Cell-Locomotion-Based Agent Migration over Distributed Systems

Abstract

This paper presents a middleware for multi-agents on a distributed system as a general test-bed for bio-inspired approaches. It is unique to other approaches, including distributed object systems, because it can maintain and migrate a dynamic federation of multiple agents on different computers. It enables each agent to explicitly define its own deployment policy as a relocation between itself and another agent. This paper describes a prototype implementation of the middleware and its practical applications to illustrate its utility and effectiveness in real distributed systems instead of using simulation-based evaluations