Heiko Rölke

An Extensible Editor and Simulation Engine for Petri Nets: Renew

Renew is a computer tool that supports the development and execution of object-oriented Petri nets, which include net instances, synchronous channels, and seamless Java integration for easy modelling. Renew is available free of charge including the Java source code. Due to the growing application area more and more requirements had to be fulfilled by the tool set. Therefore, the architecture of the tool has been refactored to gain more flexibility. Now new features allow for plug-ins on the level of concepts (net formalisms) and on the level of applications (e.g. workflow or agents).

Concurrent architecture for a multi-agent platform

A multi-agent system has a high degree of concurrency. Petri nets are a well-established means for the description of concurrent systems. Reference nets are higher level, object-oriented Petri nets. With Renew (REference NEt Workshop), there exists a tool to model and execute reference nets with seamless Java integration. So, reference nets can be used to design executable multi-agent systems while hiding the sometimes annoying details of concurrent implementations in traditional programming languages. The technique is currently used to implement a FIPA-compliant agent platform for multi-agent systems (called CAPA) focused on retaining a maximum level of concurrency in the system.

Modelling the Structure and Behaviour of Petri Net Agents

This work proposes a way to model the structure and behaviour of agents in terms of executable coloured Petri net protocols. Structure and behaviour are not all aspects of agent based computing: agents need a world to live in (mostly divided into platforms), they need a general structure (e.g. including a standard interface for communication) and their own special behaviour. Our approach tackles all three parts in terms of Petri nets. This paper skips the topic of agent platforms and handles the agent structure briefly to introduce a key concept of our work: the graphical modelling of the behaviour of autonomous and adaptive agents. A special kind of coloured Petri nets is being used throughout the work: reference nets. Complex agent behaviour is achieved via dynamic composition of simpler sub-protocols, a task that reference nets are especially well suited for. The inherent concurrency of Petri nets is another point that makes it easy to model agents: multiple threads of control are (nearly) automatically implied in Petri nets.

Properties of Object Petri Nets

In this presentation the structure of formalisms are studied that allow Petri nets as tokens. The relationship towards common Petri net models and decidability issues are studied. Especially for ”elementary object-net systems” defined by Valk [x] the decidability of the reachability and the boundedness problem is considered. It is shown that reachability becomes undecidable while boundedness remains decidable for elementary object-net systems. Furthermore it is shown that even for minimal extensions the formalism obtains the power of Turing machines.

A proposal for structuring Petri net-based agent interaction protocols

In this paper we introduce net components as means for structuring Petri net-based agent interaction protocols. We provide a tool for effortless application of net components to nets. Thus we facilitate the construction of nets and unify their appearance. Net components can be used to derive code for interaction protocols from a subset of extended AUML (Agent Unified Modeling Language) interaction protocol diagrams. This allows for a smooth integration of some traditional software development specification approaches with high-level Petri nets. By using net components we do not only unify the structure of Mulan agent protocols but also succeed to build a common language within a community of developers who share the net components.

Pattern based workflow design using reference nets

The development of workflow applications requires satisfactory concepts and tools. Workflow patterns cover the conceptual part. To base the patterns on high-level Petri nets allows for the tight integration of the modelling editor with the actual execution engine.The development process of workflow applications gains from this. We propose to use Reference nets as the modelling technique, Renew as the basic execution engine, and our workflow modelling tool for the design of workflows.The latter is a plug-in for the Renew editor, which is based on the use of workflow patterns. The development process is based on prototyping.

Reference and value semantics are equivalent for ordinary object petri nets

The concept of mobile agents imposes a great security risk for information systems. In this paper we propose object nets as a specification formalism for multi-agent systems. Since the general formalism is Turing-powerful not every analysis method that is common for Petri net can be applied. So, we define the subclass of “ordinary” object nets that allows for the application of standard P/T-net techniques, i.e. the computation of boundedness, liveness etc.

Dynamic Transition Refinement

Refinement of Petri nets is well suited for the hierarchical design of system models. It is used to represent a model at different levels of abstraction. Usually, refinement is a static concept. For many scenarios, however, it is desirable to have a more flexible form of refinement. For example in the context of service updates, e.g. version control in distributed systems, a mechanism for dynamic transition refinement is needed. The requirement of dynamic refinement at runtime is quite strong. Since we would like to redefine the system structure by itself, transition refinement cannot be implemented by a model transformation. Instead, an approach is needed which allows for dynamic net structures that can evolve as an effect of transitions firing. In previous work we introduced nets-within-nets as a formalism for the dynamic refinement of tokens. Here we consider an extension of nets-within-nets that uses special net tokens describing the refinement structure of transitions. Using this formalism it is possible to update refinements, introduce alternative refinements, etc. We present some formal properties of the extended formalism and introduce an example implementation for the tool Renew in the context of workflow modeling.

A Monitoring Toolset for Paose

Paose (Petri net-based Agent-Oriented Software Engineering) combines the paradigm of AOSE (Agent-Oriented Software Engineering, see [10]) with the expressive power of Petri nets - reference nets [12] to be more precise. While AOSE is a powerful approach when it comes to designing and developing distributed (agent) applications, it does not address the problems specific to debugging, monitoring, and testing of these applications, i.e. no global state of the system and very dynamic operating conditions. To tackle these problems, two tools have been developed in the context of Paose , which are presented in this work. Firstly, this paper will give a short overview over the interrelated set of tools, which exists already and supports Petri net-based AOSE. The tools are centered around the Petri net-based multi-agent system development and runtime environment Renew / Mulan / Capa . Secondly, Mulan -Viewer and Mulan -Sniffer will be presented in more detail --- two tools to address the issues encountered during debugging, monitoring, and testing agent applications. Both tools are first class members of the aforementioned family. The first tool, Mulan -Viewer, deals with the introspection of agents and agent behaviors, while it also offers rudimentary features for controlling the agent-system. The Mulan -Sniffer as the second tool places emphasis on tracing, visualizing, and analyzing communication between all parts of the multi-agent application and offers interfaces for more advanced methods of analysis, such as process mining. Both Mulan -Viewer and Mulan -Sniffer are realized as Renew plugins that can also be extended by other plugins.

Linking micro and macro description of scalable social systems using reference nets

Socionics attempts to release the architecture of multi-agent systems from the restrictive micro perspective viewpoint by the integration of the macro perspective in order to arrive at innovative agent systems. This paper shows how central research topics of sociology and computer science can be combined, in order to arrive at innovative agent systems. In the context of sociology the duality of micro and macro elements is relevant, while recursiveness of models appears in the perspective of computer science. These two elements are unified in our work to the socionic multi-agent architecture Sonar. The formal model, on which the representation bases, is the recursive formalism of reference nets—an extension of Petri nets that permits to understand nets again as tokens. With the help of these nets first of all a compact implementation of the multi-agent architecture Mulan is designed, secondly it serves as a description language for the sociological model, which is the fundament of Sonar. The main result here is to present an architecture based on Mulan and Sonar allowing to cover the micro as well as the macro perspective in agent-oriented modelling. Doing so, we introduce a scalable model based on agent systems.