INES IEEE 18 th International Conference on Intelligent Engineering Systems PROCEEDINGS. July 3 5, 2014 Tihany, Hungary

Transcription

1 INES 2014 IEEE 18 th International Conference on Intelligent Engineering Systems PROCEEDINGS July 3 5, 2014 Tihany, Hungary

2 Organized by Sponsored by IEEE Hungary Section IEEE Computational Intelligence Chapter, Hungary IEEE Joint Chapter of IES and RAS, Hungary IEEE SMC Chapter, Hungary Technical Co-sponsors Venue IEEE Industrial Electronics Society Hotel Club Tihany, Hungary Hungarian Fuzzy Association IEEE Catalog Number: CFP14IES-USB (pendrive); ISBN: (pendrive) Copyright and Reprint Permission: Abstracting is permitted with credit to the source. Libraries are permitted to photocopy beyond the limit of U.S. copyright law for private use of patrons those articles in this volume that carry a code at the bottom of the first page, provided the per-copy fee indicated in the code is paid through Copyright Clearance Center, 222 Rosewood Drive, Danvers, MA For reprint or republication permission, to IEEE Copyrights Manager at pubs-permissions@ieee.org. All rights reserved. Copyright 2014 by IEEE.

3 Table of Contents General information... 7 Welcome... 8 Committees... 9 Decision Support System based on Subjective Criteria...11 Hamido Fujita Iwate Prefectural University, Japan Intelligent Analysis and Trading of the Financial Markets...13 William A. Gruver Simon Fraser University, Canada Approach for Enterprise Search and Interoperability using Lightweight Semantic...15 Martin Šeleng*, Michal Laclavík*, Štefan Dlugolinský*, Marek Ciglan*, Martin Tomašek** and Ladislav Hluchý* * Institute of Informatics, Slovak Academy of Sciences, Bratislava, Slovakia; ** InterSoft, a.s., Košice, Slovakia General Regression in Cloud Computing...21 Martin Bobák, Ladislav Hluchý and Viet Tran Institute of Informatics of the Slovak Academy of Sciences, Bratislava, Slovakia Validation of Parameter Importance by Regression...27 P. Krammer, M. Kvassay, L. Hluchý Institute of Informatics, Slovak Academy of Sciences, Bratislava, Slovakia Flickr Trace Summarization: from Images to Imaginary...33 Laurent Daireaux*, Vincent Mallard*, Martin Tardy*, Maxime Wasselin*, Bérengère Branchet*, Gaël Chareyron*+, Jérôme Da Rugna*+ *Pôle Universitaire Léonard de Vinci, Paris La Défense, France; +University of Paris 1 Pantheon-Sorbonne, Paris, France Development of LabView-based HVAC Models for Net-zero Energy Strategies Implementation...37 Csaba Szász Technical University of Cluj, Cluj, Romania Implementation of Polycarbonate Material Mechanical Properties of Rapid Prototyping into System Creo, Laboratory Verification of the Results...43 J. Lipina, P. Kopec and V. Krys Technical University of Ostrava, Ostrava, Czech Republic Three Approaches for the Detection of CAD-Part Attributes as a Preparation for an Automatic Classification...49 R. Roj and H.-B. Woyand University of Wuppertal, Wuppertal, Germany Rapid Development of Application-oriented Natural Language Interfaces...55 Tamás Mészáros and Tadeusz Dobrowiecki Budapest University of Technology and Economics, Budapest, Hungary A Simple Fast Fourier Transformation Algorithm to Microcontrollers and Mini Computers...61 Jozsef Suto*, Stefan Oniga*, Gyula Hegyesi** *University of Debrecen, Debrecen, Hungary; **Hungarian Academy of Science, Institute of Nuclear Research, Debrecen, Hungary 3

