        a te n s er
                                                                                     did tio e s
                                                                                   an posi tabl path
                                                                                  C m cu n
                                                                                   co exe sitio
                                                                                    or mpo
                                                                                     co

                                                                                                                                   

                                         Fig.1 Life-cycle of resource service composition [22] 


cost, and maximize the reliability has been presented.                          increasing complexity of task to be addressed, some
Four basic resource service composite modes (RSCM),                             smarter and more intelligent methods for service
i.e., sequence model, parallel model, selective model, and                      composition are required.
circular model, for CRS are described. The principles for                       At the same time, agent technology is become a
transforming a complicated RSCM into a simple                                   popular technology in the computer science and
sequence RSCM, and the aggregating QoS evaluation                               artificial intelligent areas. It has been widely studied
method for CRS are presented. A particle swarm                                  and used in many areas such as data mining, business,
optimization (PSO) combined with a non-dominant                                 medical, and so on. One main reason for using agents
sorting technique is used to implement the optimal                              is that they can enable software models or functions to
selection of service composition paths.                                         be smarter and more intelligent because they have
     The above work has played a very important role in                         specific characteristics such as autonomy, adaptive,
addressing    service     composition.     However,            as   the         cooperative, interactivity, intelligent, learning,
increasing amount of service quantity and class, and the                        ruggedness, continuous, coordinator, and mobility.
Therefore, in order to make service composition to be                 The service composition process can be classified
smarter and more intelligent, the agent technology is           into three phases from the service composition life-cycle,
introduced into service composition and an                      and they are the design phase (Des-phase), deploy phase
agent-based service composition framework is                    (Dep-phase),      and    Execution   &    Monitor     phase
investigated in this paper.                                     (E&M-phase).
      The paper is organized as follows. Section 1                    Des-phase In the design phase, a user’s requirement
investigates the related work on service composition. The       or task is analyzed, including the operation of the task
life-cycle of service composition is studied in Section 2.      description,     task    decomposition,   task      function
A five layered agent-based service composition                  requirement extraction, and the task process requirement
framework is presented in Section 3. Section 4 summaries        extraction. The input of Des-design is a complex task or
the paper.                                                      user requirement, and its output is an abstract composite
                                                                service.
2   LIFE-CYCLE OF SERVICE COMPOSITION                                 Dep-phase Dep-phase is responsible for matching
    REQUIREMENTS                                                and searching for the qualified element candidate




                                                                                                                       
                               Fig. 2 Five layered agent-based service composition Framework
service according to the function requirements and flow        3.2    Match and search agent (MSAgent)
requirement generated in Des-phase, and evaluation of                The MSAgent is designed for searching and matches
the candidate service quality and then selecting the           the qualified service according to the task requirements
optimal service, and generating the corresponding service      generated     by      a     task    agent,     including          function
composition solutions. The operations involved in              requirements, process requirement, ability requirement,
Dep-phase are similarly computing, interface matching,         and so on. The outputs of MSAgent are qualified
function matching, flow matching, semantic matching,           candidate services. According to the service provided by
service evaluation and ranking, and the service selection.     MSAgent, it consists of five sub-agents such as algorithm
The input of Dep-pahse is the abstract composite service       agent (AlgAgent), search engine agent (SEngAgent),
generated in Des-phase, and its outputs are the candidate      function requirement match agent (FRMAgent), process
service composition solutions or the executable service        requirement match agent (PRMAgent), and ability
composition path (ESCP).                                       requirement match agent (ARMAgent).
      E&M-phase In the E&M phase, the comprehensive
quality of all candidate service composition solutions or                                    User or task
                                                                                             requirements
ESCP generated in Dep-phase are evaluated, and their
aggressing QoS are evaluated while considering the                                User

correlations among the element services. The optimal                 Task Agent              Task description
                                                                                                  agent
service composition solution or ESCP is selected and the
related concrete services are bounded and invoked by this
service composition engine. The operations involved in                                         Task description
the E&M phase including optimal-selection of service                                              document

