Development of autonomy oriented computing (AOC) systems for distributed computational problem solving

点击翻译按钮获取中文摘要

Autonomy oriented computing (AOC) is a bottom-up computing paradigm for solving computationally hard problems and for characterizing complex systems. Drawing on our previous work on AOC (Liu et al., 2002, 2004,2005), in this proposed research program, we aim to systematically model and develop autonomy oriented self-organizing systems as well as to empirically validate the capability and robustness of the AOC systems in solving large-scale computationally hard problems, such as distributed constraint satisfaction and optimization problems. In so doing, we will formally characterize the basic AOC elements, their local interaction and self-organization processes, and thereafter gain an in-depth understanding of the AOC performance from the points of view of systems optimization and topology-based computational complexity studies.   This research will have direct impacts on several important systems applications: (1) ( megascale) The development of e-transformation (e.g., e-learning) solutions, in which the service related problems are large-scale, highly-distributed in nature, and can readily be formulated, solved, and evaluated using the AOC approach; (2) (microscale) The development of autonomy oriented sensor networks that perform the tasks of collective spatial mapping (or world modeling) while satisfying their resource and energy constraints; (3) (nanoscale) The design and development of amorphous computing devices such as nanomachines and cellular computers, in which computational or programming means for generating coherent behaviors from vast numbers of unreliable parts, called "computational particles" (whether microfabricated particles on a chip or engineered cells for assembling nanoscale structures), are essential that are locally-interacting, irregularly-placed, and asynchronous.