A switched system approach to scheduling of networked control systems with communication constraints

This paper presents a scheduling strategy for a collection of discrete-time networked control systems (NCSs) subjected to communication constraints. Communication constraints under consideration include medium access constraint, network-induced delays and packet-dropouts. A feedback control system with the communication constraints is modelled as a switched delay system which switches on the open-loop and closed-loop models according to whether the feedback control loop gains access to the network or not. Delay-dependent sufficient conditions for exponential stability with L₂ gain performance are developed for the switched system. Based on the stability conditions, sufficient conditions are presented on the existence of scheduling policy that simultaneously stabilizes the collection of NCSs. Simulation on network-based control of unstable batch reactor systems is performed to demonstrated the effectiveness of the proposed scheduling strategy.

I. Introduction

Shared communication network is increasingly being used to support information exchange in control of a group of spatially distributed systems. This is very typical setups when using the base station to control and coordinate multiple mobile robots [1], unmanned aerial vehicles (UAVs) and autonomous air vehicles (AAVs) [2], through a wireless/wired network. Control systems whose control loops are closed over a wired or wireless communication network are known as networked control systems (NCSs). Motivation for using communication network in control comes from higher system testability and resource utilization, as well as lower cost, reduced weight and power, simpler installation and maintenance [3]. However, many practical networks are subjected to communication constraints that make the analysis and design of NCSs be typically harder than that of classical control systems. NCSs require novel control design methodologies that differ in nature from the tradition control viewpoint to account for the presence of communication constraints in the closed-loops. For example, (i) data packets transmitted over networks are usually subjected to random or time-varying delays, and/or even may be lost during the information transmission, see for example the recent servery paper [4] and references listed therein; and (ii) because of the limitation on channel capacity, the shared communication medium imposes access constraints: only a limited number of the network nodes are allowed to transmit their data packets at any time instant [3], [5], [6], [7].