composition solutions or ESCP are, aggressing QoS
calculation and evaluation, correlation management,
                                                                                         Task decomposition Agent
service composition coordinating, service composition
                                                                               Function                          Process
monitoring, and service composition control, etc.                          requirements pare                requirements pare
                                                                                 agent                            agent
3 AGENT-BASED SERVICE COMPOSITION
      FRAMEWORK
      There are four agents designed for this framework                              Function                        Process
                                                                                   requirements                   requirements
such as task agent, match and search Agent, QoS
processing agent, and composition agent. The framework
is shown in Figure 2.
                                                                                              Match and search
                                                                                                   agent
3.1   Task Agent
       The task agent is responsible for accepting tasks or                       Fig.3 Work flow of task agent
user requests, and realizing their digital description, then
decomposing them into corresponding subtasks if                       AlgAgent provides various algorithm services
applicable and extracting the related function and process     involved in MSAgent, such as ontology similarity match
requirements. The work flow of the task agent is               algorithms,     service       description      information          match
illustrated in figure 3, which consists of three sub-agents    algorithms, search algorithms, and so forth. The
such as task description agent, function requirements          SEngAgent is responsible for searching out the service
parse agent, and task process requirements parse agent.        qualified for the task requirements. The FRMAgent is
                                                               responsible for the function requirement match. The
PRMAgent      is   designed    for   process   requirements   service, (e) coordinating and adjusting the composition
matching and ARMAgent is responsible for the ability          solutions according to the monitoring results. The
match.                                                        structure of the composition agent is illustrated in figure 2,
                                                              and it consists of five sub-agents such as the composite
3.3   QoS processing agent                                    service generation agent (CSGAgent), composite service
      The QoS processing agent (QoSAgent) is designed         optimal-selecting         agent       (CSOSAgent),          service
to provide comprehensive process services for the             composition       control     agent     (SCCAgent),         service
specific operation to service, such as service selection,     composition coordinator agent (SCCoAgent), service
service composition. The detailed services involved in        composition monitoring agent (SCMAgent). Each agent
QoSAgent including QoS extraction are, QoS search,            is designed to address one of the problems mentioned
QoS updating, QoS evaluation, QoS monitor, and QoS            above, and the function of each agent is described as
comparison. In order to implement the above six QoS           follows its definition.
processing services, the corresponding six sub-agents are     4 CONCLUSIONS AND FUTURE WORK
designed as shown in Figure 4, and they are the QoS                  How to enable the services to be smarter, and make
extraction agent (QoSEAgent), QoS search agent                them to be deployed, delivered, invoked and used in a
(QoSSAgent), QoS updating agent (QoSUAgent), QoS              more intelligent and easy way is a key issue in the
evaluation agent (QoSEvAgent), QoS monitor agent              research field of service management. In this study, the
(QoSMAgent), and QoS comparison agent (QoSCAgent).            agent is introduced into service composition, and a four
                                                              layered agent-based service composition framework is
                                                              proposed, as well as the key functions and services
                                                              provided by each layer. In the future, the method for
                                                              second service modeling based on agent technologies will
                                                              be emphasized, and the details for each agent in the
                                                              proposed framework will be studied. The proposed
                                                              service composition method will be tested and applied in
                                                              a cloud manufacturing platform and a cloud simulation
                                                              platform, which are topics of research and developed by
                                                              the authors’ research group.

                                                              ACKNOWLEDGE

                                                                    This paper is partly supported by the NSFC
                                                              (National     Science     Foundation       of   China)      Project
                                                              No.61074144 and No.51005012 in China, and the
                                                         
                                                              Fundamental Research Funds for the Central Universities
           Fig.4 Structure of QoS processing agent
                                                              in China.

