Axel Heßler

Merging Agents and Services — the JIAC Agent Platform

The JIAC V serviceware frame work is a Java based agent framework with its emphasis on industrial requirements such as software standards, security, management, and scalability. It has been developed within industry- and government-funded projects during the last two years. JIAC combines agent technology with a service oriented approach. In this chapter we describe the main features of the framework, with a particular focus on the language JADL++ and the service matching capabilities of JIAC V.

A novel multi-agent system utilizing quantum-inspired evolution for demand side management in the future smart grid

The Smart Grid has become the future choice by many utility departments to attain bottom line goals of energy management. The Smart Grid will depend on a large number of renewable energy resources which require sophisticated control and coordination mechanisms for efficient and reliable demand side management DSM. In this paper, we propose a multi-agent based system to control and coordinate the operation among different entities within the Smart Grid. Specifically, the agents autonomously coordinate their activities to satisfy the local constraints of different entities while at the same time satisfying the underlying global goal of energy management. The novelty of our system is in formulating the coordination problem among the agents as a multi-objective optimization problem solved by a quantum-inspired evolution algorithm. We have extensively evaluated the system using the JIAC-V multi-agent platform. Experimental results show that our system is feasible and effective. Our method of coordinating the energy consumption of consumers using controller agents provides the basis for energy management at peak times. Thus, it promotes the wide application of the proposed system.

SeMa 2 :A Hybrid Semantic Service Matching Approach

The SeMa2 software module is a hybrid semantic service matchmaker based on OWL-S service descriptions and SWRL rules that participated in the S3 Contests 2008–2010. It provides syntactical as well as semantical matching techniques. Besides classifying input and output parameters it offers an approach to precondition and effect rule matching and reasoning. In this chapter we describe the architecture and the workflow of our approach and the results that have been evaluated during the last S3 Contest and with experimental test runs. Finally, we discuss the benefits and drawbacks of our implementation and propose future steps on our work.

Towards Transformations from BPMN to Heterogeneous Systems

By now, the mapping from BPMN to BPEL has been implemented in numerous tools, greatly assisting the business architect in the creation of BPEL processes. However, most of these tools are tailored especially for this transformation, neglecting the original purpose of BPMN: Providing a language independent process model. To address this shortcoming, a pure BPMN editor is needed, being dynamically extensible with several export features and added editing functionality. In this paper, we present a tool that follows this approach, not only providing a compelling transformation to BPEL but at the same time being extensible to other languages.

Towards Process-Oriented Modelling and Creation of Multi-Agent Systems

Different ways of integrating business processes and agents have been proposed, but using restricted process models or targeting only single agents, none of them is truly convincing. Nevertheless, business processes have many notions in common with agents and would be well suited for modelling complex multi-agent systems. In this paper, we combine concepts of two existing approaches to a mapping from business process diagrams to readily executable agent components. The results are well-structured and extensible, and at the same time account for nearly the entire expressiveness of the process modelling notation.

Unifying JIAC agent development with AWE

In this paper we describe the Agent World Editor, a tool for designing multi-agent systems and generating executable agent code. The tool also unifies the handling of different agent frameworks through an abstract agent model and an extensible transformation infrastructure. Currently, the tool supports three different agent frameworks of the JIAC family, and we feel confident that the approach holds for other frameworks as well as for the generation of multi-agent systems on heterogenous platforms.

A User-Centric Approach for Efficient Daily Mobility Planning in E-Vehicle Infrastructure Networks

The next generation of e-vehicles will be assessed on their ability to master the challenges posed by urbanization, resource restrictions and increasingly flexible and diverse user demands. Presumably, the vehicle-user-infrastructure network will become too dynamic and complex for an individual user to fathom without some sort of automated assistance. This paper proposes a user-centric, constraint-based, in-vehicle travel planning system, which schedules a daily travel plan of a user by exploiting knowledge about current and future states of the vehicle-user-infrastructure network. The system is implemented in an agent-based framework and is evaluated by a traffic simulator that additionally incorporates parking and charging lots together with an availability monitoring and booking service. The benefit of the planning system is assessed in a traffic simulation, where vehicles with and without a planning system compete for available resources. It can be shown that proactive conflict management and resource scheduling can reduce parking search time and overall travel time. Furthermore it can be shown that the adherence to schedule can be significantly improved.

HAI: a human agent interface for JIAC

There are many different application frameworks, which accelerate user interface development by simplifying repetitive and time consuming tasks. Most of these frameworks follow the widely accepted Model-View-Controller (MVC) architecture.Although, the existing frameworks are optimized for the implementation of object-oriented applications. The special features and possibilities offered by agent applications are not supported. Within this work, an architecture is designed, which allows the integration of any user interface in agent applications. As an extension of the MVC architecture, the advantages of agent-orientated software engineering will be combined with the advantages of the existing application frameworks. This structure provides the base for the Human Agent Interface, which allows the integration of any user interface in JIAC V agent systems.

Multi-Agent Programming Contest 2012 – TUB Team Description

We describe our contribution to the Multi-Agent Programming Contest 2012, which has been developed by students and researchers of the DAI-Labor at TU Berlin, Germany, using the JIAC V agent framework and the agile JIAC methodology.