Competence Center Agent Core Technologies (ACT)

Transcription

1 Competence Center Agent Core Technologies (ACT) DAI-Labor The Laboratory for Distributed Artificial Intelligence (DAI-Labor) is an institute of the Technische Universität Berlin; it currently is employing about 100 researchers, graduate students, postdocs and support staff. The research focus lies on agent technologies and its fast and safe deployment in diverse application areas of business and telecommunications. In particular, an open and scalable agent framework for the design, test and deployment of sophisticated services, the Java Intelligent Agent Componentware (JIAC) is developed at DAI-Labor. Additional fields of research at DAI- Labor are wireless technologies in heterogeneous networks, smart filtering algorithms and data mining, as well as next generation networks and security issues concerning network operators and end users. Competence Center Agent Core Technologies (ACT) The CC ACT works on two aspects of distributed computing. On the one hand, service engineering techniques and tools are investigated, with the goal of enabling the quick and easy creation of complex service-based applications. On the other hand, we work on agent-based serviceware frameworks which deliver the environment necessary to implement highly complex and adaptable systems. Here, we focus on the seamless integration of various technologies such as SIP, OSGi, and UPnP, as well as rich service semantics which allow the use of those technologies in a single distributed framework. We have designed and implemented an agent based serviceware framework which has been certified according to Common Criteria EAL3, and which supports smart mobile agents in dynamic environments, as well as management and AAA functionalities. Advanced tools which utilise agents support the developer from initial conceptual design down to implementation and deployment. Main Research Areas Framework and Architectures The cluster Framework and Architectures (FAC) develops new concepts and ideas for service delivery platforms (SDP), designs, and implements them. The work aims at creating different proof-of-concept platforms that fulfil different project-related requirements. The work always aims to integrate existing platforms and standards from industry and academia. Most of the research focusses on smart and self-healing services, and the implementation thereof using intelligent agents in serivce oriented settings. The

2 integration of agent technologies and insights won from service oriented architectures allows to design future-proof SDP's. Service engineering and Tools The cluster Software Engieering and Tools (SET) bundle and extend expertise in methods, techniques, and processes, as well as languages and support tools. The output of the cluster includes elements and tools to support research and development of agentand service-oriented technologies. Currently, the focus lies in the visual design of business processes, easy service creation, and generative tools. An important aspect is the extension of service delivery platforms with modern methods and tools which allow the fast and efficient creation of services, as well as the simple maintenance of deployed services. Furthermore, the cluster maintains and supports tools for the JIAC IV serviceware framework. Reference Projects JIAC Intelligent and autonomous agents are the key technology for realising flexible and distributed systems. In the JIAC project [Fricke 2001] a scalable serviceware framework for agents was developed, which allowed adaptive configuration relating to changing users, groups, security policies and interoperability. The core of the JIAC serviceware framework is a component oriented architecture, which allows creating software agent from different software components [Sesseler 2002]. Each needed ability like understanding different communication protocols, knowledge processing and action managing as well as scheduling can be found in a special component. Further more, additional collections of services, possible tasks and ontologies can be found, e.g. billing and accounting services. Unlike JIAC components, services, actions and ontologies are not written in Java, they use a special logical and declarative programming language called JIAC Agent Description Language (JIAC), which allows fast creation and configuration of agents. Application specific, additional functions can be created by extending already existing components, services, actions or ontologies using Java or JADL [Konnerth2006]. JIAC Agents base on the Belief, Desire, Intention (BDI-) Model, which means that they follow goals to achieve certain desired states. This is done more efficiently by having the ability to choose between different Actions. Communication between components of an agent is realized by asynchronous messages. For data exchange between agents and services, which base on FIPA conform messages. JIAC supports HTTP, TCP, and SSL for exchange of FIPA conform messages, where only HTTP is described in the FIPA standard, but using TCP and SSL increases security and reliability. An additional characteristic of JIAC is that security played an important role in design phase [Bsufka 1999]. Different security mechanisms are implemented in JIAC [Schmidt 2002], e.g. the protection of the transport layer with

