WP3: Agent Infrastructures Development


Infrastructures (or middleware) play a fundamental role in the engineering of complex software systems characterising modern and future information society, despite of the specific engineering paradigm adopted. Such an importance is evident when considering mainstream and standard paradigms as found in state of the art technologies: examples range from Web Services and OSGi as middlewares for service-oriented architectures, to CORBA as middleware for component-based/object-based systems, down to virtual machines such as JVM and CLR which can be thought as middleware for object runtimes. The importance of infrastructure is even more evident when considering agents and MAS as main paradigm, whose typical scenarios involves complex systems composed by a multitude ofdistributed active entities organised according to some model, immersed in specific software/hardware environments, executing some kind of individual and collective tasks which require their fruitful interaction and coordination.

Complexity of such scenarios can be framed at different levels of abstraction: from concurrency and distribution (control, space, time,..), to system/environment openness in terms of dynamisms, heterogeneity, unpredictability, and so on. As in the case of human societies, infrastructures in MAS are meant to play a fundamental role in governing such a complexity, factorising critical issues of systems in terms of services at different levels, from sustainability to security, from communication to coordination and organisation, and so on.Currently, most of the available state of the art MAS infrastructures has been conceived and designed in academic contexts, and can be still considered in its infancy, in particular with respect to mainstream middlewares. Among the others, such an immaturity affects three different aspects:
  1. abstractions and services provided
  2. available tools and technologies
  3. integration with agent-oriented methodologies

Then, we consider the research on MAS infrastructure models and supporting tools, on the integration with agent-based methodologies, and on architectures and technologies used for MAS development essential for advancing the state of the art on software engineering, in particular for what concerns complex software systems. Given these premises, the objective of this workpackage is the identification of suitable infrastructure supports for agent-based systems and MAS, for the engineering of complex software systems, in particular for their runtime and online management. We will consider existing infrastructure models, approaches, and their extention and integration, and we will develop a framework that will cover the main infrastructure issues of interest and support the methodological framework developed in WP1 and WP2. The workpackage will be articulated as follows:

A3.1: Study of the state of the art

In this activity, we will consider the main abstractions at the base of current MAS infrastructures and which play a key role in the engineering of systems on top of such infrastructures. In particular, after identifying precisely the role of infrastructures in general for the engineering of modern complex software systems, we will consider the main infrastructure approaches for agents and MAS in the state of the art: examples are JADE, TuCSoN, RETSINA, TOTA. Also, standard and mainstream infrastructures will be briefly considered in the perspective of comparisons and integrations: examples range from Web Services and OSGi for service-oriented, CORBA for OO/Component-Oriented, PKI as infrastructure for supporting security services, and so on. In particular, the main aspects that will be focussed will concern infrastructure support for systemic dimensions, organisation and coordination in particular, which will be used to frame other important aspects such as security (at different levels), trust and so on. Note that this study is different from that one proposed in WP1.

Expected Results Technical reports will be produced as an outcome of the stage, to be distributed to the other research units involved in the projects and - possibly - to the scientific community. The reports will discuss the result of the survey, focusing in particular on the weakness and strength in existing approaches with respect to the objective of this project.

Begin/End: month 1 - month 5

Deliverables: D3.1 Document describing the state of art

A3.2: Definition of an infrastructural model

In this activity an infrastructural model will be developed in order to provide some basic unifying abstractions to integrate the two worlds, and in particular the methodological framework outcome of the WP2, in armony with the meta-model identified in WP1. In particular, the key point of such a framework will be the identification of first class abstractions which on the one side could be easily mapped on primary concepts of the methodological framework devised in WP2, on the other side could be supported by infrastructures as first class entities, in order to bridge the scientific and engineering gap between methodologies and infrastructures. Such abstractions are meant to be expressive enough to cope with the heterogeneity of existing agents and MAS (meta-)models, in particular dealing both with strong and weak notion of agency, and related conceptual domains.

This activity will be devoted to identify a suitable infrastructure model, starting from existing approaches considered in A3.1. In particular we will focus on the infrastructures which either directly or indirectly are part of the research work of the groups involved in the project, in particular: TuCSoN, as an infrastructure for MAS providing services for coordination, organisation and security, and JADE, as Java-based middleware for MAS based on FIPA standard. The objective of this activity is the identification of a model of infrastructure which could support the first class abstractions identified in previous stage, and then of related supporting tools for the integration of the methodological framework identified in WP1 and WP2. Following the lines identified by the framework, the model would account also for the integration with mainstream / existing approaches.

Expected Results A result of this activity is the proposal of a model of infrastructure - either as an extension / integration of existing infrastructures or possibly a new one - exhibiting the properties as accounted by the project. The proposal will eventually result in technical reports and scientific papers to be distributed to the other research units and disseminated to the research community.

Begin/End: month 5 - month 12,5

Deliverables: D3.2 Document describing the infrastructural model

A3.3: Design and Development of Prototypes

In this activity, infrastructure prototypes - and related supporting tools - will be designed and developed, following from the model identified in the previous activity, and starting from available open-source technologies, developed both inside the research units (e.g. TuCSoN and tuProlog) and outside (e.g. JADE). The design and development of the prototypes could possibly converge to the selection of a specific prototype or the horizontal / vertical integration of the different prototypes . From what concerns the development in particular, we - following what already done with technologies developed inside the group (e.g.: tuProlog) - aim at adopting languages and tools promoting as far as possible an agile and pervasive exploitation of the infrastructure, to support the engineering of system for heterogeneous application domains. Such an attitude leads quite inevitably to the adoption of the Java language and Java (runtime) platform, as the base on which possibly stacking other languages and virtual machines (e.g. Prolog). All the technologies developed in this stage are meant to be open-source. Moreover, in order to push the extendibility and usability of the technologies, we will evaluate the possibility to distribute infrastructure tools and more generally components of the infrastructure in the form of plug-ins for well known extensible open-source IDE (Integrated Development Environment), such as Eclipse. In this activity will strongly relate to the activities of WP2, and in particular A2.3 where we will develop the methodology support tool.

Expected Results As a result of the stage, infrastructure prototypes will be realised, with related technical documentation. The design and development of the prototypes could eventually generate some technical reports and papers concerning significant aspects of the architecture and the implementation.

Begin/End: month 10 - month 17,5

Deliverables: D3.3 Infrastructure prototypes and related documentation

A3.4: Infrastructures evaluation

This activity will be dedicated to the evaluation of the proposed infrastructure. We will evaluate the usability of the infrastructure through a case-study application (WP4).

Expected Results Scientific papers will be produced to spread the results of the research activity.

Begin/End: month 15 - month 20

Deliverables: D3.4 Document describing the evaluation results

Tags: MEnSA
2008 © aliCE Research Group @ DEIS, Alma Mater Studiorum-Università di Bologna