Autonomous multi-agent systems for constrained devices in bu ilding automation domain

The automation system needs to evaluate the state of the environment and the capabilities afforded by the equipment, and consequently, based on the requirements needs to put in to action an appropriate control strategy. The properties of multi-agent systems (MAS), namely its nature of agents being autonomous, reactive to changes in the environment, proactive by being goal-driven, and social by being collaborative, fits well the needed paradigm to make BA systems flexible and responsive to needs. In particular, the possibility of resource constrained devices like sensors, actuators, and networked computation host to work in a collaborative manner under social organization rules to fulfil local and global goals is a topic of potential research (as compared to the established approaches where devices are not restricted by availability of energy or computational resources). Under such constraints, the goal of this research is to enable the collaboration among networked sensors in a smart way. Consequently, control agents are able to self-organize the automation system based on available resources. The PhD will be jointly supervised by Siemens and the University of Birmingham, based at the University but with opportunities for secondment to Siemens' data research groups in Switzerland.