4 Autonomic Monitoring Approach Based on CEP and ML for Logistic of Sensitive Goods...67 Septimiu Nechifor*, Bogdan Târnaucă*, Lucian Sasu*, Joachim Teutsch**, Walter Waterfeld**, Dan Puiu, Anca Petrescu***, Florin Moldoveanu**** *Siemens Corporate Technology Romania; **Software AG, Germany; ***Siemens Corporate Technology Romania; ****Transilvania University of Brasov, Romania Nonlinear Order-Reduced Adaptive Controller for a DC Motor Driven Electric Cart...73 József K. Tar, Tamás Haidegger, Levente Kovács, Krisztián Kósi, Balázs Botka, Imre J. Rudas A New Approach to the Use of Parametric Method of Robot Motion Planning...79 János Somló An Ontology-based Model for Vehicular Ad-hoc Networks...83 Adrian Groza, Anca Marginean, Vlad Muresan Technical University of Cluj-Napoca, Cluj-Napoca, Romania Survey on the Control of Time Delay Teleoperation Systems...89 Teréz A. Várkonyi*+, Imre J. Rudas*, Péter Pausits*, Tamás Haidegger* * ; + Universitá degli Studi di Milano, Italy Basic Approaches in Adaptive Control System Design for Small Turbo-compressor Engines...95 R. Andoga, L. Főző, L. Madarász, J. Považan and J. Judičák Technical University of Košice, Košice, Slovakia FPGA Based Hardware Implementation of a Self-Organizing Map S. T. Brassai Sapientia Hungarian University of Transylvania, Tîrgu Mureş, Romania Malware Categorization and Recognition Problem Liberios Vokorokos, Ján Hurtuk, Branislav Madoš TU of Košice, Slovak Republic New Method for Behavior Driven Product Concept Definition László Horváth and Imre J. Rudas Performance Evaluation of a Service Availability Model Tayyaba Anees, Heimo Zeilinger Vienna University of Technology, Vienna, Austria Model-based Estimation of Physiological Parameters in the Reperfusion Phase of Liver Transplantation József Homlok*, J. Geoffrey Chase**, Tibor Doktor*, Zoltán Benyó*, Balázs Benyó* *Budapest University of Technology and Economics, Hungary; **University of Canterbury, Christchurch, New Zealand Precision of 3D Body Scanners P. Elbrecht*, K. J. Palm** *Tallinn University of Technology, Tallinn, Estonia; **Incognito Ballistic Llc, Tallinn, Estonia Multidimensional Scaling Analysis of Laboratory Parameters According to Obesity Krisztina Némethy, Rita Ősz, József Gáti Experiments for Real Practice of Óbuda University Franciska Hegyesi, Gyula Kártyás 4

5 Database Schema Design for Supporting Sport Activity Monitoring Edit Tóth-Laufer Fuzzy Model-Based Cutting Parameter Combination Optimization R. Horváth*,**, E. Tóth-Laufer ** * ; ** Budapest University of Technology and Economics, Budapest, Hungary Separation Enhanced Nucleus Detection on Propidium Iodide-stained Digital Slides V. Z. Jonas*, M. Kozlovszky*, B. Molnar** * ; ** Semmelweis University, Budapest, Hungary Fuzzy Expert System for Automatic Wavelet Shrinkage Procedure Selection for Noise Suppression Adrienn Dineva, Annamária R. Várkonyi-Kóczy, József K. Tar Application of Fuzzy Logic in Hemodialysis Equipment József Klespitz, Márta Takács, Levente Kovács Example for Convex Hull Tightening Increasing the Feasible Parameter Region at Linear Matrix Inequality-based Control Design Alexandra Szöllősi*, Péter Baranyi**, Péter Várlaki*** *Budapest University of Technology and Economics; **3D Internet based Control and Communications Laboratory, Computer and Automation Research Institute, Hungarian Academy of Sciences; ***Széchenyi István University An Overview of Research Trends in CogInfoCom Peter Baranyi*, Adam Csapo*+ Peter Varlaki+ *Institute for Computer Science and Control, Hungarian Academy of Sciences; +Széchenyi István University, Győr Minimal Volume Simplex (MVS) Approach for Convex Hull Generation in TP Model Transformation József Kuti*, Péter Galambos*+, Péter Baranyi* *Institute for Computer Science and Control, Hungarian Academy of Sciences, Budapest, Hungary; + A System Model and Applications for Intelligent Campuses Attila Adamkó and Lajos Kollár University of Debrecen, Debrecen, Hungary Beyond Mashups: Graph Transformations on Web Data Gábor Imre and Gergely Mezei Budapest University of Technology and Economics, Budapest, Hungary Developing Rapid Prototype-Capable Applications for Industrial Mobile Robot Platforms Claudiu Radu Pozna*,**, Ernő Horváth*, János Kovács* * Széchenyi István University, Győr, Hungary; ** Transylvania University, Brasov, Romania Multimodal Aspects of Communication and Data Storage and Managing Systems J. Sebestyénová, P. Kurdel Institute of Informatics Slovak Academy of Sciences, Bratislava, Slovakia Stacked Generalization for Scene Analysis and Object Recognition Lorenzo Peppoloni, Massimo Satler, Emanuel Luchetti, Carlo Alberto Avizzano and Paolo Tripicchio TeCIP Institute, Scuola Superiore Sant Anna, Italy BlockImpress Viktor László Takács University of Debrecen, Debrecen, Hungary 5