3.4   Composition agent                                       REFERENCES
      The composition agent is responsible for (a)
generation of the composite service solution or executable    [1]   Web Services Business Process Execution Language Version 2.0.
path, (b) evaluating the aggressing QoS of the generated            [EB/OL]
composite service, (c) launching and controlling the                http://docs.oasis-open.org/wsbpel/2.0/CS01/wsbpel-v2.0-CS01.ht
execution of service composition, (d) monitoring the                ml.
running state and performance information of selected         [2]   OWL-S: Semantic Markup for Web Services [EB/OL]
      http://www.w3.org/Submission/OWL-S/.                                 [15] A. Segev, and E. Toch, “Context-Based Matching and Ranking of
[3]   Web Services Choreography Interface (WSCI) 1.0. [EB/OL].                   Web Services for Composition,” IEEE Transactions on Service
      http://www.w3.org/TR/wsci/ .                                               Computing, vol. 2, no.3, pp.220-222, Jul.-Sep. 2009.
[4]   Web Services Choreography Description Language Version 1.0.          [16] Z. Obrenovic, and D. Gasevic, “End-User Service Computing:
      [EB/OL].                                                                   Spreadsheets as a Service Composition Tool,” IEEE Transactions
       http://www.w3.org/TR/2004/WD-ws-cdl-10-20041217                           on Service Computing, vol.1, no.4, pp.229-242, Oct.-Dec. 2008.
[5]   J. Hu, J. Li, and Z. Zeng, “SWSCF: A semantic-based Web              [17] I. Estévez-Ayres, P. Basanta-Val, M. García-Valls, J. A. Fisteus,
      service composition framework,” Journal of Networks, vol. 4, no.           and L. Almeida, “QoS-Aware Real-Time Composition Algorithms
      4, pp. 290-297, June 2009.                                                 for Service-Based Applications”, IEEE Transactions on Industrial
[6]   M. Al-Nory, A. Brodsky, and H. Nash, “A service composition                Informatics, vol.5 , no.3, pp.278-288, Aug.2009
      framework for decision making under uncertainty,” Lecture Notes      [18] H. Guo, F. Tao, L. Zhang, S. Su, N. Si. Correlation-aware web
      in Business Information Processing, vol.24, pp.363-375, 2009.              services composition and QoS computation model in virtual
[7]   V. Tosic, D. Mennie, and B. Pagurek, “On dynamic service                   enterprise [J]. International Journal of Advanced Manufacturing
      composition and its applicability to e-business software systems,”         Technology, 2010, 51 :817-827
      In   Proceedings    of   the    WOOBS      2001   (Workshop     on   [19] J. El Hadad, M. Manouvrier, and M. Rukoz, “TQoS:
      Object-Oriented Business Solutions) workshop (at ECOOP 2001),              Transactional and QoS-Aware Selection Algorithm for Automatic
      June 18-22, 2001, pp. 95–108.                                              Web Service Composition,” IEEE Transactions on Service
[8]   B. Benatallah, Q. Z. Sheng, and M. Dumas, “The self-serv                   Computing, vol. 3, no. 1, pp.73-85, Jan.-Mar. 2010.
      environment for web services composition,”         IEEE Internet     [20] W. Gaaloul, S. Bhiri, and M. Rouached, “Event-Based Design
      Computing, vol.7, no. 1, pp.40-48, 2003.                                   and Runtime Verification of Composite Service Transactional
[9]   F. Baader, C. Lutz, M. Miličić, U. Sattler, and F. Wolter, “A              Behavior,” IEEE Transactions on Service Computing, vol. 3, no.
      Description Logic Based Approach to Reasoning about Web                    1, pp.32-45, Jan.-Mar. 2010.
      Services,” In Proceedings of the WWW 2005 Workshop on Web            [21] A. Liu, Q. Li, L. Huang, and M. Xiao, “FACTS: A Framework
      Service Semantics (WSS2005), May 10–14, 2005                               for Fault-Tolerant Composition of Transactional Web Services,”
[10] S. McIlraith, and T. C. Son, “Adapting golog for composition of             IEEE Transactions on Service Computing, vol. 3, no. 1, pp.46-59,
      semantic web services,” In Proceedings of the 8th International            Jan.-Mar. 2010.
      Conference on Knowledge Representation and Reasoning, Apr.           [22] F. Tao, D. Zhao, Y.F. Hu, Z. Zhou. Resource service composition
      22-25,                                                                     and its optimal-selection based on particle swarm optimization in
[11] R. Zhang, I.B. Arpinar, and B. Aleman-Meza, “Automatic                      manufacturing grid system [J]. IEEE Transactions on Industrial
      Composition of Semantic Web Services”. In Proceeding of the                Informatics, 2008,4(4):315-327
      International Conference on Web Services (ICWS03), Amsterdam,        [23] F. Tao, Y.F. Hu, L. Zhang. Theory and practice: Optimal
      Feb. 14-16, 2003, pp. 38-41.                                               Resource Service Allocation in Manufacturing Grid [M]. China
[12] J. Rao, P. Küngas, and M. Matskin, “Composition of Semantic                 Machine Press, April. 2010.
      Web services using Linear Logic theorem proving,” Information
      Systems, no. 31, pp.340–360, 2006                                    BIOGRAPHY
[13] Q.Z. Sheng., B. Benatallah., Z. Maamar, and A.H.H. Ngu,
                                                                               Lin Zhang received the B.S. degree in 1986 from the Department
