| Collection Mémoires et thèses électroniques |
|
| Accueil À propos Nous joindre |
| CHAPTER 3: A Generic Method for Developing 2D-3D Multiagent Geosimulations and its Application: The Main Steps and the Case Study | ||
|---|---|---|
|
|
|
Table des matières
In this chapter, we present the main steps of a generic method that can be followed when developing 2D and 3D multiagent geosimulation applications, simulating various kinds of systems/behaviors in virtual geographic environments. This chapter also presents the case study, which will be used to illustrate the application of this method: The customers’ shopping behavior in a mall.
Our literature review revealed that several methods and approaches have been proposed to develop simulation applications using various techniques such as discrete event simulation, continuous event simulation, object-oriented simulation, and multiagent based simulation. However, we did not find any paper or research work dealing with an approach or method that can be followed in order to develop geosimulation or multiagent geosimulation applications, simulating spatial phenomena or behaviors in virtual geographic environments. This lack of methodological works in the multiagent geosimulation field motivated us to propose a generic method to develop multiagent geosimulations in virtual geographic environments. This chapter aims to present an overview of the proposed method. In Section 3.2, we briefly describe the main steps of this method which have been inspired by different methods presented in our literature review (sub-sections 2.6.1 and 2.6.2). Section 3.3 aims to present the case study that will be used to illustrate the proposed method: the customers’ shopping behavior in a mall . The detailed illustration of the various steps of the method will be presented in the following chapters. Finally, the last section of the chapter states our conclusion.
This section aims to briefly present the main steps of a method that we propose for developing multiagent geosimulation applications aiming at simulating phenomena in geographic environments. These steps are depicted in Fig 3.1 and will be detailed in the following chapters (chapters 4 to 10) using the shopping behavior case study as an illustration.
Simulation applications are generally used to support decision making. In geosimulation applications, decisions are influenced by the spatial characteristics of the simulated phenomenon and the geographic features of its environment. Before developing a multiagent geosimulation application, we must study in detail the needs and goals of its future users. This step is very important because it helps us to identify the future users, the limits of the system and, of course, the spatial limits of the simulation. Identifying the limits of the phenomenon to be simulated and its environment is a very important task because it defines internal and external factors influencing the phenomenon to be simulated as well as the simulation inputs and outputs. This can help to reduce the level of complexity by reducing the size of the phenomenon to be simulated or the complexity of the simulation environment. This step is illustrated in the Chapter 4 of this dissertation using the shopping behavior in a mall as case study.
Based on the users' needs, we must identify the characteristics of the phenomenon to be simulated and those of its environment, including all the relevant spatial and non-spatial features within the limits that were defined in the previous step. This step is important because it prepares the ground for the following steps. This step is detailed in the Chapter 4 of this dissertation using the shopping behavior in a mall case study as an illustration.
In order to be able to simulate the studied system/behavior in a computer, we must model it, as well as its environment, taking into account the spatial and non-spatial aspects. Depending upon the users’ needs specified in the first step of the method, we must choose the appropriate level of detail or granularity of the model. Since our method is based upon the multiagent approach to simulate systems, we must use agent-oriented design techniques to create the models and represent the entities of the simulation. The Agent-Based Unified Modeling Language (AUML) (http://www.auml.org/) provides such techniques.
In this step of the proposed method, we can distinguish two categories of entities:
■ Passive Agents (PA) : This category of agents represents entities that have a structure, without a behavior. Usually, a large part of the elements contained in the simulation environment belongs to this category. Here, we must characterize the spatial and non-spatial structures of the passive agents in both 2D and 3D modes (see Fig 3.2).
■ Active agents (AA) : This category of agents represents entities, which have structures and behaviors: These entities actively participate in the simulation. For this category, we must specify the data structures of the entities (spatial and non-spatial structures) as well as their behaviors (spatial and non-spatial behaviors) (see Fig 3.3).
This step is illustrated in the Chapter 5 of this dissertation using the shopping behavior in a mall as case study.
The conceptual model generated during the last step must be executed on a computer. For this reason, we have two choices:
(1) to implement our simulation models using a standard language, such as C, C++ or Java for example. In this case, we need to select the language to be used for the implementation.
(2) to use an existing simulation tool/platform/language. In this case, we must choose a simulation tool/platform/language that will be used to execute the simulation models. There are several simulation tools/platforms/languages that can be used to create computer simulations. For example, the interested reader can visit the web site http://www.idsia.ch/~andrea/simtools.html, which presents a good collection of modeling and simulation resources on the Internet. Naturally, a question arises: How to select the appropriate simulation tool/platform/language for a given application? Metrics to evaluate simulation tools/platforms/languages include modeling flexibility, ease of use, modeling structure (objects, agents, etc.), code reusability, graphic user interface, animation, hardware and software requirements, output reports, graphical images, customer support, and documentation. It is important to mention that the choice of a suitable simulation tools/platform/language depends upon the characteristics of the system to be simulated and its environment.
