An Autonomic Approach in Swarm-Array Computing

Abstract

Can autonomic computing concepts be applied to traditional multi-core systems found in high performance computing environments? In this paper, we propose a novel synergy between parallel computing and swarm robotics to offer a new computing paradigm, ‘Swarm-Array Computing’ that can harness and apply autonomic computing for parallel computing systems. One approach among three proposed approaches in swarm-array computing based on landscapes of intelligent cores, in which the cores of a parallel computing system are abstracted to swarm agents, is investigated. A task gets executed and transferred seamlessly between cores in the proposed approach thereby achieving self-ware properties that characterize autonomic computing. FPGAs are considered as an experimental platform taking into account its application in space robotics. The feasibility of the proposed approach is validated on the SeSAm multi-agent simulator.