3 SSL, as well as the protection of the application layer. Further more, mechanisms like authorization and authentication of agents, agents owners and users are included. All these mentioned mechanisms base on the Public Key Infrastructure (PKI) mechanisms in JIAC [Bsufka 2006]. JIAC supports the PKI standards x.509 and SPKI. The migration of mobile agents basing on trusting systems for mobile agents [Albayrak 2001] is included, too. Further more, JIAC version is evaluated and certified for Common Criteria EAL 3 [DAILabor2004, Geissler 2004, BSI 2005]. Beside the ability of reading private keys from filesystem, in version agents are able to use private keys from chip cards with a special enhancement [Scheunemann2002]. With JIAC, development is not only supported by collections of components, services, actions and ontologies, several additional tools [Hirsch 2006] that help to create agents and to define their roles and communication protocols, are provided MAMS The development of new services for telecommunication applications and other IT systems is one of the most important vehicles of innovation for telecommunication and other service providers. Competition and ever growing expectations of users require faster service developments and the means to deliver them promptly. Thus, the acceleration of the service development process requires the development of new concepts and tools. The MAMS project develops different approaches as answers to these problems. On the one hand, a development environment for non-expert users is created that permits the composition and deployment of new services. Thus, the non-experts become developers and can apply their creativity and domain knowledge independently of service providers, developing new services and applications. Furthermore, tools for experts are developed which allow the fast and efficient development of new components for non experts, providing new building blocks or complete applications altogether. Standards like Web services or process modelling languages such as BPEL are the basic technologies that are used and extended. Second, the development tools and the execution environment are integrated into a unified framework. This permits a smooth deployment of the developed services and applications and aims to achieve a further acceleration of the development process. The execution- and the development environment offer the integration of most modern infrastructures, so that services and applications can rely on up to date technologies and infrastructures. For example, the IP Multimedia subsystem (CIM) belongs to the infrastructures for the integration of multi media connections, e.g. for video telephony or conference calls, etc. Thus, the MAMS Project offers a complete vertical integration of all layers, from the service development over execution and network layer down to the hardware. The integration of the different components is thereby a core element allowing a quick transition from development to deployment and execution of services and applications.

4 Besides these technical aspects the project also supports the development of solutions by integrating small to medium enterprises (SMEs), in order to receive direct feedback for the research process. These enterprises provide testing information and quality checking on the project results that flows directly back into the ongoing project. The MAMS project is financed by German Federal Ministry of Education and Research (BMBF) and led by prominent industrial partners from the telecommunication domain such as Deutsche Telekom AG, Alcatel AG and Infineon AG, as well as a group of research institutes, namely the Fraunhofer Gesellschaft and the Technische Universität Berlin. The work of the DAI-Labor takes place in close co-operation with the Institut für Übersetzerbau und Programmiersprachen (ÜBB) of the faculty IV of the Technische Universität Berlin LOMS The main topic of the project LOMS (Local Mobile Services) is to look into methods and tools for development, deployment, and use of location-based mobile services. LOMS is an ITEA project. Participating partners come from Belgium, Germany, and Spain. D-LOMS is the german part of LOMS and is sponsored by the German Federal Ministry of Education and Research (BMBF). The goal of the project LOMS is to define and develop an open service architecture through which new innovative local mobile services can easily be created, deployed, and consumed by mobile users. The aim is to lower the threshold for small and medium-sized enterprises (SMEs) and non-professional users to set up their own mobile services and allow charging from the content or service provisioning in business-to-consumer (B2C) or business-to-business (B2B) scenarios. The major project outcomes include the definition, creation, and the development of business models, user scenarios, and a Serviceware with associated support functions and tools for the provisioning of mobile content in local services, such as in shops, tourism locations, airports, car parks, and trains. A number of new innovative mobile applications will be created in the project to demonstrate the feasibility of the developed platform. Various enabling services and service building blocks will be promoted to a broad audience, allowing for the development of smarter, context-aware, and personalised mobile services. Selected Publications [Albayrak 2001] Sahin Albayrak and Karsten Bsufka. Integration von Public-Key- Infrastruktur-Funktionalitäten in Agenten-Toolkits. In S. Jablonski, S. Kirn, M. Plaha, E.J. Sinz, A. Ulbrich vom Ende, and G. Weiß, editors, Verteile Informationssysteme auf der Grundlage von Objekten, Komponenten und Agenten. PROCEEDINGS der Verbundtagung VertIS 2001, pages , October 2001.

