A Formal Hybrid Analysis Technique for Composite Web Services Verification

Transcription

1 A Formal Hybrid Analysis Tehnique for Composite Web Servies Verifiation MAY HAIDAR 1,2, HICHAM H. HALLAL 1 1 Computer Siene Department / Department of Eletrial Engineering Fahad Bin Sultan University P.O Box Tabuk, SAUDI ARABIA 2 Département d informatique et de la reherhe operationnelle Université de Montréal Pavillon André-Aisenstadt, CP 6128 su Centre-Ville, Montréal QC, H3C 3J7 CANADA {mhaidar,hhallal}@fbsu.edu.sa Abstrat: - In this work, we propose to develop an integrated formal framework where both stati and dynami analysis tehniques omplement eah other in enhaning the verifiation proess of an existing web servies based appliation. The proposed framework onsists of three omponents. A Library of Property Patterns, whih we intend to build on existing work and ompile a library and a lassifiation of web servies properties (patterns and antipatterns). These would inlude BPEL4WS and WISCI requirements in the form of property patterns whih an be instantiated in different ontexts and for different purposes like verifying orretness, seurity, and performane related issues. The property library will be based on an easy to use template that depits mainly the type, formal model, and example of a property. The seond omponent is the development of Stati Analysis Tehniques that inlude diret ode inspetion, abstration based tehniques, and model based tehniques. The third omponent is the development of dynami analysis tehniques that inlude extrating behavioral models of appliations from observed exeutions and verifying them (mainly using model heking) against behavioral properties. These properties speify defets that annot be deteted using stati analysis tehniques. We elaborate in this paper the formal approah used to extrat an automata based model of a web servie omposition from exeution traes that are observed and olleted using a monitoring tool. We also outline the omponents of a prototype to realize the proposed approahes for stati analysis, modeling, and dynami verifiation of the appliations under test. In this paper we present the initial implementation of the dynami approah. Key-Words: - Stati Analysis, Dynami Analysis, Property Patterns, Web Servies, Model Cheking, Automata Models 1 Introdution Businesses are inreasingly adopting servie orientation to shape the arhiteture of their enterprise solutions and to inrease the effiieny of their software appliations. At the foundation of this ever more popular paradigm, web servies are heavily used to enhane deentralization and ross platform and language portability. The power of servies resides mainly in the high degree of dynamism and flexibility they exhibit throughout their lifeyle: publiation, disovery, and binding are all dynami ativities that make a servie an evolving entity apable of adapting to ontinuously hanging and new requirements. In addition, ompositions of servies, whih an also be dynami, have added to the power of servies in building larger enterprise solutions for heterogeneous businesses. However, the fast paed growth of servie implementation and deployment in various ontexts has resulted in a growing gap between the development and verifiation of servie based appliations. On one hand, stati analysis tehniques [1, 13] remain insuffiient to detet behavioral flaws and defets that are exhibited only when servies, espeially omposite ones, are exeuted. In partiular, suh tehniques fae two major problems: diffiulty of generating exeutable ISSN: Volume 2, 2017

2 models that an be used in the analysis, and limited overage of defets that are exhibited only during runtime, e.g., onurreny inurred problems. On the other hand, dynami and runtime tehniques, whih depend mainly on monitoring, an only laim to detet errors and flaws in the observable behavior of a servie, or a dynami omposition of servies. Currently, formal methods have gained momentum as a reliable solution to automate the analysis of various systems. In partiular, software development ommunities are inreasingly adopting formal tehniques to perform different development ativities suh as requirement definition and eluidation, modeling and model transformation, testing, and property verifiation [14,17]. For example, model heking has been used to verify various properties on models of systems. Model heking an be fully automati and produe ounterexamples that point to the violations when a model does not satisfy a given property. Yet, as is the ase for most formal analysis methods, adoption of model heking tools remains relatively limited due mainly to problems like the lak