This step is detailed in the Chapter 6 of this dissertation using the shopping behavior in a mall case study as an illustration.
In this step, we collect data and transform it in order to feed the simulation models. If we provide the simulation models with random data, this step can be very simple. However, if we want to use real data we must collect and analyse it before feeding it in the system. Since we deal with geosimulations, we must collect and analyze both non-spatial and spatial data. In our method, we use spatial and non-spatial multidimensional analysis techniques to analyze the multiagent geosimulation input data.
Several techniques exist, which can be used to collect non-spatial and spatial data, as for example:
- Questionnaires and surveys (pen and paper) : If the system to be simulated contains people that can be interviewed, we may use surveys and questionnaires to collect non-spatial and spatial data about it. The disadvantage of this technique is that the gathered data is on paper and cannot be used directly by a computer simulation. Hence, the need to digitalize it.
- Digital questionnaires or Web-based questionnaires : Some researchers used digital or web-based questionnaires and surveys in order to gather digital data that can be directly used by a simulation.
- Observations : We can observe the system/behavior intended to be simulated in order to better understand it. Then, based upon observation results, we can gather data about this system/behavior. The gathered data can be in a digital form or not.
- Consulting experts : We can gather data about the system or the behavior to be simulated directly from experts who know the system/behavior.
- GPS (Global Positioning Systems) : Taking advantage of the GPS technology, we may collect geographic data about the system/behavior to be simulated. The advantage is that the gathered data can be directly stored in databases or geographic information systems (GIS). Then, after some manipulations (validation, filtering, etc.), this data can be transferred in a computer simulation. However, this technique has a disadvantage because we can gather geographic data about the system/behavior only in areas where GPS signals are available.
After gathering simulation input data, it must be analyzed in order to determine some behavior patterns of the system to be simulated. Since we deal with both non-spatial and spatial data, we propose to analyze these two kinds of data using their respective analysis techniques. In this step, we use a multi-variables analysis approach to analyze the simulation input data. This step is illustrated in the Chapter 7 of this dissertation using the shopping behavior in a mall as case study.
During this step, we perform the simulation models on the selected simulation tool/platform/language, using the data characterizing the system and its environment. During this step, we must respect the constraints and limitations of the selected tool/platform/language, such as the input data structure. This step is illustrated in Chapter 7 of this dissertation using the shopping behavior in a mall as case study.
To be useful, the simulation application must return meaningful results. Based upon the analysis of these results, the users may make informed decisions. In our method, and in order to analyze the simulation output data, we use the multidimensional analysis techniques OLAP ( On Line Analytical Processing ) and SOLAP ( Spatial On Line Analytical Processing ) (Bédard et al., 2001). It is important to indicate that the simulation output data is generated by specific types of software agents called Observer agents . This step is detailed in Chapter 8 of this dissertation using the shopping behavior in a mall as an illustration.
The verification and validation processes insure that the simulation models accurately represent the real system/behavior to be simulated. These two processes which are theoretically distinct are closely related in practice (Arthur and Nance, 1996) (Balci, 1988). During this step of our method, we can compare the model’s performance under known conditions with that of a simulated real system. This step not only insures that the model assumptions are correct, complete, and consistent, but also enhances the users' confidence in the simulation models (Anu, 1997). Based upon the simulation input data and the simulation results, we can verify and validate our simulation models. This step is illustrated in Chapter 9 using shopping behavior in a mall as case study.
During this step, we document and test the simulation. In the documentation, we present the results of the system analysis, the simulation models, the selected tool/platform/language, a guide to use the simulation interface, the input/output data analysis results, etc. This step is illustrated in Chapter 9 using the shopping behavior in a mall as case study.
The last step of our method is the use of the multiagent geosimulator. According to (Anu, 1997), multiagent geosimulators can be used to:
- understand the system to be simulated by observing various simulations carried out over long periods of time, using the geosimulation platform.
- test the hypotheses about the system to be simulated. These hypotheses can be fixed by the simulation users.
- compress time in order to observe a system over long periods, or expand time to observe it in detail; to this end, the user can control the simulation time step.
- experiment with the system in new situations or contexts in order to assess the influence of different decisions. As an example, we can mention the assessment of the spatial environment. A user may need to evaluate the influence of important changes of the spatial configuration of the physical environment. Users can use the same simulation scenarios with different spatial configurations and compare the simulation results. The analysis techniques, presented in Section 3.3.7, can be used again for these comparisons.
