Jan Sudeikat

Evaluation of agent–oriented software methodologies – examination of the gap between modeling and platform

More and more effort is made to provide methodologies for the development of agent-based systems. Awareness has grown that these are necessary to develop high quality agent systems. In recent years a number of proposals have been given. Based on our experiences we argue that a complete evaluation of methodologies cannot be done without considering target platforms, because the differences between available implementations are too fundamental to be ignored. In order to conduct a suitable comparison we present a flexible evaluation framework that takes platform specific criteria into account. Part of this framework is a procedure to derive relevant criteria from the evaluated platforms and methodologies. In combination with a set of platform dependent and independent criteria our framework allows evaluation of the appropriateness of methodologies with respect to platforms. As a consequence, also the suitability of methodologies for an individual platform, or vice versa of several platforms for an individual methodology can be examined. To show the usefulness of our proposal, we evaluate the suitability of different methodologies for an example platform.

Validation of BDI agents

Testing and Debugging multi-agent systems (MAS) - which are inherently concurrent and distributed - is a challenging task. While complex application scenarios demand intelligent entities with autonomous reasoning capabilities, the applied reasoning mechanisms impair current approaches to validate MAS implementations. Reactive planning systems, namely the well-known Belief Desire Intention (BDI) architecture, have been successfully applied to implement these intelligent entities by means of goal directed agents. Despite testing and debugging, used to validate whether implementations behave as intended, are crucial to serious development efforts, only minor attention has been payed to corresponding tool support and testing procedures for BDI-based MAS. In this paper, we examine how the reasoning mechanism inside agent implementations can be checked and how static analysis of agent declarations can be used to visualize and check the overall communication structure in closed MAS. We present corresponding tool support, which relies on the definition of crosscutting concerns in BDI agents and enables both approaches to the Jadex Agent Platform.

MASDynamics: Toward Systemic Modeling of Decentralized Agent Coordination

Enabling distributed software systems to purposefully self-organize, i.e. to adapt to dynamically changing execution contexts by the collective adjustment of individual components, challenges current development practices. Since the dynamics of self-organizing systems arise from agent coaction, developers cannot directly infer the macroscopic system behavior from established agent design models. This paper plays a part in an ongoing research effort that addresses the provision of self-organizing processes as design elements, i.e. reusable patterns of agent interrelations. We propose a systemic modeling approach and support the application independent description of (inter-) agent coordination patterns by a domain specific language that allows to map interrelations of agent activity to detailed agent design models. This facilitates the separation of decentralized coordination strategies from domain specific agent implementations and enables development teams to treat nature-inspired coordination strategies, which steer self-organizing dynamics, as design concepts. In addition, we show how this modeling conception provides a declarative programming approach by the automated supplementation of conventional developed agent models with non-linear, inter-agent coordination mechanisms.

Systematically Engineering Self-Organizing Systems: The SodekoVS Approach

Self-organizing systems promise new software quality attributes that are very hard to obtain using standard software engineering approaches. In accordance with the visions of e.g. autonomic computing and organic computing, self-organizing systems promote self-adaptability as one major property helping to realize software that can manage itself at runtime. In this respect, self-adaptability can be seen as a necessary foundation for realizing e.g. self* properties such as self-configuration or self-protection. However, the systematic development of systems exhibiting such properties challenges current development practices. The SodekoVS project addresses the challenge to purposefully engineer adaptivity by proposing a new approach that considers the system architecture as well as the software development methodology as integral intertwined aspects for system construction. Following the proposed process, self-organizing dynamics, inspired by biological, physical and social systems, can be integrated into applications by composing modules that distribute feedback control structures among system entities. These compositions support hierarchical as well as completely decentralized solutions without a single point of failure. This novel development conception is supported by a reference architecture, a tailored programming model as well as a library of ready to use self-organizing patterns. The key challenges, recent research activities, application scenarios as well as intermediate results are discussed.

Modeling Feedback within MAS: A Systemic Approach to Organizational Dynamics

Organization---oriented modeling approaches are established tools for Agent---Oriented Software Engineering (AOSE) efforts. Role and Group concepts are prevalent concepts in the design of agent---based applications. These allow the definition and partition of static organizational structures and facilitate the description of agent behaviors in terms of role/group changing activities. Due to a growing interest in the construction of adaptive and self---organizing dynamics within MAS --- i.e. applications that adjust their organizational structure at runtime --- developers require tools for expressing the dynamics of MAS organizations, that result from individual agent activity and adaptiveness. In this paper we discuss how the macroscopic behavior of organizational structures can be modeled by relating systemic modeling techniques to MAS designs. Particularly the notions of causal loop diagrams , composed of system properties connected by causal links are applied to express the timely behavior of role and group occupations. Corresponding modeling activities are facilitated by a graphical notation that highlights feedback loops. Since simulations are indispensable to examine complex, non---linear behaviors, we discuss how the systemic semantics can be translated into systems of stochastic process algebra terms, therefore enabling model simulation.

