www.ijcsi.org 201 Control Theory baed Approach for the Improvement of Integrated Buine Proce Interoperability Abderrahim Taoudi 1, Bouchaib Bounabat 2 and Badr Elmir 3 1 Al-Qualadi Reearch & Development Team, ENSIAS, Univerité Mohammed V Souii BP 713, Agdal Rabat, Maroc 2 Al-Qualadi Reearch & Development Team, ENSIAS, Univerité Mohammed V Souii BP 713, Agdal Rabat, Maroc 3 Al-Qualadi Reearch & Development Team, ENSIAS, Univerité Mohammed V Souii BP 713, Agdal Rabat, Maroc Abtract Many framework are available today to help organizing and performing Enterprie Interoperability project efficiently. There are alo many metric available to meaure the interoperability degree between ytem. However, there i a real lack in methodolog to control Enterprie Interoperability improvement project execution. The aim of thi paper i to introduce a new approach to control interoperability improvement project execution by uing control theory, project planning theory and. Keyword: ; Control theory; Enterprie Interoperability; Project planning theory; Interoperability improvement; Automated Buine Procee. 1. Introduction Interoperability can be defined a the ability for two (or more) ytem or component to exchange information and to ue the information that ha been exchanged [1]. In the current buine environment, haring information and competenc internally, between department and employee, and externally with partner make compan much more competitive. A ucceful implementation of interoperability will help compan to optimize their buine procee, reduce their cot, and maximize ervice quality. In the Enterprie Interoperability area, many reearch project have been launched in the lat decade i.e. ATHENA [2], INTEROP [3]. Today, there i a number of framework that were developed and validated and are available to ue i.e. Chen et al. [4], ATHENA [2], LISI [5], IDEAS [6], EIF [7]. Concerning enterprie Interoperability meaurement, many approache and meaure are alo available. Ford et al. [8] lited a number of them. There are alo other new meaure like Chen et al [9] and [10]. focue on meauring the interoperability degree of an automated buine proce with it environment. It take into account three main apect: Interoperability maturity level of the environment where the tudied proce i located. Compatibility degree of the external interface of the buine proce with it ecoytem. Operational performance of the upport ytem. Managing and controlling the execution of interoperability improvement project raie many challenge. Given the current and targeted interoperability degree a well a the available reource (i.e. Budget Allocation, Human Reource), the firt challenge conit in finding the optimal plan for an efficient management of thee project. The econd challenge i the ability to handle unexpected event that can be encountered during project execution, o that the manager can know exactly how many additional reource ha to be allocated to correct the deviation from the project optimal plan. The available framework and metric are not currently ufficient to handle the aforementioned challenge. The aim of thi paper i to propoe a new approach to Control the execution of interoperability improvement project. The propoed approach will be baed on mature and proven tool: the framework of chen et al [4] (currently under CEN/ISO tandardization proce) a the interoperability framework, [10] a the interoperability quantitative metric, Project Planning theory to define the optimal plan and Control theory to control project execution. 2. Overview of i a new quantitative ratio metric to meaure interoperability between automated buine procee that wa developed in [10]. With thi ratio, an organiation can
www.ijcsi.org 202 evaluate, at any time and in a quantitative way, the degree of interoperability of it automated buine procee. take into account three kind of interoperability meaurement a o a: 1. to quantify the firt kind of interoperability, Interoperability potentiality, by uing the five level of IMML (Interoperation Maturity Model Level) [10] calculated a bellow: 2. to quantify the econd kind of interoperability, Interoperability compatibility, by uing a modified matrix of Chen et al [10], ee Table 1. Buine Table 1: Interoperability compatibility Conceptual Organizational Technology ynt actic em antic Aut horit rep oab ilitie organi ation platf orm com muni catio n 0/1 0/1 0/1 0/1 0/1 0/1 Proce 0/1 0/1 0/1 0/1 0/1 0/1 Service 0/1 0/1 0/1 0/1 0/1 0/1 Data 0/1 0/1 0/1 0/1 0/1 0/1 (1) By noting dc ij the element of thi matrix, thi potential i calculated a bellow: dc ij i given the value 0 if the criteria in an area marked atifaction; otherwie if a lot of incompatibilit are met, the value 1 i aigned to dc ij. 3. to quantify the third kind of interoperability, Interoperability performance, by uing thee three element: DS : the overall availability rate of application erver. QoS : the ervice quality of different network ued for interacting component communication. TS : the end uer atifaction level about interoperation. (2) Thi potential i: Uing thee three previou indicator, i calculated a bellow: (4) Uing thi ratio, [11] define a tool, Interoperability Monitoring Tool (IMT), which ha three module: Module 1: For aeing interoperability at a pecific period. Module 2: For propoing a cenario to reach a planned degree of interoperability. Module 3: For giving the prerequiite of going from a maturity level to the next one. 3. Defining the optimal plan of the interoperability improvement project Project planning ha different meaning in project management. In thi paper, Project Planning i the act of building the tak by tak chedule which we will call the Project Plan. The optimal plan i the project plan that minimize one or more optimization criteria: Cot, Reource and Time. The high level objective of the interoperability improvement project i to improve interoperability by paing from an initial R i, which i the actual tate of interoperability, to a targeted R t. To define the optimal plan of thee project, we propoe to follow thee tep: Definition of the project objective Definition of the optimal plan uing project planning theory. 3.1 Project objective definition The high level objective of the interoperability improvement project defined above mut be decompoed to clear, concie and meaurable objective which will be ued to plan the project properly. To do o, the Periodic Interoperability Monitoring Tool (IMT) [11], can be ued to define a clear cenario to reach the deired Rt. the propoed cenario will define: (3) The target Maturity Level. The prerequiite to reach thi target Maturity Level. The incompatibilit to remove.
