Controlled Experimentation with Agents — Models and Implementations

Agents: Agenthood, Agent Architectures, and Agent Taxonomies

“…Abstractions are developed that take the purpose of testing into account (Broy et al, 2005). Simulation-based testing of software agents makes abstraction explicit including data abstraction, temporal abstraction, communication abstraction, and functional abstraction (Röhl and Uhrmacher, 2005). This simulation-based testing can be applied to the evaluation of a deliberative agent in its virtual environment, for example, in (Gierke et al, 2006) for evaluating the Autominder (Pollack, 2006) software (Figure 3.6), and for evaluating communities of reactive agents, for example, different user models and protocols in mobile ad hoc networks .…”

PASSIM: a simulation-based process for the development of multi-agent systems

“…This includes SIMDEI, a simulation tool which allows for the animation and analysis of the specification of the rules and protocols in an EI. In Rohl and Uhrmacher (2004) a modelling and simulation framework (DynDEVS) for supporting the development process of MASs from specification to implementation is proposed. The exploited simulation framework is JAMES, a Java-Based Agent Modelling Environment for Discrete Event Systems Specification (DEVS)-based Simulation, which aims at exploring the integration of the agents paradigm within a general modelling and simulation formalism for discrete-event systems.…”

“…To answer the second question we need to quantify the claimed added value in using simulation for MAS development through actual experimentations covering the whole software development life cycle of MASs: requirements capture, analysis, design, implementation, deployment and testing. To date a few MAS development processes have been proposed in the literature, such as Electronic Institutions (Sierra et al, 2004), DynDEVS/James (Rohl and Uhrmacher, 2004), CaseLP (Martelli et al, 1999), GAIA/MASSIMO (Fortino et al, 2005b), TuCSon/pi (Gardelli et al, 2005) and Joint Measure (Sarjoughian et al, 2001), which incorporate simulation to support the design phase of the MAS development life cycle, with the main focus on the validation and performance evaluation of the designed MAS model. However, to quantify the benefits of using simulation for MAS development, further research work needs to be carried out in the aforementioned direction and in further directions encompassing all the phases of the MAS development life cycle.…”

Distributed architectures for surrogate clustering in CDNs

“…To evaluate the performances of the proposed architectures, the agent-based modeling and simulation approach [10,14,2], which has been successfully applied in different application domains ranging from computers and telecommunications to biological and economic systems, is used. In particular, an agentbased modeling and simulation methodology is employed [5]: the modeling phase is driven by a Statecharts-based modeling language and supported by a visual tool which automatically translates visual specifications into code; the simulation phase is based on an agent-oriented and event-driven simulation framework which executes the code produced in the modeling phase in the context of purposely set distributed scenarios.…”