Agent-Based Development of Multisensory Monitoring Systems - Foundations on Natural and Artificial Computation - page 458

is ICARO-T in VigilAgent , as it provides the necessary functionalities for devel-
oping new modules capable to carry out this task. These modules are developed
following a general process based on both the definition of specific templates for
each target platform, and procedures to extract information from INGENIAS
models. In the literature there is no proposal for mapping INGENIAS models to
ICARO-T code. So, our contribution to solve this gap has been to identify them
and to develop modules to automatically carry out the Model-to-Text transfor-
mation from INGENIAS to ICARO-T.
The process for using the IDK modules in order to automatically generate
ICARO-T code of the system being developed that can be updated with new
code introduced by the user when necessary is carried out as follows. (1) The
INGENIAS ICARO-T Framework generator module ( IIF ) is used to automat-
ically generate code for the detailed design specification. IIF generates several
XML files that describe the behavior of each agent, java classes for each agent
and application, and the XML file describing the application deployment. (2)
The developer manually inserts code in the protected regions of the generated
java classes and implements those new classes he/she needs. (3) The developer
uses the ICAROTCodeUploader module in order to update the model with the
modifications introduced in the protected regions. It allows the developers to
always keep the model and the source code synchronized, so that those changes
introduced in the source code are kept when code is regenerated again from
the model. Finally, the script file generated by IIF module is executed by the
deployment manager to launch the developed application.
4 Conclusions
The lack of consensus about which the best methodology to develop agent-
oriented applications is has driven the main idea presented in this paper: not to
propose a new methodology from scratch but to combine several methodologies,
VigilAgent , to take advantage of each one of the analyzed methodologies. The
learning curve of VigilAgent can be steep at first because users must get used to
different terms that have the same meaning depending on the technology used in
each phase (Prometheus and INGENIAS for modeling, and ICARO-T for imple-
mentation). However, this disadvantage is overcome thanks to the two transfor-
mations that are executed automatically. First, a Model-to-Model transforma-
tion executed by means of Medini automatically transforms Prometheus struc-
tures into their equivalent INGENIAS structures. Second, the Model-to-Text
transformation tool transforms INGENIAS models into code for the ICARO-T
framework.
It is worth pointing out that the time spent learning how to develop and
implement the INGENIAS ICARO-T Framework generator and the ICAROT-
CodeUploader modules was two months and fifteen days. This effort is rewarded
as new applications are modeled and implemented because (i) the quality of the
developed system is improved as the automatically generated code does “not”
have bugs; (ii) productivity is improved as the time necessary for coding is re-
duced because the developer does not need to learn the structure, location and