6 MasterBroker: REST-oriented Service Broker Dijana Kosmajac*, Vladimir Vujović*, Mirjana Maksimović*, Nikola Davidović*, Branko Perišić** * University of East Sarajevo, East Sarajevo, Bosnia and Herzegovina; **University of Novi Sad, Novi Sad, Serbia Sirius: A Rapid Development of DSM Graphical Editor Vladimir Vujović*, Mirjana Maksimović* and Branko Perišić** * Faculty of Electrical Engineering, East Sarajevo, Bosnia and Herzegovina; ** Faculty of Technical Sciences, Novi Sad, Serbia Improving Degradation Prediction Models for Failure Analysis in Topside Piping: A Neuro-Fuzzy Approach A. M. N. D. B. Seneviratne and R. M. Chandima Ratnayake University of Stavanger, Norway Modified Clustering Algorithm for Projective ART Neural Network Roman Krakovsky*, Radoslav Forgac**, Igor Mokris** * Catholic University, Ruzomberok, Slovakia; ** Institute of Informatics Slovak Academy of Sciences, Bratislava, Slovakia Performance Evaluation of a Live, Crowdsensing Based Transit Feed Service Architecture Károly Farkas, Róbert Szabó and Bernát Wiandt Budapest University of Technology and Economics, Budapest, Hungary On Practical Constraints of Approximation Using Neural Networks on Current Digital Computers Michal Puheim, Ladislav Nyulászi, Ladislav Madarász, Vladimír Gašpar Technical University of Kosice, Košice, Slovak Republic Quantum Fuzzy Inference Gate Design in Robust Intelligent Control of Robotics and Mechatronics S. V. Ulyanov*, A. G. Reshetnikov*, A. V. Nikolaeva*, I. A. Barchatova* and V. A. Albu** * International University Dubna / Institute of System Analysis and Control, Dubna, Moscow Region, Russia; ** Institute of Mathematics and Computer Science, Chisinau, Republic of Moldova Authors Index

7 General Information Date and Place INES 2014 will take place on July 3-5, 2014, at Hotel Club Tihany, Hungary. Official Language The official language of the conference is English. All presentations, including discussions and submissions, must be made in the official language. No translation will be provided. Proceedings Each accepted paper reaching us in time will be published in pendrive proceedings, and it will be distributed at the registration desk. Conference organizers do not mail the proceedings to authors who do not show up. At least one author of each paper must attend the conference. Opening Hours of the Registration Desk July 3, Thursday: 9:00 15:00 July 4, Friday: 9:00 15:00 Presentation Presentations can be made by using data projector. All authors are kindly asked to take their presentation on CD or USB. To present the paper it is not allowed to use own computer. All conference rooms are supplied with data projector with PC. Smoking Please, be so kind to your lungs and your colleagues by not smoking during the sessions and social events. 7

8 WELCOME MESSAGE FROM THE GENERAL CHAIRS On behalf of INES 2014 Committees, it is my pleasure to welcome you to the 18 th International Conference on Intelligent Engineering Systems (INES) to be held in Tihany, Hungary. The growing international competition in the industrial arena has created a demand for the introduction of intelligent techniques to various industrial problems to improve product quality and production process efficiency, as well as reduce production costs. The aim of the INES conference series is to provide researchers and practitioners from industry and academia with a platform to report on recent developments in the area of computational intelligence. INES 2014 focuses on the application of state-of-the-art intelligent techniques to engineering systems. The conference location is Hotel Club Tihany by the beautiful coast of lake Balaton. I would like to acknowledge the efforts of the Technical Program Chairman, the Organizing Committee Chairman, the Technical Program Committee members and all those persons responsible for the background activities from local arrangements to conference secretariat. I also want to thank many volunteers who have contributed lots of time and effort to bring INES 2014 to you. My pleasant duty is to gratefully acknowledge the support provided by the sponsors of the conference: IEEE Industrial Electronics Society, IEEE Hungary Section, IEEE Joint Chapter of IES and RAS, Hungary, IEEE Computational Intelligence Chapter, Hungary, IEEE SMC Chapter, Hungary,, and Hungarian Fuzzy Association. I hope that all in attendance at INES 2014 will find this event intellectually stimulating and professionally rewarding. János Fodor INES 2014 General Chair 8

9 Committees INES FOUNDING HONORARY CHAIR Imre J. Rudas, INES 2014 HONORARY CHAIRS Lotfi A. Zadeh, USA Bogdan M. Wilamowski, Auburn University, AL, USA INES 2014 GENERAL CHAIR János Fodor, INES 2014 TECHNICAL PROGRAM COMMITTEE CHAIRS Tamás Haidegger, Levente Kovács, INES 2014 TECHNICAL PROGRAM COMMITTEE Rudolf Andoga, Technical University of Košice, Slovakia Valentina Balas, "Aurel Vlaicu" University, Arad, Romania Ildar Batyrshin, Mexican Petroleum Institute, Mexico Barnabás Bede, Attila L. Bencsik, Balázs Benyó, BME, Hungary Tom Gedeon, Australian National University, Australia Gerhard Hancke, University of Pretoria, South Africa Gábor Hegedűs, Kaoru Hirota, Tokyo Inst. of Tech., Japan László Horváth, Karel Jezernik, University of Maribor, Slovenia Zsolt Csaba Johanyák, Kecskemét College, Hungary Janusz Kacprzyk, Polish Academy of Sciences, Poland Oussama Khatib, Stanford University, USA László T. Kóczy, Széchenyi István University, Győr, Hungary George Kovács, CAI of HAS, Hungary Levente Kovács, Szilveszter Kovács, University of Miskolc, Hungary Krzysztof Kozlowski, University of Poznan, Poland Krisztián Lamár, Jose Antonio Tenreiro Machado, Institute of Engineering of Porto, Portugal Ladislav Madarász, Technical University of Košice, Slovakia Rita Ősz, Endre Pap, University of Novi Sad, Serbia Béla Pátkai, Tampere University of Technology, Finland Emil M. Petriu, University of Ottawa, Canada Jozef Považan, Technical University of Košice, Slovakia Radu-Emil Precup, Politehnica University of Timisoara, Romania Stefan Preitl, Politehnica University of Timisoara, Romania Octavian Prostean, Politehnica University of Timisoara, Romania János Somló, BME, Hungary 9

