Distributed Multi-Agent Architecture for Dependable Supervision over WSAN: Oil Refinery Tests

In supervision applications over wireless sensor and actuator networks the communication channels' reliability, along with the network quality of service and readings are critical issues. As such, in the design and implementation stages of such schemes preventive measures should be taken into account for the sake of control systems dependability. This paper proposes a multi-agent based framework implemented in some of the nodes of the network, for dealing with intermittent communication link breakdowns on the forward channel and accommodating outliers on the raw data taken from the environment. Experiments on a real oil re⬚nery process demonstrate the feasibility of such approaches in the context of critical industrial supervision tasks.