www.ijcsi.org 203 The target operational performance ratio: Availability rate of application erver, The QoS of different network and end uer atifaction level. 3.2 Optimal plan definition Uing the objective a defined above, there are many planning method and tool to define the optimal plan taking into account reource, cot and time. The paper [12] lit many determinitic and non determinitic mathematical model ued to define optimal plan. Mot of thee model are already automated. Bellow ome example of thee model: The tandard Project Management model, PMBOK [13]. Critical Path Method, CPM, and PERT. Non-reource-contrained NPV maximization. The Reource-Contrained Project Scheduling Problem, RCPSP. The Multi-mode Reource-Contrained Project Scheduling Problem, MRCPSP. The project planning theory will help u define the optimal plan to atify the project objective lited in the ection 3.1. The following table 2 will preent a template incorporating the core element ued to define optimal plan: Id Decription Precedent tak Table 2: Decription Layout Duration (in week) Reource need Element Initial Value Target Value element are the elementary component ued to calculate which are: IMML, DS, QoS, TS and the twenty four dc ij (i take value from 1..4, and j take value from 1..6). 4. Control of interoperability improvement project execution Without careful monitoring and control, many project fail to achieve the expected reult. The aim of thi phae i to meaure actual execution, compare it with the optimal plan, analyze it and correct the deviation. To achieve thi goal, we will ue the feedback control theory. 4.1 Feedback control theory Feedback control theory i widely ued in many domain i.e. manufacturing, electronic and phyic. It ued alo in computer cience i.e. apache [14], web erver [15], lotu note [16], internet [17] and network [18]. A feedback control ytem, alo known a cloed loop control ytem, i a control mechanim that maintain a deired ytem output cloe to a reference uing information from meaurement of output. The feedback control diagram adopted by thi paper i illutrated in Figure 1. r(t) + - e(t) = r(t) y(t) w(t) Controller Fig. 1 Feedback control diagram. The plant i the ytem to be controlled. In our cae, it the interoperability improvement project. It ha a controlled input (denoted by w(t)), and a meaured output (denoted by y(t)). The controller take a input the control error (denoted by e(t), which i the difference between the oberved value and the reference value), and it adjut the input of the plant ytem to minimize thi error. Becaue of the dicrete nature of the ytem, we will adopt a dicrete time approach with uniform interval ize (Day, Week, two Week, or Month). 4.2 reference Definition The reference i the of the ytem. It curve will be derived from the optimal plan. We will take into account the finihed tak to calculate the projected at a time t. The objective of the control ytem i to minimize the deviation between the deired baed on the optimal plan and the meaured. 4.3 Modeling the plant ytem y(t) Plant The plant ytem i the interoperability improvement project. The input of the plant ytem, at a time t, i the effort conumption at thi time to releae the project. It can be the reource of the project or budget allocation. The
www.ijcsi.org 204 output of the plant ytem, at a time t, i the at thi time. Bellow i the definition of the characteritic of the plant ytem illutrated in Figure 2: w(t) = the effort conumption at time t to releae the project (reource of the project, budget allocation). y(t) = meaured of the ytem at time t r(t) = the deired of the ytem at time t baed on the optimal plan. We will model the plant ytem a a black-box. We will focu on the behavior of the ytem not on the internal ytem contruction detail which are conidered complex. To do o, we will ue a tatitical approach. The model adopted i the tatitical model ARMA. To keep thing imple, we will adopt ARMA Model of firt order. a and b are contant which will be etimated tatitically. Thee contant can be etimated by varying input (w(t)), and calculating the reulting (y(t)). For each value of the effort w (reource, budget allocation), an automated project planning oftware can be ued to calculate the optimal plan and derive the value for the (y(t)). Uing thee experiment, we can etimate the contant a and b tatitically. The ue of an ARMA model with greater order will give a more precie approximation of the plant ytem. The tranfer function of equation (5) i 4.4 Modeling the controller According to [19], there are four propert of feedback control ytem to verify: Stability: a ytem i aid to be table if for any bounded input the