10 Sándor Szénási, József K. Tar, Annamária R. Várkonyi-Kóczy, Jozef Živčák, Technical University of Košice, Slovakia INES 2014 ORGANIZING COMMITTEE CHAIR József Gáti, INES 2014 ORGANIZING COMMITTEE Gyula Kártyás, Árpád Takács, Teréz Várkonyi, Péter Zentay, INES SERIES LIFE SECRETARY GENERAL Anikó Szakál PROCEEDINGS EDITOR Anikó Szakál PRODUCTION PUBLISHER IEEE Hungary Section 10

11 INES 2014 IEEE 18th International Conference on Intelligent Engineering Systems July 3-5, 2014, Tihany, Hungary Decision Support System based on Subjective Criteria Hamido Fujita Director of Intelligent Software Systems Iwate Prefectural University, Japan Editor-in-Chief: Knowledge-Based Systems PLENARY TALK Abstract Decision Support Systems (DSS) are important research field that overlapping different research paradigm in intelligent system granular computing, soft set and multicriteria decision making. Due to the characteristics nature of criteria and parameters, these are represented using fuzzy set (linguistics or crisp values), or fuzzy soft set, or fuzzy interval or else. Different presentations have reported and contributed to establish a wide variety of solution and approaches in providing best selection to alternatives in relation to these defined criteria. The common among these theoretical and practical approaches is that; Decision Makers (DMs) should provide values of these objective criteria based on their experience and else. In order to provide feasible aggregation functions or operators that rank set of alternatives based on expert decision preferences (i.e., decision matrix) we need to assume that these defined criteria are independent. These approaches are additive decision preferences (like TOPISIS) or multiplicative preferences to characterize DMs preferences and relative weight of attributes. Researchers have provided a solution by introducing support function among DMs; providing arithmetic means solution to define the group decision matrix. However, even in this case, the weights given by expert are static, and are not allowed to be changed. These are making the DSS not very applicable in real-practices, like DSS for medical applications in which criteria may have subjective change to patient preferences. The subjectivity of criteria and dependency among them, as well as the mental projection in relation to system users or decision makers, would all contribute to add additional mechanism on DSS. In this I am highlighting (theoretically and practically) these hot research challenge through a project we called it as Virtual Doctor System, which is DSS for medical diagnosis based on subjective and metal analysis of medical doctor (expert) and patients as users /14/$ IEEE 11

12

13 INES 2014 IEEE 18th International Conference on Intelligent Engineering Systems July 3-5, 2014, Tihany, Hungary Intelligent Analysis and Trading of the Financial Markets William A. Gruver, PhD, FIEEE Professor Emeritus, Simon Fraser University PLENARY TALK Abstract Although the U.S. financial markets have experienced a massive rebound since the crash of 2008, many people believe that the world markets are overbought and due for a major correction. Some even claim that the markets are rigged. Nevertheless, it is acknowledged that the change from open outcry pits to fully electronic exchanges, and the rise of the high speed electronic trading systems, requires a new approach to trading and investment decisions. The purpose of this presentation is to provide an overview to the practical aspects of data analysis and trading of the financial markets. Topics include market moving events, market psychology, high frequency trading (HFT) and dark pools, supply/demand levels, technical indicators and odds enhancers, and examples of swing trading and macro-trend trading. Examples will be presented from the live markets to illustrate the techniques being presented. All examples will be taken from data on the New York and Nasdaq stock exchanges, however, the methodology is applicable to the analysis of markets worldwide /14/$ IEEE 13

