A novel multi-objective multi-constraint genetic algorithms approach for co-ordinating embedded agents

In this paper, we present a distributed, fault tolerant and adaptive software architecture for cooperative multi-embedded agent systems that operate in ubiquitous computing environments. The system is based on a novel genetic algorithm that learns to co-ordinate a set of embedded agents whilst satisfying a set of local and global objectives and constraints. The system operates in a lifelong learning mode which adapts to changes in the environment or users´ requirements. We experimented on various embedded agents situated in a ubiquitous computing environment test bed which is the Essex intelligent dormitory. The results manifested that the system can converge within a short time interval to a coordination strategy that satisfies the global and local objectives and constraints.