of formal models, the inherent state spae explosion problem, and the lak of proper justifiation for its use espeially for lasses of properties whose verifiation does not explore onurrent behavior of the models Error! Referene soure not found.error! Referene soure not found.. Yet, for many distributed appliations and properties, espeially those speified in terms of events issued by onurrent proesses, the need for model heking beomes lear and outweighs the doubts ast over its use. Formal verifiation tehniques are urrently used in several domains inluding ommuniation systems, software and program analysis [13], and web based appliations [2, 14]. In the ase of omposite Web Servies, the reasoning about the use of model heking is similar. While analyzing simple web servies does not neessarily require the use of model heking tehniques, the use of model heking in the analysis of web servies featuring underlying dynamially omposite servies is learly needed and justified. The latter is speifially true for servies whose omposition is speified through WS-BPEL [20] (Web Servies Business Proess Exeution Language) and WSCI [21] (Web Servies Choreography Interfae). As to the lak of models, espeially in the ase of inaessible ode, analysis in general has been applied to the traes an appliation/system produes when it is used. For this, predefined properties are used to analyze the appliation under test using model heking, when needed, or less omplex tehniques like searh based methods or even manual inspetion. In this paper, we propose to develop an integrated formal framework where both stati and dynami analysis tehniques omplement eah other in enhaning the property testing proess of an existing web servies based appliation. 2 Formal Framework We develop an integrated formal framework as illustrated in Figure 1, where both stati and dynami analysis tehniques omplement eah other in enhaning the verifiation proess of an existing web servies based appliation. The proposed framework onsists of the following main omponents. Library of Formal Speifiations Quality Requirements Library of Patterns / Antipatterns Web Servies Code Formal Speifiations Exeution Traes Automated Formal Models Extration Communiating Automata Based Model Model Cheking Figure 1. Formal Framework for Web Servie Composition Analysis 2.1 Library of Property Patterns Patterns have long been used in the development of software appliations, and servie oriented arhitetures as well, sine they introdue lever and insightful ways to solve ommon problems. Along with patterns, whih are intended to failitate the design and development proesses, the term antipattern is defined. An antipattern is simply a solution to a problem that does not work as intended (in terms of orretness and/or effiieny). Tool Stati and Dynami Modeling Formal Verifiation Property Results: 1. Satisfied 2. Counter Example ISSN: Volume 2, 2017

3 Following the definition, efforts exist to doument antipatterns in atalogs (similar to design patterns) so that they an be avoided. In the proposed framework, we intend to build on existing work [7,8,13,16,17] and ompile a library and a lassifiation of web servies properties (patterns and antipatterns). The lassifiation of properties will be hierarhial: stati/dynami, orretness/funtional, style/performane, et. Suh lassifiation should help developers identify the antipatterns to better avoid them, and testers detet them in the appliation using the appropriate tehniques. On the other hand, doumented properties, whih would inlude BPEL4WS and WISCI requirements in the form of property patterns, an be instantiated in different ontexts and for different purposes like verifying orretness, seurity, and performane related issues. The property library will be based on an easy to use template that depits mainly the type, formal model, and example of a property. For example, in our previous work [16,17], we have defined a pattern template and identified 119 patterns and property speifiation for the verifiation of Web appliations (WAs). Figure 2 shows an example of suh patterns. Eah pattern is speified in Linear Temporal Logi (LTL), whih makes it diretly usable in many model hekers. ID Pattern desription Category Page Attributes LTL Mapping Comments Soure FGS6 Banking information is entered no more than one before submitting form Funtional General Seurity and Authentiation Banking_info: Boolean identifying the presene of fields for banking information Submit: identifiation of page where form submit ation exists PreedeneGlobally (( ( banking_info) W (banking_info W (G (banking_info)))), submit) Newly introdued Figure 2. Example of Web Appliations Speifiation Patterns 2.2 Stati Analysis Approah In general, stati analysis tehniques for software, mainly targeting (ompiled) ode and/or existing speifiations or textual desriptions, are independent of speifi input data sets or individual exeution paths. They are usually lassified into: 1. Diret ode inspetion, where suspiious ode segments are diretly identified in the ode (through linear sanning for example). 