DeCoMAS: An Architecture for Supplementing MAS with Systemic Models of Decentralized Agent Coordination

The systematic conception of decentralized and self-adaptive software systems, as demanded for the run-time management of today’s distributed, multi-tier software architectures, is an active research area. Here, we discuss a coordination architecture that facilitates the enactment of externalized coordination models in Multi-agent Systems (MAS). Coordination is configured by systemic models, i.e. structures of agent-behaviour interdependencies that denote the mutual influences of agent activities. The systematic, iterative application development is supported by a non-intrusive integration approach that enables developers to adjust application dynamics by complementing agent-based applications with systemic coordination models.

Toward Systemic MAS Development: Enforcing Decentralized Self---organization by Composition and Refinement of Archetype Dynamics

The utilization of self-organizing processes promises scalability, robustness and adaptivity in Multi-Agent Systems (MAS), solely based on decentralized coordination of individual actors. Bionicdevelopment approaches have been established, which reuse decentralized coordination mechanisms that are derived from natural self---organizing systems. In this paper, we address analysis activities in incremental MAS development, concerning with the derivation of system architectures that enable applications to meet system requirements. As the functional requirements to self---organizing MAS comprise recurring types of system wide dynamics, we propose a systemic approach to analysis and architectural design activities by the iterative refinement of macroscopic dynamics. Based on a catalog of dynamic models of currently applied environment---mediated design metaphors, we discuss how intended MAS dynamics can be modeled and refined to decentralized MAS designs. A systemic design procedure is proposed and exemplified in a case study that demands the combination of two established design metaphors to enable an projected level of MAS adaptivity.

A Systemic Approach to the Validation of Self---Organizing Dynamics within MAS

Conceiving applications as sets of autonomous agents is a prominent approach to the construction of complex distributed systems. Particularly attractive are decentralized application designs that enable adaptive, robust and scalable applications by allowing agents to self---organize. Tools to the construction of self---organizing MAS, e.g. decentralized coordination strategies, catch increasing attention in MAS research. However, their purposeful utilization challenges current development practices. The intended non---linear macroscopic dynamics hinder top---down designs on the drawing board and corresponding development procedures rely on sequences of manual system simulation. In order to stimulate methodical development and facilitate the validation of complex MAS by simulation, we present a systemic approach to the qualitative validation of macroscopic MAS dynamics. Describing MAS as dynamical systems enables developers to formulate hypotheses on the intended macroscopic MAS behaviors that guide system simulations. We discuss and exemplify how to (1) derive systemic models as well as hypotheses from MAS designs, (2) infer appropriate simulation settings to their validation and (3) interpret the obtained results. In addition, work in progress on the automation of both system simulations and their interpretation is outlined.

Monitoring group behavior in goal-directed agents using co-efficient plan observation

Purposeful, time- and cost-oriented engineering of Multi-Agent Systems (MAS) requires developers to understand the relationships between the numerous behaviors exhibited by individual agents and the resulting global MAS behavior. While development methodologies have drawn attention to verification and debugging of single agents, software producing organizations need to validate that the MAS, as a cooperative system exhibiting group behavior, is behaving as expected. Recent research has proposed techniques to infer mathematical descriptions of macroscopic MAS behavior from microscopic reactive and adaptive agent behaviors. In this paper, we show how similar descriptions can be adjusted to MAS composed of goal-directed agent architectures. We argue that goal-hierarchies found in Requirements Engineering and Belief Desire Intention (BDI) architectures are suitable data structures to facilitate a stochastic modeling approach. To enable monitoring of agent behaviors, we introduce an enhancement to the well-known capability concept for BDI agents. So-called co-efficient capabilities are a novel approach to modularize crosscutting concerns in BDI agent implementations. A case study applies co-efficient plan observation to exemplify and confirm our modeling approach.

Toward Requirements Engineering for Self - Organizing Multi - Agent Systems

Since Multi-Agent Systems (MAS) are composed of sets of (inter-)acting autonomous and pro-active agents, they provide suitable concepts to the construction of self- organizing processes, promising robust and adaptive system behaviors. An examination of current software-engineering approaches to self-organizing MAS identified a lack of requirement engineering approaches, established in general- purpose engineering methodologies. As self-organizing dynamics are typically utilized to enforce continuous MAS adjustment we discuss how expectations on MAS adaptivity can be expressed and validated, considering both the intended system configurations as well as the system dynamics that steer system (re-)configuration. We argue that self-organizing MAS development typically intends recurring pattern of system dynamics and exemplify their specification and validation, transferring requirements engineering practices to self-organizing MAS development.