14

15 INES 2014 IEEE 18th International Conference on Intelligent Engineering Systems July 3-5, 2014, Tihany, Hungary Performance Evaluation of a Service Availability Model Tayyaba Anees, Heimo Zeilinger Institute of Computer Technology Vienna University of Technology Gußhausstraße 27-29, 1040 Vienna, Austria {anees, zeilinger}@ict.tuwien.ac.at Abstract Adoption of Service-oriented architecture (SOA) is increasing due to easy service integration and expected cost reductions due to reusability. Nowadays, safety critical, telecommunication and business systems are using SOA. SOA based systems still face challenges related to performance and quality of service such as availability. These challenges are due to complexity, distributed services and XML-based message processing. Response time is used to analyze performance of SOA applications and its analysis for web services is required. The solutions for increasing service availability can impose performance overhead as extra resources are required. We have proposed a SOA based model for increasing web services availability, which uses monitoring. In this paper, we have analyzed impact of monitoring on performance of web services in both LAN and WAN environments. We experimentally prove that monitoring can be used to increase service availability. Acceptable service loads and response time can be estimated based on our model. Keywords- performance; service-oriented architecture;service availability; monitoring; safety critical systems; business systems; telecommunication systems I. INTRODUCTION Nowadays, SOA adoption is increasingly seen in diverse areas of application such as safety critical, telecommunication and business systems. The reason for the wide-scale adoptability of SOA is reusability, loose coupling and reduced costs. SOA enables easy integration of distributed services and legacy systems to newly build systems due to which costs are reduced in SOA based systems. SOA based systems can be trusted for quality by service consumers due to standard based development approach. Considering quality of service, for service consumers, system response time and jitter are the most important concerns. In case of safety critical systems, unresponsive services can cause safety hazards and human lives can be endangered. Service-oriented environments are confronted with performance problems due to distributed services, middleware based solutions and XML based message processing. Due to distribution of services in SOA, network conditions can vary between service providers and service consumers, which can decrease performance and availability of services. In SOA environments, a common approach is introducing a middleware to the communication process also known as Enterprise Service Bus [1]. An ESB facilitates routing and transformation of messages and manages the orchestration mechanism in SOA for service compositions. Despite, the advantages of a middleware, disadvantages can be expected as well: performance overhead and functionality issues to name only two. Both reduce service availability and therefore represent critical parameter. Furthermore, SOA environments use XML based messaging format that comes with performance overhead [13]. In addition service availability is not a characteristic typically addressed in SOA and therefore must be investigated in more detail. Though, mechanisms which are used to increase service availability may also impose a performance overhead. Anees et al. proposed a SOA based model, which can be used to increase web services availability [2]. In the current paper, we focus on analyzing the performance impact of monitoring used in our model to increase service availability. Monitoring mechanisms are commonly included to at least increase the reliability of communication solutions. However, solutions like group communication systems, or clustered systems vary and monitoring mechanisms must differ with respect to system architectures and processes. In this work the availability model mainly bases on the system response time as performance indicator which is analyzed in LAN and WAN environments. For increasing service availability, a monitoring service has been added based on a heartbeat mechanism for failure detection, instead of timeouts used by service consumers, for improving the process of failure detection and accuracy. The system is based on synchronous web services integrating the minimum set of participants in a service-oriented environment: service consumers, service providers and service registry. The work focuses on atomic services, respectively the performance of atomic web services whose reusability, in the authors opinion is higher than of service compositions. In case the availability of atomic services is increased, it will consequently increase availability of service compositions.. The remainder of the paper is organized in Sections. Section II describes the state of the art research work. In Section III we discuss our proposed SOA model for improving service availability and performance. Section IV presents results and discussion. Section V concludes the paper /14/$ IEEE 115