2. Abstration based tehniques, where ode representations (lass diagrams, all graphs, et.) are used to math the exhibition of ertain predefined patterns (or antipatterns). In the ase of web servies based appliations; stati analysis tehniques would be applied to the available douments ontaining the desriptions of individual and omposite servies. In doing this, we follow in the steps of the work in [13]; the main deviation being the ustomization of the antipattern library developed to handle mutlithreaded Java appliations to the ontext of web servies and web servies ompositions. In addition, the library will be extended to over patterns/antipatterns like the one shown in Figure 2. However, some omplex faults annot be deteted with stati analysis approahes or only at a high ost (like deadloks). Moreover, stati analysis tehniques are prone to produing signifiant numbers of false warnings (mainly false positives) while not being able to detet some behavioral errors like in the ase of exeption handling. This justifies the need for the third omponent, a set of dynami analysis tehniques. 2.3 Dynami Analysis Approah Dynami Analysis tehniques have emerged as omplementary to stati analysis tehniques, espeially when onurreny and large arhitetural strutures of appliations make the latter ineffiient and rather inomplete. Dynami analysis tehniques do not neessarily rely on existing speifiations or textual desriptions of the appliations under test. Instead, they are applied to exeutable behavioral models that are derived from the appliation s observed exeutions (traes or logfiles). Suh approah to analysis is partiularly effiient in the ase of web servies based appliations; often haraterized by their readiness to ompose web servies, espeially dynamially. The ommuniation between web servies normally uses Internet protools, suh as HTTP, SMTP, and ISSN: Volume 2, 2017

4 FTP. However, all messages an be strutured aording to SOAP (Simple Objet Aess Protool) whih is a protool speifiation for exhanging strutured information in the implementation of web servies. Although many standards have been introdued to address the problem of web servie omposition, inluding BPEL4WS (Business Proess Exeution Language for Web Servies) and WSCI (Web Servie Choreography Interfae), they address mainly the desription and exeution of workflow speifiations for web servie ompositions. Yet, they are not suffiient to support automated verifiation tehniques based on stati analysis. The proposed tehniques inlude extrating behavioral models of appliations from observed exeutions and verifying them (mainly using model heking) against behavioral properties speifying defets that annot be deteted using stati analysis tehniques. The known tehniques in the field inlude: 1. Offline (postmortem) tehniques, where reorded exeutions of an appliation are stored and later used in modeling and verifying the appliation under test. 2. Online (runtime) tehniques, where an appliation under test is analyzed as the exeutions are generated. In our previous work [14,16], we designed a framework for formal modeling and verifiation of web appliations WAs using the model heker Spin. We interepted HTTP requests/responses that depit the behavior of web appliations and extrated ommuniating automata models translated into Promela (the modeling language of Spin). The properties verified inluded properties of onurrent behavior of WAs featuring multiple displays (windows/frames). We use a similar approah for model extrations from behavioral exeutions of omposite Web servies. The major differene is the availability of multiple traes reording the behavior of different servies in the omposition. The olleted traes from all servies are analyzed and abstrated as ommuniating automata models depiting the behavior of the respetive servies. Eah automaton (inferred from a single trae) depits the behavior of one servie where requests are modeled as events and responses as states. Events will be distinguished as loal and ommon. Common events represent ommuniations among the servies in a omposition. It is important to note that while the dynami approah in this paper relies on model heking, models are derived from the observed behavior of the