Within the framework of "emerging information technologies", we have developed a new kind of distributed computing substrate that could appear in the near future, with the advent of a new era of computing, coming after the mainframe era, and the PC era. In this 3rd era of computing, we are likely to see a myriad of ubiquitous devices with large computing and sensory capacities in our environment. Such devices will interact with each other using wireless communication and coordinate to seamlessly help people in their daily tasks.
Anticipating this new era of computing, we have envisioned a solution for simulating complex systems, inspired from the distributed computing projects, but this time using the computational resources of ubiquitous devices in our environment: our cellular phones, PDAs, the alarm clock radio on our night table, our children's electronic toys and robots, our microwave oven, etc.
PERvasive computing framework for modeling comPLEX virtually-Unbounded Systems
A flyer illustrating the concept and current achievements of the project is available for download here.
We have set up a demonstration that shows the first prototype of what is meant to be a "scale model" of what a SETI-like framework for ubiquitous computing would look like. This prototype is a pervasive hardware platform (a series of PDA-like devices) made of custom reconfigurable devices endowed with bio-inspired capabilities, such as growth, learning, and evolution, that shall enable the simulation of large-scale complex systems and the study of emergent complex behaviors. We have shown that a series of complex models in the domains of neuroscience and social sciences can take advantage of such a modeling framework for studying the emergent behaviors arising from such models. Moreover, such ubidules also control autonomous mobile marXbot robots that interact with the user and with other robots, thus providing a means for simulating embedded and situated versions of complex systems.
PERPLEXUS hardware platform for modeling complex systems
Ubiquitous computing modules for complex systems modelling
We have developed a set of ubidules with the aim of demonstrating a "scale model" of tomorrow's use of ubiquitous computing for modeling complex systems. An illustration of the PERPLEXUS platform can be downloaded here.
An ubidule is a PDA-like device running an agent framework that enables users to run complex systems models in a distributed way, interacting with the environment and using an ubichip, a custom reconfigurable integrated circuit exhibiting bio-inspired capabilities.
Ubidule - UBIquitous computing moDULE
At the heart of our ubidules lies custom reconfigurable devices, called ubichips, endowed with bio-inspired capabilities, such as growth, learning, and evolution, that shall enable the simulation of large-scale complex systems and the study of emergent complex behaviors.
Ubichip layout
More information about the ubichip architecture can be found here. The specification of the bio-inspired features supported by the device are thoroughly described in the downloadable report.
The modeling platform programming methodology extensively relies on the JADE Agent-based Modeling Framework. It provides a unified solution for application specification, software-based simulation and hard- ware implementation / monitoring. The cornerstone of the so-called BAF - Bio-inspired Agent Framework is an agent compiler that enables compiling JADE application agents into polymorphic agents capable of being executed in software, or hybrid hardware / software mode. The set of tools includes the so-called JubiSplitter, Jubicompiler, and Jubiassembler. Together with the Ubimanager (the software tool for managing ubichips), these tools enable the user to implement software, hardware and hybrid soft- ware/hardware complex system models, even if certain applications may require of additional tools, like the SpinDek for spiking neural models, or ASEBA for the marXbot robot-based embodied models. For more information on the modeling platform, please refer to a dedicated document.
Platform programming tools - Illustration of the Ubimanager running on the ubidules, on the marXbot robots and on a personal computer
A series of modeling applications have been implemented on the hardware substrate: culture dissemination models, biologically-plausible spiking neural models, synaptogenesis models, as well as evolvable hardware applications and an embedded system for human activity identification. Out of these applications, two main applications are being used for benchmarking purposes: a biologically-plausible spiking neural network model and a culture-dissemination model. For more information about these modeling problems, please refer to a report on the specification of biologically plausible developing neural networks modeling problem, a report on specification of the culture-dissemination modeling problem and a report on the set of benchmarks obtained by the software simulation of the modeling application.
Dynamic-structure neural and complex networks and emergent behaviors
As far as the marXbot robots are concerned, we are currently building 20 robots (a total of 70 robots are being built for several european projects), as well as the experimental arena for running a collective exploration application, taking advantage of self-organizing principles and information dissemination.
marXbot robot and experimentation arena
For more information on the MarXbot application, download a poster here. Our ubidules can control marXbot robots that interact with the user and with other robots (see video), also enabling the simulation of embodied and situated complex systems models.