      “Configurable Composition and Adaptive Provisioning of Web
                                                                           of Computer and System Science at Nankai University, China. He
      Services,” IEEE Transactions on Service Computing, vol. 2, no. 1,
                                                                           received the M.S. degree and the Ph.D. degree in 1989 and 1992 from
      pp.34-49, Jan.-Mar. 2009.
                                                                           the Department of Automation at Tsinghua University, China, where he
[14] A.H.H. Ngu, M.P. Carlson, Q.Z. Sheng, and H-Y. Paik,
                                                                           worked as an associate professor from 1994. He served as the director of
      “Semantic-Based Mashup of Composite Applications,” IEEE
                                                                           CIMS Office, National 863 Program, China Ministry of Science and
      Transactions on Service Computing, vol. 3, no. 1, pp.2-15,
                                                                           Technology, from December 1997 to August 2001. From 2002 to 2005
      Jan.-Mar. 2010.
                                                                           he worked at the US Naval Postgraduate School as a senior research
associate of the US National Research Council. Now he is a full              [SCS] and a member of the [SCS] international board of directors, he is
professor in Beijing University of Aeronautics and Astronautics. He is       a member of the Global Advisory Board of the International Federation
an Editor of “International Journal of Modeling, Simulation, and             of Non-Linear Analysis, he is amenber of the editorial board of the
Scientific Computing”, and “Simulation in Research and Development”.         International Journal of Computer Science and SystemsBiology, he is a
His research interests include integrated manufacturing systems, system      Professeur Associe au LITIS, Laboratoire d'Iformatique du Traitment de
modeling and simulation, and software engineering.                           l'Information et des Systemes INSA-ROUEN (Rouen, Haute Normandie)
     Yongqiang Guan received his M.S. degree in applied mathematics          Institut National des Sciences Appliquees de Rouen FRANCE. He is an
From LangFang Normal College and XiangTan University, LangFang               Adjunct Full Professor, Department of Computing Science, Humboldt
China in 2007 and 2010, respectively. He is currently working toward         State University, California, The California State University System,
the Ph.D. degree in complex systems at Beihang University. His current       College of Natural Resources and Sciences USA, He is a Registered
research interests are in the fields of coordination of multiagent systems   Chemical Professional Engineer (MT2530E). He the Founder and
and consensus problems.                                                      International Ambassador for the International McLeod Institute of
     Fei Tao is currently an associate professor at Beihang Universigy       Simuation Sciences. His current research interest is in High Level
since April 2009. His research interests include service-oriented            Simulation Languages for the computer analysis of complex engineering
manufacturing, intelligent optimizaiton theroy and algorithm, and            systems.
resource servcie management. He is author of one monograph and over
20 journal articles of these subjects. Dr. Tao was awarded the Excellent
Doctoral Dissertation Award of Hubei Province, China and was elected
to be a research affiliate of CIRP in 2009. He is the founder and
editore-in-chief   of    Internatioanl    Journal    of    Servcie    and
Computing-Oriented Manufacturing (IJSCOM).
     Yongliang Luo received the B.S. degree and the M.S. degree in
2006 and 2009 from the Department of Computer Science at Shandong
University of Science and Technology, China. He is currently working
for the Ph.D. degree in modeling simulation theory and technology at
Beihang University. His research interests include service-oriented
manufacturing and integrated manufacturing system.
     Anrui Hu receiced his B.E. degree in electrical engineering and
automation from Shandong Agriculture University, China in 2005. He
received his M.E degree in control theory and control engineering from
Shandong University of Science and Technology in China in 2010. He is
working for the Ph.D. degree in modeling simulation theory and
technology at Beihang University at present. His research interests are
service-oriented manufacturing and cloud manufacturing.
     Ralph C Huntsinger, Ph.D. is Emeritus Professor of Mechanical
Engineering and Emeritus Professor of Computer Science, California
State University, Chico USA. He received his M.S. and Ph.D.degrees in
Chemical Engineering from Montana State University, Bozeman MT
USA.     He is an Editor-in-Chief of the “International Journal of
Modeling, Simulation, and Scientific Computing”, and “Simulation and
Modeling in Research and Development”, and a member of the Senior
Advisory board of SIMULATION, the transactions of the Society for
Modeling and Simulation International [SCS], he is a past prsident of

				
DOCUMENT INFO
Shared By:
Categories:
Tags:
Stats:
views:34
posted:10/15/2011
language:English
pages:7