appliation. Thus, the approah ould be seen as passive testing. Sine results of verifiation ould be ompromised when a WSUT does not meet the assumptions desribed previously, this approah does not eliminate the need for traditional testing and should be onsidered as a omplimentary ativity rather than an alternative. For instane, the approah ould be enhaned by additional testing of the appliation with model heking ounterexamples, in order to verify whether properties are indeed violated. Also, behavioral models derived by this approah enable model based test generation Error! Referene soure not found.error! Referene soure not found.. In this paper, we propose a model heking based approah to the verifiation of web servies omposition whose soure ode is inaessible against user defined properties. The model of the appliation under test is obtained from traes of the web servies exeution while properties of interest relate to both the business logi and ergonomis of the web servies. More speifially, the proposed approah breaks down into the following main steps: 1. Modeling the Web servies omposition in a language aeptable by a hosen model heker. We use Spin Error! Referene soure not found., the open soure model heker that is used in many researh and industrial projets. As desribed earlier, we use the exeution traes of the web servies omposition reorded using a relevant monitoring tool, e.g., a proxy server that is apable of interepting HTTP and SOAP ommuniations. The traes are then onverted into a ommuniating automata model representing the behavior of all the omponents of the web servies based appliation. 2. Speifying properties of interest. These properties an represent both desired and undesired behaviors of the web servies. Properties will be mainly user defined and expressed in the property speifiation language of Spin, LTL. 3. Cheking the obtained model against the given properties. To do so, Spin omputes the omposition of all the omponent automata in the derived model and builds a graph ontaining the global states of the appliation. The graph is then inspeted against the language of a property for ontainment. ISSN: Volume 2, 2017

5 3 Automated Model Extration of Web Servies The purpose of building a formal model for a web servie under test (WSUT) is to verify whether the servie omposition exhibits ertain predefined properties using model heking tehniques. It is assumed in this paper that the properties speified in a temporal logi of a hosen model heker are omposed of atomi propositions and for eah SOAP/HTTP servie request, the value of eah proposition is uniquely determined by the ontent of the servie response. These propositions refer to attributes that are user defined and have to be heked (and of ourse refleted in a model). Attributes an be of various types, for instane: a numerial type to ount the ourrenes of a ertain element, a string type to denote the domain name of a response. To build a formal model of a web servie omposition whose soure ode is aessible, one may use abstration tehniques developed in software reverse engineering following a the stati, white box approah Error! Referene soure not found.error! Referene soure not found. as desribed in the previous setion. However, the soure ode is not always available, or aess to the ode ould breah opyrights or trade serets (espeially when verifiation is performed by a third party). Moreover, a web servie omposition an be written using different languages and even different paradigms whih makes stati analysis diffiult to perform. When the ode is not available for modeling, one an build a formal model following a dynami, blak-box based approah, by exeuting the appliation and using only the observations of an external behavior of the servie omposition Error! Referene soure not found. over a ertain period of time. Verifiation of suh models (resulting from finite trae of an appliation) is alled run-time verifiation Error! Referene soure not found.error! Referene soure not found.. In ase of web servies that rely on the SOAP or HTTP protool onsidered in this work, an observable behavior onsists of requests and responses, assuming that the flow of requests and responses between a lient side and a server in the WSUT is observable. One possible way of ahieving this is to use a proxy server Error! Referene soure not found.. A proxy server monitors the traffi between the lient and the server and reords it in proxy logs. The proxy logs, i.e., traes, ontain the requests for omposing servies and the responses to these requests. In the next setion, we present our approah to derive automata based models from traes of web servies. 