Skip to Main Content
A key problem of component-based grid application configuration is to map services onto the execution nodes of the grid environment such that all services of the application satisfy some minimum quality requirements. This problem is known to be NP-hard. This paper presents two extensions to our previous ant-based application service mapping heuristic, in order to establish some coordination among the working agents in a decentralized environment and improve its convergence rate. The first extension proposes to use gatekeepers that learn to guide the movement of intelligent foraging ants. The second extension is a coordination mechanism to allow an ant to draw on the learned knowledge of its co-workers. The simulation evaluation of the proposed extensions shows that inclusion of scaled sleep time further improves the convergence rates while the gatekeepers seem to have a negative effect.