output i alo bounded. Accuracy: a ytem i accurate if the meaured output converge to the reference input. Settling time: a ytem ha hort ettling time if it converge quickly to it teady tate value. Overhooting: a ytem that achieve it objective without overhoot, that i without exceeding an upper limit. There are three baic controller model: Proportional Controller: w(t) = K*e(t) Integral Controller: w(t) = w(t-1) + K*e(t) Differential Controller: w(t) = K*(e(t)-e(t-1)) The contant K i called the gain. To achieve the four propert of our tudied feedback control ytem, the (5) (6) model that we will adopt i the Proportional-Integral model (PI Model): The tranfer function of thi PI controller i: Thu, we can define the following objective for our deign: The ytem i table. The teady tate error i minimized The ettling time doe not exceed a contant value K. Maximum overhoot doe not exceed a contant value Mp. Uing thee objective, [19] dicue in detail the procedure to calculate the appropriate Kp and Ki of the model. With the plant and controller modelled, the control ytem of interoperability improvement project i totally defined. 5. Cae tudy To illutrate the approach, we will ue the ame e- government example a in [11]. Thi cae conit of an online payment for health care ervice in a public hopital. It wa ued in [11] to illutrate aement and the uage of the IMT Tool. Thi ytem i decribed in Figure 2. Fig. 2 Online payment buine proce. The main objective of thi cae tudy i to illutrate the detail of tep and calculation ued by the approach preented in thi paper. (7) (8)
www.ijcsi.org 205 5.1 Initial aement During the implementation phae, three incompatibilit were detected: Exchange with mutual ervant: Infratructure are not compatible. It a Buine/Technology platform and communication incompatibility. Exchange with National ocial ecurity fund: Period for data up-dating not-ynchronized. It a Data/Organizational incompatibility. Exchange with private inurance: Proce decription model can t exchange information. It a Proce/Conceptual yntactic and emantic incompatibility. The initial interoperability compatibility matrix i lited in Table 3: Buine Table 3: Interoperability compatibility Conceptual Organizational Technology ynt actic em antic Aut horit rep oab ilitie organi ation platf orm 0 0 0 0 1 1 Proce 1 1 0 0 0 0 Service 0 0 0 0 0 0 Data 0 0 0 1 0 0 com muni catio n Uing the framework defined in [10] and in ection 2 of thi paper, the initial interoperability aement i decribed in Table 4: Table 4: Initial value Metric Decription Value Maturity Level IMML 0,4 Interoperability compatibility DC 0,79 (Baed on Table 3) Overall application DS 0,9 erver availability Network quality of QoS 1 ervice End uer TS 0,8 atifaction metric 0,69 Ta k Id 1 2 3 ak4 5 5.2 Project objective definition The targeted i 0,8. The propoed cenario to reach thi targeted i: Remove the tree incompatibilit of the ytem. Improve the Overall application erver availability to be 1. Improve the end uer atifaction level to be 1. 5.3 Optimal Plan Definition Uing thee objective, the project tak are defined in Table 5. The duration unit i the week: Decription Removing exchange with mutual ervant incompatibilit Removing exchange with National ocial ecurity fund incompatibilit Removing exchange with private inurance incompatibilit Improving the Overall application erver availability Improving the end uer atifaction level Table 5: decription Precede Durati Reourc nt tak on (in e need week) Elemen t Initi al Valu e Targ et Valu e 3 5 dc 15, dc 16 0 1 5 6 dc 44 0 1 5 6 dc 21, dc 22 0 1 2 2 DS (From 0,9 to 1) 3 5 TS (From 0,8 to 1) 0,9 1 0,8 1 All thee tak are independent. The total reource for the project are 6. The optimal plan i decribed in figure 3. Fig. 3 Optimal plan 5.4 reference Definition Uing thi optimal plan, the reference i decribed in figure 4:
www.ijcsi.org 206 Fig. 4 Reference 5.5 Modeling the plant ytem Figure 5 illutrate the evolution or depending on the reource. In our cae, the characteritic polynomial i: (12) (13) Step 3: Contruct and expand the modelled characteritic polynomial The modelled characteritic polynomial i (14) Where Fig 5 over time and reource Uing the leat quare regreion method, the plant ytem parameter etimation i: 5.5 Modeling the controller The objective of our deign are: The ytem i table The teady tate error i minimized The ettling time K doe not exceed a contant value 20 Maximum overhoot Mp doe not exceed a contant value 20%. Uing thee objective, [19] dicue in detail the procedure to calculate the appropriate Kp and Ki of the model. In our cae, the tep followed are: Step 1: Calculate r and ɵ uing the following equation (9) (15) K(z) i the tranfer function of the PI Controller in equation (8). G(z) i the tranfer function in equation (5) (16) Expending (14) and eliminating all fraction in the denominator will give u the following polynomial: (17) Step 4: Solve Kp and Ki by reolving the equation (12) = (17). So And (18) (10) (11) In our cae: a=1, b=0,1, r= 0,819 and ɵ= 0,39 Reolving thee two equation will give u: (19) In our cae, K=20 and Mp=0,2 So: r= 0,819 and ɵ= 0,39 Kp=3,3 and Ki=1,56 So our controller i modelled a: Step 2: Calculate the deired characteritic polynomial uing the following equation: (20)