3.1 Modeling Approah Figure 3 shows the workflow of the proposed approah. The main omponents are: A Monitoring module. It interepts SOAP/HTTP requests and responses during the navigation of the WSUT performed by the user/rawler. An analysis module. It takes the interepted traes as input and generates an automata model in XML/Promela. This module is realized as a prototype tool whih is desribed in Setion 4. A model heking module, in this ase Spin. It verifies user defined properties against the generated model and produes a ounterexample for eah violated property. Properties to Chek SOAP/HTTP Requests /Responses Graphial User Interfae Formal Speifiations SPIN Model Cheker Monitor / Intereptor SOAP/HTTP Reader Servie Analyzer & Modeler System in XML/PROMELA XML/Promela Generator Property Results: 1. Satisfied 2. Violated Figure 3. Workflow of the approah Communiating Automata Model With this approah, a behavior of a WSUT, alled an exeution session, aka Request/Response Sequene (RRS), is interpreted as a possible sequene of web servies responses intermittent with the orresponding requests. Usually, many of these requests are triggered by the user s ations (liking links, submitting forms), while others an be triggered by the servie itself. 3.2 Exeution Session as Communiating Automata Here, the method for modeling an observed exeution session by a system of ommuniating automata is desribed in general. Further development of the detailed model will be attempted in the future. Given the exeution session, first loal a s 0 f 1 f s 2 1 s 2 2 u 0 u 1 u 2 b f 1 v 0 v 1 f 2 ISSN: Volume 2, 2017

6 exeution sessions that orrespond to the behaviors of the omposition web servies of the WSUT are determined, eah of whih is modeled by an automaton. Automata ommuniate synhronously by rendezvous, exeuting ommon (rendezvous) ations. Suh ommuniation is formalized by the parallel omposition operator on automata. Formally, two ommuniating automata A 1 = < S 1, s 01, 1, T 1 > and A 2 = < S 2, s 02, 2, T 2 > are omposed using the operator. The resulting automaton, denoted A 1 A 2, is a tuple < S, s 0,, T >, where s 0 = (s 01, s 02 ) and s 0 S; = 1 2 ; and S S 1 S 2 and T are the smallest sets whih satisfy the following rules: If (s 1, e, s' 1 ) T 1, e 2, and (s 1, s 2 ) S, then (s' 1, s 2 ) S, and ((s 1, s 2 ), e, (s' 1, s 2 )) T. If (s 2, e, s' 2 ) T 2, e 1, and (s 1, s 2 ) S, then (s 1, s' 2 ) S, and ((s 1, s 2 ), e, (s 1, s' 2 )) T. If (s 1, e, s' 1 ) T 1, (s 2, e, s' 2 ) T 2, and (s 1, s 2 ) S, then (s' 1, s' 2 ) S, and ((s 1, s 2 ), e, (s' 1, s' 2 )) T. The omposition is assoiative and an be applied to finitely many automata. Loal Exeution Sessions. An exeution session represents the behavior of ommuniating servies denoted o 1, o 2,, o k, where o 1 orresponds to the main omposing servie and k is the number of ommuniating servies. Given an exeution session, the number of ommuniating entities k and their relationship are determined and a proedure is used to partition the browsing session into loal exeution session, denoted (RRS 1,, RRS k ). 4 Implementation of the Approah The implementation of the proposed framework inludes the following main tasks: 1. Surveying the literature and ommon praties of various developers of web servies based appliations to ompile a set of most frequently enountered properties (patterns and antipatterns). 2. Formulation of properties in speifiation languages that an be used in both stati and dynami analysis tehniques. 3. Identifying proper stati analysis tehniques for eah lass of properties and evaluating their effiieny and robustness. In partiular, this task inludes identifying the proper abstrations, along with methods to extrat them, to be used in deteting orresponding antipatterns in the ode. 4. Reord exeution traes from the appliations under test. This task inludes studying the instrumentation based and intereption based tehniques. 5. Extrating models from monitored exeutions. This inludes extrating models from ompleted traes and inremental models in the ase of runtime analysis that an be used in known model heking tools. 