The fast-moving advances on Internet technology along with more recent developments delivered by ubiquitous computing, are making possible a surge of emerging application domains. Thus, not surprisingly applications on ambient intelligence, sensor networks, virtual organisations, and P2P, to name a few, are nowadays the matter of intense research and development. Typically, these applications are composed of a wealth of physical devices and/or software agents that act to achieve some global and/or individual goals. The achievement of such goals normally requires the coordination, and eventually the continuous adaptation, of their components? activities. This must be regarded as a tremendous challenge because of the lack of a centralised control, the partial observability of devices and (human and software) agents (and therefore of uncertainty), the openness of the system, and the high degree of dynamicity due to environmental changes (e.g. due to varying types of failures).

Hence, a fundamental question in this type of domains is how to achieve effective coordination among participants and how to sustain it over time. Notice that since these systems are situated in dynamic environments, coordination must be adaptive to continue being effective under unexpected (unplanned at design time) conditions. Because of the enormous complexity of such a task, we argue that it is fundamental to engineer these applications as systems capable of managing themselves, namely as self-* systems, where * stands for a number of properties: self-organisation, self-configuration, self-adaptation, self-diagnose, self-repair, etc. Since we can regard these systems as a kind of open systems, the multiagent system (MAS) view, where each participant is regarded as an autonomous entity, appears as suitable for them. Thus, we propose to engineer self-* systems as multiagent systems with self-* capabilities. In order to succeed in this endeavour, we focus on the following topics:

  • exploiting bio-inspired mechanisms for self-organising systems
  • engineering self-organising and self-adaptive systems
  • robust emergence of cooperation, converntions and social norms

ENGINEERING SELF-* VIRTUALLY-EMBEDDED SYSTEMS

Initial/final date: 
01 January 2010 to 31 December 2012
Main researcher: 
Project type: 
Plan Nacional
Funding Entity: 
MICIN TIN2009-14702-C02-01 (€ 145.321.00)
Description: 
Self-* systems, where * stands for a number of properties such as self-organisation, self-configuration, self-diagnose or self-repair, are systems capable of managing themselves. They provide the most convenient alternative for software development in open, unstructured and dynamic environments. These are the characteristics of a new surge of emerging application domains composed with a wealth of devices and software and human agents. These applications require the effective coordination among participants to achieve their global and individual goals. Furthermore, coordination mechanisms have to be adapted over time to continue being effective under environmental changes. Virtual worlds (VW) are interactive, computer generated 3D environments where participants come together for a variety of self and group determined purposes. They are a successful technology to integrate humans in computer systems. The main aim of this project is to apply multi-agent technologies to the engineering of self-* systems composed of software and/or human agents, while paying special attention to those that are to be virtually embedded. We propose to achieve that aim by fulfilling the following three objectives: 1. To design new computational models and algorithms for their application to the achievement of coordination in self-* systems. 2. To develop a simulation environment to test virtually-embedded self-* systems. 3. To construct three proof-of-concept prototypes that illustrate the algorithms and technologies created in the project.
Phd Students: 
Tomas Trescak
Daniel Villatoro
Norman Salazar-Ramirez
Meritxell Vinyals
José Luis Fernández
Research line: 
Self-* Systems
Acronym: 
EVE
External researchers: 
Nicholas R. Jennings,Simeon Simoff,
In collaboration with: 
Universitat de Barcelona, Birchman (EPO), Plastia (EPO), DLM Solutions (EPO)

Mecanismos de Autoorganización y de Control social generadores de normas sociales

Initial/final date: 
01 September 2008 to 31 August 2010
Project researchers: 
Main researcher: 
Project type: 
Intramural
Funding Entity: 
CSIC
Description: 
The main objective of this project is to explore the emergent scientific space generated by the intersection of three different paradigms dedicated to the study of self-organizing mechanisms in social systems: prehistoric etno-anthropology and social anthropology, experimental economics, and multiagent systems. The project aims to study the mechanisms of self-organization in societies where it does not exist a centralized control mechanism and where the institutional level is almost inexistent. We will study the mechanisms in charge of self-organization and distributed social control (as it is the case of reputation) that generate social norms in human societies and that can be easily applicable to virtual environments, either entirely populated by artificial entities (autonomous agents) or virtual societies, where both humans and virtual agents coexist. The work developed should serve as a guide to treat similar research topics where the behaviour of a group of individuals responds to the individual adoption of certain social conventions. This research can lead to three different and complementary paths: (i) a theoretical work where different social coordination mechanisms are identified by using the models and simulations suggested in this project; (ii) an instrumental work where the methodology and tools developed under this project are refined to build a technological corpus for social scientists; (iii) and finally, the application of several mechanisms for social coordination and the technological corpus in virtual communities as the integration of web services, ubiquitous and pervasive computing applications, and the online gaming applications.
Research line: 
Self-* Systems
Acronym: 
MacNorms