Distributed multi-agent architecture for real-time wireless control networks of multiple plants

This paper describes an interesting model architecture based on the tree topology, as well as its analysis through an associated validation method for control networks designed for low-rate wireless sensor and actuator networks (WSAN), consisting of multiple control loops closed over a wireless communication network, taking advantage of the flexible communications features of IEEE 802.15.4 protocol. The main problem to be addressed is the control of multiple or parallel plants, which should have an intelligent and flexible architecture based on multi-agent systems (MAS), that allows to add or remove new plants or nodes online, without the need of reconfiguring the system, while maintaining temporal and functional restrictions in the system. Important issues to consider in the research are tackled herein like topology, optimal static routing, end-to-end test analysis of the real-time flows on the WSAN, real-time kernel with Earliest Deadline First (EDF) scheduler, periodic and aperiodic tasking models for the nodes, lightweight and flexible compensation-based control algorithms for WSAN that exhibit packet dropouts, and an event-triggered sampling scheme. Additionally, as example of an application in a simulation environment is presented, in which is intended to guarantee an optimal parallel fed-batch bioreactor operations model controlled over a WSAN, using a master-slave synchronization system as control strategy, obtaining an accurate end-to-end communication analysis of the proposed plant with Wireless Sensor and Actuator Networks.