6. Integrating the ompiled library and developed tools in a user friendly toolset whih masks the details of the underlying analysis tehniques form the users and makes the dissemination of the produed framework easier. The proposed framework is implemented using Spin model heker [18]. The automata models are represented using Promela language and the patterns/antipatterns are represented in LTL. We use the Java Elipse environment for the toolset implementation. The omplete toolset will inlude integrated omponents as follows: a. A library of ompiled patterns/antipatterns translated in LTL. b. Exeution intereption and monitoring: the tool interepts requests and responses of a web servies omposition using an open soure proxy 1. The monitoring module an operate in two modes: online and offline mode. In online mode, the monitor interepts the exeutions and feeds them to the analysis module. In the offline mode, the monitor registers an exeution trae in a log file.. Property based attribute seletion: through the graphial user interfae of the tool, a number of predefined attributes that haraterize web servies are provided. The user seleted attributes are evaluated in eah web servie and are refleted in the automata model. d. Analyzing exeution traes and model generation: the tool parses and analyzes the exeution traes and evaluates the user defined attributes in eah visited page. An internal data struture of the automata model of the web servies omposition is built. The model an be generated either in Promela language or XML- Promela. 1 SOLEX, Web Appliation Testing with Elipse. ISSN: Volume 2, 2017

7 e. Automata model visualization and statistial data: the tool has a model visualization feature. The built model of a web servies omposition an be visualized in two different graphial modes as well as one textual model. In the graphial mode, whih is based on existing Java graph libraries, both single automaton, and ommuniating automata models are visualized, whih an be manipulated by the user. For instane, the user an zoom in/out, pik displayed states and drag them, visualize the ontent of eah state, and optionally show/hide transition labels. Also, the tool provides numerial data about the model, namely the number of proesses (automata), total number of states, total number of transitions. Figure 4 illustrates our initial toolset prototype for the dynami modeling of Web servies. Figure 4. Prototype Tool for Web Servies Monitoring and Modelling 5 Related Work Run time verifiation of software appliations has grown as a major field overing major ativities related to the development of software. At the same time, webbed, and web servie-based, appliations have gained a lot of attention in many researh ativities both in aademia and in the industry given the role suh appliations have in the shaping of today s eonomy based on e-ommere and e- servies. Our related work that is losely onneted to this new proposed work is published in [8,14,16,18]. We have implemented an integrated formal framework for run-time verifiation of web appliations. Results were interesting and we were able to verify properties that ould not be verified using other approahes. Reently, a large body of researh has been produed with a fous on formal modeling of web servies based appliations in order to indue automation in the analysis of the developed appliations against some predefined properties speified from the desription and requirements texts. Derived models are often generated from textual desriptions of appliations (BPEL, BPEL4WS, and WSCI), and an be used mainly to hek stati properties that relate to the struture and ontent of the appliation, usually desribed as a omposition of servies. Examples of suh researh inlude the work of Foster et al. [1,2], whih models BPEL desriptions as Finite State Proess models, whih an be verified against properties that are mainly derived from design speifiations written in UML notations like the Message Sequene Chart (MSC) or ativity diagrams. Properties sought for verifiation inlude mostly semanti failures and diffiulties in providing neessary ompensation handling sequenes that are tough to detet diretly in ommon workflow languages like BPEL. Other attempts have been desribed in the literature as well inluding the work of Breugel and Koshkina [3, 4] who introdue the BPE-alulus to apture ontrol flow in BPEL desriptions and programs. The servie desriptions in the proposed language allow for heking against properties like dead path elimination and ontrol yles. The verifiation, mainly formal model heking, is performed in the toolset Conurreny Workbenh (CWB). However, as disussed in Setion 1, proposed verifiation approahes based mainly on the stati analysis of an existing soure ode, where different types of models like EFA, Promela, and ommuniating FSMs [11, 12] are used, have their limitations and impratialities. Consequently, more efforts are being spent on performing run-time verifiation of web servie appliations based on monitoring and model extration. Also, [5] address the run-time monitoring of funtional harateristis of omposed Web servies, as well as for individual servies [6]. 6 Conlusion In this paper, we proposed an integrated formal framework for the analysis and verifiation of Web servies omposition. We propose a hybrid of both stati and dynami analysis tehniques, whih omplement eah other. We also intend to develop a library of patterns and antipatterns of interesting ISSN: Volume 2, 2017