This step is detailed in Chapter 10 using the shopping behavior in a mall as an illustration.
In order to illustrate our method, we use the case study of customers’ shopping behavior in a mall. Since the shopping behavior is carried out by people in geographic environment representing a mall, it is relevant to combine multiagent technology and geographic information systems (GIS) in order to simulate this behavior.
As it will be shown in Chapter 4, the shopping behavior in a mall is very complex given that it is influenced by a large number of factors, some of which are related to the shopper (internal factors), while others come from the mall’s environment (situational/contextual factors). The shopping behavior is also composed of several processes which are of different natures: cognitive, spatial, etc. In the following points, we present several aspects of this behavior.
a) The shopping behavior as a spatial and environmental behavior : The shopping behavior is performed in a geographic environment (the mall). In a shopping experience, shoppers visit stores, kiosks, and other places. Therefore, their behavior is easily influenced by the geographic characteristics of the environment (mall). The interactions, taking place between the shoppers and the spatial characteristics of the environment, let us consider the shopping behavior as a spatial behavior. In the literature, some researchers such as (Dogu and Erkip, 2001), presented the shopping behavior in a mall as a wayfinding/orientation behavior and mentioned some spatial factors affecting it.
b) The shopping behavior as an economic behavior : People go to the mall to visit stores/kiosk in order to purchase items or services. Sometimes, they are confronted with the choice of numerous stores/kiosks to visit, or items to buy. This choice generally depends upon their goals, preferences, and especially, upon the purchasing power of the shopper. Therefore, some researchers consider the shopping activity as an economic behavior, given the fact that it involves financial aspects (spending money).
c) The shopping behavior as a psychological behavior : As it will be shown in Chapter 4, the shopping behavior in a mall is influenced by various psychological variables, such as the personality, self-concept, and emotional state of the shopper.
d) The shopping behavior as a social behavior : (Fox, 2004) describes well that the mall is not only a place where we spend money. It is also a place to socialize, to meet people, etc.: « The mall became the new high street, the new town center, the new village green. The mall is usable in all weathers, there is no entrance fee, and there is no harassment. It is all very public and, hence, safe. Malls are colourful and exciting social places. The bustle and the movement, the tremendous variety of things and places, the food people love, and even movies. But most of all, just hanging out, meeting friends, making friends, drinking, gossiping, joking, making dates, etc. And all in a concentrated space – not stretched out over thousands of acres of suburb» . An important dimension of the shopping behavior in a mall is the social aspect. In addition, the shopping activity can be accomplished individually or in groups (couples, families, with colleagues or friends). The presence of other people in the environment (mall) also shows the social aspect of such behavior.
The method’s steps discussed in this chapter are based upon the majority of those existing in the literature, and presented in the previous chapter. Our method stands out from these methods and approaches by several characteristics which are presented in Table 3.1.
As shown in Table 3.1, the majority of methods and approaches can be used to simulate complex phenomena. What distinguishes our method from the other approaches is the fact that it takes into account the spatial characteristics of these phenomena, as well as the geographic features of the simulated environment. Like the methods proposed by (Drogoul et al., 2002) and (Ramanath and Gilbert, 2003), our method is based upon agent technology. It benefits from advanced capabilities of the agents. Unlike the methods proposed by (Drogoul et al., 2002) and (Ramanath and Gilbert, 2003), our method ca be used:
(1) to gather geographic data relative to the spatialized behaviors to be simulated,
(2) to specify agent models that emphasize spatial behaviors (such as navigation capabilities taking into account the geographic properties of the environment) and plausible interactions with the geographic environment (such as perception and memorization of objects’ locations, etc.), and
(3) to carry out practical analyses of the spatial dimensions of the simulation results.
Developing such method with such characteristics distinguishing it from others, can be considered as a contribution in computer simulation field, and more specifically, in the multiagent-based simulation field.
As mentioned in Table 3.1, we think that our method is most suitable for developing 2D-3D multiagent geosimulation applications of phenomena in geographic environments.
Besides presenting our method’s steps, this chapter also briefly presented the case study, which will be used to illustrate this method: the customers’ shopping behavior in a mall . The following chapters present details of the illustration of these steps using the shopping behavior case study.
© Walid Ali, 2006
|
|
|
|
| CHAPTER 2: Literature Review: Computer Simulation, Multiagent Based Simulation and Multiagent Geosimulation |
|
CHAPTER 4: Identification of the Application Domain, the Users’ Needs, the Characteristics of the Phenomenon to be Simulated and its Environment |
