Reliable PID dual-rate controller based on LoRaWAN for long-range distributed systems

Abstract

With the advances in digital communication systems, Network Control Systems (NSC) appeared as a feasible control solution for automation tasks. More recently, the exploitation of ubiquitous Internet connectivity for implementing a NCS has been suggested, paving the way for the Industrial Internet-of-Things paradigm. In particular, moving towards the Control as a Service (CaaS) would allow to benefit from the adaptability and scalability offered by cloud computing. Unfortunately, the needs of automation systems are very different from those dictated by the office and enterprise scenarios. The performance and availability of communication infrastructure used for connecting with the cloud can severely affect the control strategy effectiveness. This study assesses the implementation of a dual-rate controller, enhanced by an appropriate predictor, to facilitate NCS utilizing a potentially unstable LoRaWAN network. The selection of LoRaWAN, a type of Low Power Wide Area Network, is due to its extensive utilization in academic and industrial sectors. Moreover, the standard explicitly delineates backend entities, facilitating novel CaaS business models. A case study of a control system composed of a single LoRaWAN end-node device placed in a local plant to be controlled and connected to a single gateway network is employed to simulate the DR-PID control strategy on a second-order time-continuous system. The simulation model incorporates LoRaWAN characteristics, such as node-initiated transactions, and synthesizes measurable parameters, such as up-link and down-link losses. This model has been employed to assess control performance relative to a reference ideal situation in which communication losses are absent. A figure of merit is defined, enabling the assertion of the suggested approach’s superiority over the plain LoRaWAN NCS. Furthermore, the quasi-orthogonality of time and frequency superposed frames with varying Spreading Factors (SF) is proposed as an alternate method to leverage the enhanced noise immunity provided by elevated SF values.