Alan Lopez

Pervasive environments are characterized by the including number of devices in a specific area that can be abstracted as services. The devices included in this specific area have different characteristics that make the environment very dynamic.
For instance, among the characteristics of these devices are that they enter and leave the area of interest dynamically, have limited quantity of energy, have little resilience, among others. These devices offer their services to anyone who requests them. However, the services that potential users require in general are complex and require the cooperation of several devices. We call this cooperation, a pervasive
service composition.
There are several open problems to satisfactorily perform a pervasive service composition. One of the main challenge is how could a system maintain useful composite solutions for as long as they are required given the dynamism of the environment? For this to be possible, the system needs to be highly adaptable.
To achieve this objective, the system must be able to detect changes in the environment, to know the available services to implement the adaptation, determine the most useful or least costly options to maintain o replace the required service.
In order to improve performance, we propose to use the system’s experience to face similar situations, that is to learn from the use of the system and its resources.
The proposal presented in this dissertation is based on the observation that human beings face similar problems frequently using their ability to adapt their behavior upon the environment’s changes, even at runtime. Based on this fact, our proposal make use of the Cuayollotl cognitive architecture, which is inspired by the functioning of the human mind- That is, we propose the use of Cuayollotl as a service composition system for pervasive environments. In this work we focus on the adaptability mechanism within the brain known as cognitive f lexibility, which implements the human ability to disengage from a behavior when necessary and generate a new behavior for solving the task at hand. This new behavior must meets the requirements to continue with the solution to the task at hand according to the new environmental conditions, or propose a complete new solution satisfying the solution of the task.