16 T. Anees and H. Zeilinger Performance Evaluation of a Service Availability Model II. STATE OF THE ART The tendency of SOA adoption by safety critical, telecommunication and business systems is increasing due to reusability, improved integration and cost reductions. For instance, a SOA approach has been selected as an architectural solution for a renewed system of the Finnish Defense Forces [3] including command, control, surveillance and reconnaissance. Telecommunication networks are service centric [4]. The integration of SOA based applications reduces the dependency of different types of IT systems, decreasing the cost of system maintenance [5] and increasing the agility [6], [7]. In authors opinion, the standard based development approach in SOA makes it more acceptable to service consumers. If the service is based on standards, there is a high probability that a service is subscribed by a higher number of consumers [8]. Despite the advantages offered by SOA, performance monitoring of SOA based solutions is challenging and a complex task [9]. The SOA approach can have a negative impact on the performance of an application due to physical distribution of services, overhead caused by intermediaries that handle communication and due to use of standard messaging formats [10] that are not designed specifically for one application. Communication failures, routing errors and network traffic congestion must be addressed as the performance of service oriented systems are commonly characterized by instability [11] [12]. Service-oriented systems generally base on web services. XML-based SOAP Web Services are wide spread technology, which is currently adapted in Service Oriented Architectures, cloud interfaces, management of federated identities, egovernment, or military services [14]. Web services have features such as interoperability, decoupling and just-in-time integration [15]. In the authors opinion, performance evaluation of web services is required and response time can be used for performance analysis. Web service based application providers will face challenging features related to performance [16]. Response time, which for this work is interpreted as the elapsed time between sending request and receiving the response, is an effective performance indicator in a SOA implementation [17]. A high number of concurrent requests results in a decrease of service availability for the customer and performance. Mechanisms applied to increase service availability include machine virtualization [18], clusters [19] and group communication systems [20]. These solutions use monitoring mechanism to increase availability but the solutions can vary by their implementation. Availability of web services have become a great concern in SOA and monitoring mechanisms are deployed to detect and recover failures for service compositions [21]. While monitoring mechanism could improve availability to some extent, they may cost more resources and increase the response time perceived by end users. In this work monitoring is applied and the system wide response time is analyzed to predict acceptable service loads and performance. III. MODIFIED SOA MODEL AND PERFORMANCE METRICS In this Section, SOA model for increasing service availability is proposed. In addition, performance and response time with respect to the model are discussed. A. Modified SOAModel for Service Provider Availability Easy integration, reusability and a standard-based development approach has made service-oriented systems widely acceptable. SOA provides benefits but face challenges related to service availability, performance and safety in SOA. A SOA based model is proposed based on synchronous web services. Figure 1, shows the basic participants in a service-oriented environment, which are service consumers, service providers and service registry. Figure 1. Basic SOA model In Figure 1, the interaction between service consumers and service providers is explained. Service providers publish (1.publish) the description of their services into a service registry. Service consumers find (2.find) the service from the service registry. Service consumers bind (3.bind) to service provider. Service consumers and service provider start to interact (4.interact) with each other. The basic SOA model shown in Figure 1 is changed by adding a monitoring service (see Figure 2). Service providers are realized in a redundant way and send heartbeat messages to the monitoring service. In Figure 2, for simplification, only one redundant service provider as well as heartbeats from only the primary service provider are shown. Also, all service providers publish their services in the service registry and the failure detection process is similar for redundant service providers. A monitoring service detects the failure of the service provider on the basis of 3 consecutive missed heartbeats. Monitoring service detects failure, notifies failure to service consumers and initiates the failover process. The duration of the failover process includes the failure detection time and the recovery time [22]. This work mainly concentrates on optimizing the failure detection. The monitoring service deletes the information of the failed service provider from the service registry so that failed service provider information is not distributed to more service consumers. Monitoring service restarts the failed service provider to make it available again. 116

17 INES 2014 IEEE 18th International Conference on Intelligent Engineering Systems July 3-5, 2014, Tihany, Hungary A middleware component was not applied to the model as it can impose additional performance overhead. Furthermore a UDDI based service registry is used. The focus of the approach is reducing failure detection time through monitoring and heartbeat mechanism. decided on the basis of consecutive missed heartbeats, generally three missed heartbeats for accurate failure detection. Using heartbeat mechanisms, non-failure states of service provider can be correctly identified. In case of timeouts, failure is ensured only once and it is possible that service provider is available but it is unable to send response within a specified timeout interval. In heartbeat mechanisms, failure is ensured several times, so non-failure states can be easily identified. For instance, in our approach service providers send heartbeats at periodic interval of 100 ms and failure is decided on 3 consecutive missed heartbeats. There will be 3 attempts from service provider to send heartbeats. If 2 heartbeats are missed and a third heartbeat is received, a non-failure state can be identified and failure is not assumed by checking once as is the case with the approach of timeouts. IV. EXPERIMENTAL RESULTS AND DISCUSSION The following experiments have been conducted to evaluate the performance of the proposed web services availability model [2]. Figure 2. Modified SOA model In the following, the performance impact of monitoring on response time of web services with various service loads is analyzed. The proposed model bases on synchronous web services. Thereby a timely response is required in order to avoid the service to be considered as unavailable. Hence, service consumers can use timeouts for failure detection. If a response is not retrieved within a specified timeout, the service provider is assumed to be failed and unavailable. The critical parameter is the timeout period used by the service consumer retrying to contact the provider. In case it is too large, the failover time increases, perhaps up to an unacceptable limit. On the other hand, lowering the timeout period may reduce the chance to contact the required service provider. This results into a failover which may be less efficient than staying with the original service provider. Too short timeout periods can result in false failure detections. Understanding the uncertainty arising in web services is crucial for choosing the right recovery techniques, setting timeouts, and adopting system architecture and its behavior to a changing environment such as the Internet and SOA [11]. An appropriate timeout value selection is difficult for web services as they rely on the network and at the same time, it is essential for accurate and efficient failure detection. Heartbeat mechanisms are used in the proposed approach for failure detection instead of timeouts. The service provider send heartbeat messages to a dedicated monitoring service which manages service failures. Of course, as the service consumer evaluates the failure possibility wrong failure detections are possible. In heartbeat mechanism, failures are Figure 3. Implementation environment for SOA model The test environment for evaluating the proposed model is shown in Figure 3. WANem [24] is used for emulating the model in LAN and WAN environments under different network conditions. All traffic between service consumers, service providers and monitoring service passes through the emulator. Service consumers send requests in parallel to service providers and reside on one node. Monitoring service is placed on a separate node. Service consumers and service providers are deployed on separate node on Glassfish server [25]. We have used juddi [26] service registry and we have used Jakarta-Tomcat server [27] for registry access. We have used mysql server [28] for storing registry information. We 117