5 [BSI 2005] Bundesamt für Sicherheit in der Informationstechnik. Certification Report BSI- DSZCC for Java Intelligent Agent Componentware IV Version from DAI- Labor Technische Universität Berlin [Bsufka 1999] Karsten Bsufka, Stefan Holst, and Torge Schmidt. Realization of an Agent- Based Certificate Authority and Key Distribution Center. In Sahin Albayrak, editor, Intelligent Agents for Telecommunication Applications, number 1699 in Lecture Notes in Artifical Intelligence, pages , Third International Workshop, IATA 99, Stockholm, Sweden, August [Bsufka 2006] Karsten Bsufka. Public Key Infrastrukturen in Agentenarchitekturen zur Realisierung dienstbasierter Anwendungen. Dissertation, Technische Universität Berlin, [DAI-Labor 2004] DAI-Labor. Security Target Java Intelligent Agent Componentware IV [Fricke 2001] Stefan Fricke, Karsten Bsufka, Jan Keiser, Torge Schmidt, Ralf Sesseler, and Sahin Albayrak. Agent-based telematic services and telecom applications. Communications of the ACM, 44(4):43 48, April [Geissler 2004] Tim Geissler and Olaf Kroll-Peters. Applying Security Standards to Multi Agent Systems. In The First International Workshop on Safety and Security in Multiagent Systems (SASEMAS) Part of AAMAS held at Columbia University New York City, 20 July 2004, July [Hirsch 2006] B. Hirsch, T. Konnerth, A. Hessler, and S. Albayrak. A serviceware framework for designing ambient services. In A. Mana and V. Lotz, editors, Proceedings of the First International Conference on Ambient Intelligence Developments (AmID 06), pages Springer France, [Konnerth 2006] Konnerth, T.; Hirsch, B. & Albayrak, S.Baldoni, M. & Endriss, U. (ed.) JADL --- an Agent Description Language for Smart Agents Springer Verlag, LNCS 4327, , 2006 [Scheunemann 2002] Heiko Scheunemann. Einbindung der TCOS-Chipkarten- Funktionalitäten In der Agentenarchitektur JIAC IV. Diplomarbeit, Technische Universität Berlin, 2002 [Schmidt 2002] Torge Schmidt. ASITA: Advanced Security Infrastructure for Multi-Agent Applications in the Telematic Area. Dissertation, Technische Universität Berlin, [Sesseler 2002] Ralf Sesseler. Eine modulare Architektur für dienstbasierte Interaktionen zwischen Agenten. Dissertation, Technische Universität Berlin, [Stender 2006] Mobility-based Runtime Load Balancing in Multi-Agent Systems Jan Stender, Silvan Kaiser, Sahin Albayrak 18th International Conference on Software Engineering and Knowledge Engineering (SEKE'06), Reedwood City, CA, USA

6 Persons Dr. Benjamin Hirsch studied Artificial Intelligence at the University of Amsterdam, and obtained his PhD in 2005 at the University of Liverpool (with distinction). His thesis has been nominated for the distinguished dissertation award 2006 of the British computer science society. Dr. Hirsch is the Director of the Competence Center Agent Core Technologies at the DAI Labor, which is part of the Technische Universität Berlin. The Competence Center consists of roughly 10 doctoral students as well as 5 graduate students, and focuses on next generation service-centric software. Under his supervision, the competence center staffs several projects related to serviceware frameworks and service engineering. He is also managing director of the project offices of the DAI Labor. Apart from managerial duties and supervising his team he is actively involved in research on agent technologies and has authored a number of papers, as well as reviewed in numerous agent related conferences. Dipl.-Inform. Thomas Konnerth studied computer science at the Technische Universität Berlin. Since finishing his Diploma, he has been working as a Researcher at the DAI- Labor on several projects including the JIAC IV agent-platform. Currently he is working on his PhD thesis which is focused on the subject of integrating agent-oriented techniques with service oriented approaches. Dipl.-Inform. Axel Heßler studied computer science at the Technische Universität Berlin. He works as a researcher at the DAI-Labor of TU Berlin. He has worked on a number of projects within the laboratory, including the JIAC IV agent-platform. Currently, he is finalising his PhD thesis about agent-oriented methodologies in the context of complex projects. Dipl.-Inform. Claus Schenk studied computer science at the Technische Universität Berlin. He works as a researcher at the DAI-Labor of TU Berlin. He has worked on a number of projects within the laboratory. Among them were several distributed agent applications, ranging from mobile applications and ehealth to the JIAC IV agent-platform. Currently, he is working on developing a new agent based service delivery platform and is finishing his PhD thesis about service discovery in distributed multi-agent systems. Dipl.-Inform. Silvan Kaiser studied Computer Science at the Technische Universität Berlin focussing on Computation and Information Structures (CIS) and artificial intelligence. Concluding with a master thesis on unified messaging based on software agents in 2000 he continued to work as researcher in Java based software agents and later on as research project manager. His PhD thesis focusses on the integration of software agents and grid computing.

7 Dipl.-Inform. Erdene-Ochir Tuguldur studied computer science at the Technische Universität Berlin. Since 2006 he works as a researcher at the DAI-Labor, where he focuses on real time agent systems. Dipl.-Inform. Alexander Thiele studied Computer Science and Physics at the University of Leipzig. After graduating in October 2006 he started working as a researcher at the DAI-Labor. From 2001 till 2006 he has been running his own company which focused on the development of custom software and webapplications. He has been working on several software projects including open source software. Currently he is preparing his PhD thesis. Dipl.-Inform. Janko Dimitroff studied Computer Science at the Technische Universität Berlin and graduated with a master thesis on supporting complex projects using enterprise architecture frameworks. His interests include distributed architectures, software architectures and software development. Dipl.-Inform. Jan Keiser studied computer science at the Technische Universität Berlin. He works as a researcher at the DAI-Labor of TU Berlin. He has worked on a number of projects within the laboratory, including the JIAC IV agent-platform. Currently, he is finalising his PhD thesis about management infrastructure for agent-based systems.