NIVEL 3 DE DESEMPEÑO PISA CIENCIAS
XI.4. Segundo periodo escolar, al concluir el tercer grado de primaria, entre 8 y 9 años de edad 1 Estándares de Español
XI.4.4. Aprendizajes esperados de Lengua Indígena
Agent based model is useful for simulating people’s movement in smart environment. It
is a modeling paradigm that models a system by modeling its individual components as agents.
An agent is an autonomous entity that performs certain actions and interacts with other agents in
will perform and how will they interact with each other. The dynamic of the system is the sum of
dynamics of all of the agents in the system. The philosophy of agent-based model is “the whole
is greater than the sum of parts” because complex system behaviors can be generated from simple rules of each agent. Typically, an agent-based model consists a set of agents, a set of
rules each agent will follow, a set of interaction topology connecting one agent to another and a
non-agent environment. For example, an agent-based model that models the social network
would contain a set of agents where each of the agents represent a single person. Each agent is
assigned with the behavior such as talk and argue. Assuming that agents are all interconnected with each other and a variable called “relationship” represents the weights of the connections. When two agents talk to each other, the weight of the connection between the two agents
increases. When two agents argue with each other, the weight of the connection between the two
agents decreases. It would be interesting to see how the social network will evolve given certain
intitial distribution for the agent to perform different behaviors. This is a typical agent based
model. The agent-based modeling and simulation has many advantages over the traditional
system dynamic modeling paradigm. The most important features are probably that the agent-
based model can represent the heterogeneity of the entity of the system and the ability of agent-
based model to simulate the interaction among different agents. These two features are otherwise
impossible to implement using a top-down approach. The heterogeneity of the entity of system
means that each agent can have its own preference of performing behavior or even its own
behavioral rules to follow. Without agent-based model, the entities of a system have to be
modeled in a homogenous manner, which is not practical in many real world scenario. (For
social system should be modeled in a heterogeneous manner). Also, the interaction among
entities of a system is difficult to specify without a proper agent-based model.
Because of the features of agent-based modeling, it is particularly suitable in simulating people’s movement behavior. The movement of people involves a process that is highly heterogeneous and interactive. Each individual can have his own choice of route, speed and
destination and acts according to its own preference to move to a destination. When individuals
are in close proximity, they will have to interact and perform maneuver to avoid each other.
Agent-based model captures the two features of the movement behavior of occupants naturally. A number of researchers have conducted their work on building models for peoples’ movement behavior in building environment [20][78]. Besides homogeneity and interaction,
agent-based modeling and simulation also has other features. For example, each agent can be
considered as a self-contained entity with its own attributes. These attributes can be added,
removed or shared with other agents easily and define the identity of the agent. With these
attributes, the agent can be recognized by other agents and the environment. An agent is also
autonomous, which means that its action or behavior can be easily defined using computer
program. Their functionalities are independent of environment and other agents. By observing
the environment and other agents, an agent determines its behavior to perform. The agent’s behavior can be of any form. It can be as simple as a single equation or it can be complex
procedure. Furthermore, an agent has its own state, which determines the behavior it will
perform. The aggregation of the state of each agent becomes the state of the whole agent-based
model. The state of the system at some time step is determined by the state of system at previous
time step and determines the behavior of the system as a whole. Moreover, the agent can have
behavior and action rules according to a learning procedure. The learning procedure involves
utilizing accumulated experience of the agent to decide its preference of behavior. This requires
that the agent has some sort of memorizing mechanism to store its perceived information from
the environment and other agents. Finally, an agent can be goal directed. This means that each
agent can be assigned with a task. The behaviors of the agents are designed specifically for
accomplishing this task. This is particularly useful in simulating movement behaviors of people
because people normally move according to a predefined route which can be considered as a
goal directed behavior. [85]
In the previous research of agent-based modeling and simulation, agent-based simulation
is usually used as an offline tool to help the system design without incorporating observation
data. However, the trend of utilizing observation data in design and application of agent based
models has received much attention in recent years. For instance, the work of [79] carries out
micro-simulation and initializes an agent based model using contextual data. The work of [80]
discusses the potential to combine data mining technology and agent based model together so
that the two technologies utilize each other to solve problems in their own domains. A
framework to integrate data into the procedure of modeling, validating, calibrating and analyzing
of agent-based models is proposed in [81]. Another framework focusing on integrating data into
the modeling procedure of agent-based model is presented in [82]. Furthermore, how agent-
based models can work with data and observation models from other discipline such as GIS is
discussed in [83][84]. Nevertheless, most of these works focus on incorporating observation data
into modeling or validation procedure; none of them utilizes real time data to calibrate the
simulation and to estimate of the system’s state under study. In this dissertation, we construct an agent-based model to simulate the movement of occupants in building and develop a data
assimilation framework to incorporate observation data into the model to estimate the positions
of occupants in the building.