8 speifiations of web servies. These speifiations will be automatially translatable to a formal speifiation language namely LTL. We presented the formal framework for run-time verifiation of Web servies omposition as well as the extrated automata model. Based on our previous experiene and the initial results obtained in the use of our formal approah for run-time verifiation, we believe that results of this proposed work are promising. Referenes [1] Foster, H. (2008). Tool Support for Safety Analysis of Servie Composition and Deployment Models. Proeedings of the 2008 IEEE International Conferene on Web Servies, pp IEEE Computer Soiety. [2] Foster, H., Uhitel, S., Magee, J., & Kramer, J. (2003). Model-based Verifiation of Web Servie Compositions. Pro. of 18th IEEE International Conferene on Automated Software Engineering, pp Montreal, Canada. [3] Koshkina, M., & van Breugel, F. (2004). Modeling and verifying web servie orhestration by means of the onurreny workbenh. SIGSOFT Software Engineering Notes, 29(5):1-10. ACM. [4] Van Breugel, F., & Koshkina, M. (2005). Dead- Path-Elimination in BPEL4WS. Proeedings of the 5th International Conferene on Appliation of Conurreny to System Design, pp IEEE Computer Soiety. [5] Kallel, S., Char_, A., Dinkelaker, T., Mezini, M., Jmaiel, M.: Speifying and Monitoring Temporal Properties in Web servies Compositions. Proeedings of the 7th IEEE European Conferene on Web Servies (ECOWS). (2009). [6] Simmonds, J., Gan, Y., Chehik, M., Nejati, S., O'Farrell, B., Litani, E., Waterhouse, J.: Runtime Monitoring of Web Servie Conversations. IEEE Transations on Servies Computing. 99, (2009). [7] May Haydar, Sergiy Boroday, Alexandre Petrenko, and Houari Sahraoui. "Properties and Sopes in Web Model Cheking". In Pro. of 20th IEEE/ACM International Conferene on Automated Software Engineering (ASE 05). Long Beah, California, USA, November [8] May Haydar, Sergiy Boroday, Alexandre Petrenko, and Houari Sahraoui. "Propositional Sopes in Lenear Temporal Logi". In Pro. of 5th International Conferene on New Tehnologies of Distributed Systems (NOTERE 05). Gatineau, Quebe, Canada, August [9] Dwyer M, Avrunin GS, Corbett JC. Patterns in Property Speifiations for Finite-state Verifiation. 21st Int. Conferene on Software Engineering, May, [10] X. Fu et al, Analysis of interating BPEL web servies. 13th Int. World Wide Web Conferene, [11] Nakajima, S. (2006, May). Model-Cheking Behavioral Speifiation of BPEL Appliations. Proeedings of the International Workshop on Web Languages and Formal Methods, 2(151): ENTCS. [12] Fu, X., Bultan, T., & Su, J. (2004). Analysis of interating BPEL Web Servies. Proeedings of the 13th International World Wide Web Conferene, pp ACM Press. [13] H. H. Hallal, E. Alikaem, W. P. Tunney, S. Boroday, A. Petrenko,(2004) "Antipattern- Based Detetion of Defiienies in Java Multithreaded Software," qsi, pp , Quality Software, Fourth International Conferene on (QSIC'04). [14] Haydar, M., Petrenko, A. and Sahraoui, H. (2004) "Formal Verifiation of Web Appliations Modeled by Communiating Automata" In Proeedings of 24th IFIP WG 6.1 IFIP International Conferene on Formal Tehniques for Networked and Distributed Systems (FORTE 2004), pp Madrid, Spain. [LNCS, vol. 3235] [15] Boroday, S., Petrenko, A., Sing, J. and Hallal, H. (2005) "Dynami Analysis of Java Appliations for MultiThreaded Antipatterns" In Proeedings of the Third International Workshops on Dynamis Analysis (WODA 2005). St-Louis, MI, USA. [16] May Haydar. A Formal Framework for Run-Time Verifiation of Web Appliations: An Approah Supported by Sope Extended Linear Temporal Logi. VDM Verlag, Germany, September ISBN: [17] May Haydar, Houari Sahraoui, and Alexandre Petrenko. "Speifiation Patterns for Formal Web Verifiation". In Pro. of 8th International Conferene on Web Engineering (ICWE 08). Yorktown Heights, New York, USA, July [18] Gerarld Holzmann. The SPIN Model Cheker: Primer and Referene Manual. ISBN- 10: Addison-Wesley, Sptember ISSN: Volume 2, 2017

9 [19] A. Andrews, J. Offutt J, R. Alexander, Testing Web Appliations by Modeling with FSMs, Software Systems and Modeling, 4(3): , July [20] OASIS. (2007). OASIS Web Servies Business Proess Exeution Language. =wsbpel. [21] Arkin, A., Askary, S., Fordin, S., & al. (2002). Web Servie Choreography Interfae (WSCI) 1.0. Retrieved on April 10, 2005 from ISSN: Volume 2, 2017