Distributed Embedded Intelligence Room with Multi-agent Cooperative Learning

In this paper, a novel Multi-agent control system with fuzzy inference learning and its physical testbed are presented. In the Multi-agent system, distributed controlling, monitoring and cooperative learning are achieved through ubiquitous computing paradigm. The physical testbed named Distributed Embedded Intelligence Room (DEIR) is equipped with a fair amount of embedded devices interconnected in three types of physical networks, namely LonWorks network, RS-485 network and IP network. The changes of environment states and user actions are recorded by software agents and are processed by fuzzy inference learning algorithm to form fuzzy rules that capture user behaviour. With these rules, fuzzy logic controllers can perform user preferred control actions. Comparative analysis shows our control system has achieved noticeable improvement in control accuracy compared to the other offline control system.