18 T. Anees and H. Zeilinger Performance Evaluation of a Service Availability Model have used synchronous web services in implementation. We have used four nodes in test setup. In the first experiment, the impact of redundancy and number of service consumers on response time in a LAN and WAN environment have been analyzed. As proposed, the response time represents the performance indicator in this case. The number of service consumers differs in order to see the correlation to the response time. Results in Figure 4, indicate a predictable system behavior for response time where an acceptable service load can be identified. For instance, results indicate that using our model 300 service consumer s can send requests in parallel to service provider with 100 ms response time, which should be acceptable to service consumers in most of the cases. Above 300 service consumers there are fluctuations in response time due to garbage collection as our implementation is based on Java. In the second experiment, we have taken measurements to see the impact of redundancy and variable number of service consumers on response time in a WAN environment. We have used same parameters in measurements as used in LAN environment, which can reduce service provider availability and performance such as different number of service consumers and different number of service providers. Results in Figure 6, indicate that by adding service consumers, response time increases. Results show that response time in WAN is similar to LAN but number of service consumers are significantly reduced irrespective of applied redundancy. Results show that performance is not decreased by adding redundancy. Results indicate that acceptable service load is similar in WAN as LAN and 300 service consumers can send requests in parallel to service provider. Results show that with 10 service consumers performance is better than 1 service consumer. In author s opinion, it is due to network delay and as synchronous web services are used. Due to synchronous communication, service consumer can send requests one after the other. Figure 4. Average response time in LAN with different redundancies Results in Figure 5, show the average request rates per second for the first experiment. Results indicate that request rates go down by adding more service consumers and also by adding redundancy. Results indicate that service provider can handle less service requests with more service consumers. Request rates with different redundancies are though similar. Figure 6. Average response time in WAN with different redundancies With 10 service consumers, 10 requests are sent at same time approximately and with 1 service consumer 1 request is sent at same time. After sending requests service consumer wait for response. 1 service consumer send 1 requests and waits for sending next request whereas 10 service consumers send 10 requests and wait for sending next requests. Results in Figure 7 show that request rates in WAN are significantly reduced in comparison to LAN. Performance is similar in WAN as LAN irrespective of redundancy. Results indicate that as service provider handles less number of requests in WAN, response time stays similar to LAN. It can be concluded that performance or response time stays similar in LAN and WAN environments. However, in LAN more service consumers can be supported in comparison to WAN. Redundancy can be used according to requirement. For, instance systems which expect high failure rate can add more redundancy and if failures are infrequent, then redundancy should not be applied. Figure 5. Average request rate in LAN with different redundancies 118

