Interdependent distinguishing sensors and their applications on discrete event systems control

Abstract

A distinguishing sensor (DS) is a formal artefact which works virtually like a sensor. Its purpose is to provide details about certain events in a Discrete Event System (DES) according to their context of occurrence, which allows the coding of cognitive aspects of a DES. Sometimes, however, certain event distinctions may depend on other distinctions, provided by other DSs, in a context dependency chain that makes the design of the sensor itself complex. This paper presents the theoretical foundation for multiple interdependent DSs to be designed by Finite State Machines (FSM) and integrated within the same DES modelling scope. As this approach may result in very large models, we also show how to synthesise supervisors using the Local Modular Control (LMC) framework combined with abstraction-based synthesis and related extensions to accommodate interdependent DSs. After synthesis, large DS models usually overload also the implementation phase, since they have to be integrated with the controller. Therefore, this work also proposes a Internet of Things (IoT)-based implementation strategy that treats DSs as loose-coupled software artefacts that can be separated from the controller implementation. In this way, the unfolded composition of controllers and DSs is never actually needed, and the conceptual DSs act like physical sensors, informing the controller only and whenever required. A further contribution of this work is developing and testing a tool that converts FSMs controllers and DSs into implementable code C or Python, with support for both monolithic and modular synthesis. The approach is illustrated using a real dispatch control of a poultry company.