www.ijcsi.org 207 We can ee that the value 4,86 i approximately the mean value of tak reource. If the i le than the reference, the controller will ugget adding thi quantity of reource to begin a pending tak. Thi will accelerate the advancement of the project. The propoed approach will be more efficient if thee condition are met: Project are medium to large (more than 50 tak). Chooing the unit of time the larget poible. In the plant model, chooing an ARMA model with greater order. 6. Concluion and Future work In thi paper, we have propoed a complete approach to control the execution of interoperability improvement project. It baed on proved mathematical model (feedback control theory and tatitic) in addition to the metric. We have modeled the interoperability improvement project a a black box ytem without entering deeply into the relationhip between input (i.e. work effort) and output (). In future work, we will try to model the ytem in more detail. We will work alo on the applicability of other branche of control theory, like optimal control. Reference [1] IEEE. (1990). IEEE tandard computer dictionary: a compilation of IEEE tandard computer gloar. [2] ATHENA (2003), Advanced Technolog for Interoperability of Heterogeneou Enterprie Network and their Application, FP6-2002-IST-1, Integrated Project. [3] INTEROP (2003), Interoperability Reearch for Networked Enterprie Application and Software, network of excellence. [4] Chen, D., Daclin, N. (2006). Framework for enterprie interoperability. In IFAC TC5.3 workhop EI2N, Bordeaux, France. [5] C4ISR (1998), Architecture Working Group (AWG), Level of Information Sytem Interoperability (LISI). [6] IDEAS (2003), IDEAS Project Deliverable (WP1-WP7), Public report, Retrieved from www.idea-roadmap.net. [7] EIF (2004), European Interoperability Framework, White Paper, Bruel, http://www.comptia.org. [8] Ford, T. C., Colomb, J., Grahamr, S. R., Jacque, D. R. (2007, June). A urvey on interoperability meaurement. In Proceeding of 12th International Command and Control Reearch and Technology Sympoium, Newport, RI. [9] Chen, D., Vallepir, B., Daclin, N. (2008). An approach for enterprie interoperability meaurement. In Proceeding of MoDISE-EUS. France. [10] Elmir B., Bounabat, B. (2011). A Novel Approach for Periodic Aement of Buine Proce Interoperability. IJCSI International Journal of Computer Science Iue, 8(4), ISSN (Online): 1694-0814, www.ijcsi.org. [11] Elmir, B., Alrajeh N.A., Bounabat, B. (2011). Interoperability monitoring for e-government ervice delivery baed on enterprie architecture. In International Conference on Information Management and Evaluation (ICIME), Toronto, Canada. [12] WILLIAMS, T.M. (2003). The contribution of mathematical modelling to the practice of project management. IMA Journal of Management Mathematic 14, 3 30. [13] Project Management Intitute (2000). A Guide to the Project Management Body of Knowledge (PMBOK). Project Management Intitute, Upper Darby, PA, US. [14] Gandhi, N. Tilbury, D. M., Diao, Y., Hellertein, J., Parekh, S. (2002). Mimo control of an apache web erver: Modeling and controller deign. In Proceeding of the American Control Conference. [15] Lu, C., Abdelzaher, T., Stankovic, J., Son, S. (2001). A Feedback Control Approach for Guaranteeing Relative Delay in Web Server. In IEEE Real-Time Technology and Application Sympoium, Taipei, Taiwan. [16] Gandhi, N., Parekh, S., Hellertein, J., Tilbury, D. M. (2001). Feedback Control of a Lotu Note Server: Modeling and Control Deign. In American Control Conference, Arlington, VA, USA. [17] Macolo, S. (1999). Claical Control Theory for Congetion Avoidance in High-peed Internet. In Proceeding of the 38th Conference on Deciion & Control, Phoenix, Arizona, US. [18] Chiu, D., Jain, R. (1989). Analyi of the Increae and Decreae Algorithm for Congetion Avoidance in Computer Network. Computer Network and ISDN Sytem, 17(1). [19] Hellertein, J. L., Diao, Y., Parekh, S., Tilbury, D. M. (2004). Feedback Control of Computing Sytem. ISBN 0-471-26637-X, John Wiley & Son.