19 INES 2014 IEEE 18th International Conference on Intelligent Engineering Systems July 3-5, 2014, Tihany, Hungary REFERENCES Figure 7. Average request rate in WAN with different redundancies V. CONCLUSION Wide acceptability of service-oriented architecture by safety critical, telecommunication and business systems diverts attention towards improving quality of service attributes such as availability and safety of SOA based systems. From service consumers point of view performance and response time are always important. A sluggish service can result in poor end user experience as service can easily be deemed as unavailable. With early recognition of performance bottlenecks for web services, poor end user experience can be avoided and it can reduce system downtime and costs. Performance of SOA based systems is not generally considered good because of service distribution, intermediaries and XML-based message processing. Solutions which are used to increase availability such as clustering, machine virtualization and group communication systems use system resources, which can decrease performance. When availability solutions are applied performance assessment becomes necessary such as for web services. We have proposes a SOA model for web services availability. Availability of services is increased in our model by reducing failure detection time using monitoring and heartbeat mechanism. In this paper, we have analyzed the performance impact of monitoring mechanism regarding service response time. Experimental results show the efficacy of the approach. In the paper, we have experimentally proven that monitoring can be used to increase service availability as acceptable response time can be estimated based on our model to avoid performance bottlenecks. We have emulated our model and tested it in both LAN and WAN environments to bring it close to reality. Experimental analysis indicates that performance or response time in LAN and WAN environments is similar but number of service consumers and service requests in WAN are significantly reduced in comparison to LAN. [1] L. Garces-Erice, Building an enterprise service bus for real-time SOA: A messaging middleware stack, IEEE International Computer Software and Applications Conference (COMPSAC), United States, Jul. 2009, pp [2] T. Anees, and H. Zeilinger, SOA Model for High Availability of Services, IARIA Conference on Communication Theory, Reliability, and Quality of Service, Italy, Apr. 2013, pp [3] J. Jormakka, and J. Lucenius, Possibilities for improving dependability of c4i2sr systems based on service-oriented architecture, Computation World: Future Computing, Service Computation, Cognitive, Adaptive, Content, Patterns, Athens, Nov 2009, pp [4] R. Levenshteyn, J. Niemöller, E. Freiter, K. Vandikas, R. Quinet, and I. Fikouras, Multi-technology service composition for the telecommunication domain - concepts and experiences, IEEE International Conference on Next Generation Mobile Applications, Services and Technologies, Amman, Jul. 2010, pp [5] X. Zhang, and X. Yin, Research of B2B E- business application & development technology based on Web services, IEEE International Conference on Computer Science and Information Technology(ICCSIT), China, Aug 2009, pp [6] P. Bhallamudi, and S. Tilley, SOA migration case studies and lessons learned, IEEE International Systems Conference (SysCon),Canada, Apr. 2011, pp [7] Y. Chee, L. Chen, N. Zhou, and L. Zhang, Legacy asset analysis and integration in model-driven SOA solution, IEEE International Conference on Services Computing, Miami, Jul. 2010, pp [8] S. W. Choi, and S. D. Kim, A quality model for evaluating reusability of services in SOA, IEEE International Conference on E-Commerce Technology and the Fifth IEEE Conference on Enterprise Computing, E-Commerce and E-Services, United States, Jul. 2008, pp [9] G. Trlin, and I. Zoraja, Online SOA chain performance monitoring in mixed environments, IEEE International Conference on Software, Telecommunications and Computer Networks (SoftCOM), Split, Sep. 2012, pp.1-5. [10] S. H. Oh, J. S. Her, S. W. Choi, and S. D. Kim, A framework for measuring performance in service-oriented architecture, IEEE International Conference on Next Generation Web Services Practices (NWeSP), Seoul, Oct 2007, pp [11] V. Kharchenko, S. Mamutov, O. Tarasyuk, C. Yuhui, and A. Romanovsky, Real distribution of response time instability in service-oriented architecture, IEEE Symposium on Reliable Distributed Systems, India,, Nov. 2010, pp [12] A. Wrzosk, Applying NAF for performance analysis: Performance analysis of SOA systems using LQN models, Communications and Information Systems Conference (MCC), Gdansk, Oct. 2012, pp [13] L. O. Lero, P. Merson, and L. Bass, Quality Attributes for Service-Oriented Architectures, Systems Development in SOA Environments (SDSOA), Minneapolis, May. 2007, p.3. [14] C. Mainka, J. Somorovsky, and J. Schwenk, Penetration Testing Tool for Web Services Security, IEEE International Congress on Services (SERVICES), Honolulu, Jun. 2012, pp [15] M. Tang,and L. Ai, A hybrid genetic algorithm for the optimal constrained web service selection problem in web service composition, IEEE Congress on Evolutionary Computation (CEC), Barcelona, Jul. 2010, pp [16] K. Drira, R. B. Halima, E. Fki, and M. Jmaiel, A large-scale monitoring and measurement campaign for web services-based applications, Concurrency and Computation: Practice & Experience - Performance Evaluation of Communications in Distributed Systems and Web-Based Service Architectures, vol. 22 Issue 10, Jul. 2010, pp [17] P. Chodhury, P. De, and S. Choudhury, A framework for performance analysis of client/server based SOA and p2p SOA, IEEE International Conference on Computer and Network Technology (ICCNT), Bangkok, Apr. 2010, pp

20 T. Anees and H. Zeilinger Performance Evaluation of a Service Availability Model [18] S. Mohapatra, J. Sahoo, and R. Lath, Virtualization: A survey on concepts, taxonomy and associated security issues, IEEE International Conference on Computer and Network Technology (ICCNT), Bankok, Apr. 2010, pp [19] H. Liu, H. Wang, Z. Wu, and J. Dong, A new method to improve the dependability of key nodes in pervasive computing environment, IEEE International Conference on Pervasive Computing Signal Processing and Applications (PCSPA), Harbin, Sep 2010, pp [20] Y. Krasny, A. Krits, E. Farchi, G. Kliot, and R. Vitenberg, Effective testing and debugging techniques for a group communication system, IEEE International Conference on Dependable Systems and Networks (DSN), Jul 2005, pp [21] J. Wen, P. He, K. Wu, and J. Xu, Monitoring resources allocation for service composition under different monitoring mechanisms, IEEE International Conference on Complex, Intelligent and Software Intensive Systems (CISIS), China, Jul. 2011, pp [22] M. Yin, Assessing availability impact caused by switchover in database failover, Annual Reliability and Maintainability Symposium (RAMS), Fort Worth, Jan. 2009, pp [24] WANem, Wide Area Network Emulator, [retrieved: Feb 14, 2013] from [25] Glassfish homepage, [retrieved: Feb 16, 2013] from [26] juddi, An Apache Project Homepage, [retrieved: Feb 11, 2013] from [27] Apache Tomcat, [retrieved: Feb 14, 2013] from [28] MySQL, [retrieved: Feb 22, 2013] from 120