Switched Sliding Mode Control Strategy for Networked Systems Antonella Ferrara, Gian Paolo Incremona and Veronica Stocchetti Abstract—In this paper, a networked switched control strat- egy based on Sliding Mode Control is presented. The idea pursued in this work is to reduce to a minimum the packet rate over the network, in order to limit the problems induced by the transmission of the state measurement between the sensor and the controller, while providing performance comparable with that of a non networked Sliding Mode Control scheme. The proposed scheme includes a model based controller which contains the nominal model of the plant, and relies on a suitably defined triggering condition. The latter considers the amplitude of a sliding variable determined relying on nominal model, and enables the actual state transmission only when the sliding variable is within a predefined boundary layer. When the plant state is not transmitted, the model state is used to determine the control action. In this way, it is possible to guarantee the same robustness with respect to matched uncertainties as in conventional sliding mode control schemes, as well as the exponential stability of the origin of the controlled system state space, even if the actual system state is not always used to close the feedback. Moreover, in steady-state, when the boundary layer is reached, in order to avoid a continuous transmission of the actual state measurement, a mechanism based on a moving average of the current sliding variable is adopted, which allows to suitably deactivate the state transmission even within the boundary layer, yet maintaining some robustness. Simulation results demonstrates the effectiveness of the proposed strategy. I. I NTRODUCTION Sliding Mode Control (SMC) is a widely appreciated strategy because of its capability of guaranteeing satisfactory performance of the controlled system under critical uncer- tainty conditions [1], [2], [3], [4]. For this reason, it can be regarded as a good candidate to be used in Networked Control Systems (NCSs), i.e., feedback systems including data networks. NCSs present several advantages compared with traditional configurations, such as reconfigurability, low installation costs, and the possibility to create a wide interconnected grid to transmit information [5]. However, the presence of the network in the control loop can cause the occurrence of packet loss, jitter, and delayed transmissions, which deteriorate the performance of control systems designed in the conventional way. As a consequence, suitable data communication protocols, new fault detection strategies, and control schemes designed explicitly taking into account the network presence have been proposed in the literature in recent years [5], [6], [7]. A. Ferrara, G. P. Incremona and V. Stocchetti are with the Dipartimento di Ingegneria Industriale e dell’Informazione, University of Pavia, via Ferrata 1, 27100 Pavia, Italy (e-mail: antonella.ferrara@unipv.it, {gp.incremona, veronica.stocchetti}@gmail.com). This is the final version of the accepted paper submitted for inclusion in the Proceedings of the IEEE Mediterranean Control Conference, Palermo, Italy, Jun. 2014. The aim of this paper is the design of a networked control scheme, which is able to solve the trade-off between the stability of the controlled system and the bandwidth allocation over the network [8]. In order to obtain satisfactory performance, in this framework, the use of a robust control strategy is mandatory. In this paper, the SMC methodology is adopted to design the controller. This is of model based type [9], [10], so that, relying on a suitably defined event triggered strategy [11], [12], [13], [14], [15], [16], it is possible to switch between the use of the actual plant state and of the model state to close the feedback. More specifically, the model based controller contains three elements: the controller itself, the nominal model of the plant and a triggering condition. The triggering condition verifies if a sliding variable, determined relying on the nominal model state, belongs to a predefined boundary layer, and enables the actual state transmission only when this occurs. When the condition does not hold, then the model state is used to determine the control action. As a result, the state of the nominal model is used during the reaching phase, i.e. the phase during which the sliding variable is steered to the boundary layer, while the actual plant state is used within the boundary layer. This could cause an intensification of state transmission in steady-state. To circumvent this drawback, and to keep the packet rate to a minimum, a mechanism based on a moving average of the current sliding variable is adopted. This mechanism enables to deactivate the state transmission even within the boundary layer. Indeed, if the moving average remains almost constant for a certain number of sampling time instants, then the model state is used again. The theoretical assessment of the proposed scheme is provided in the paper, by addressing both robustness and stability issues. Assuming that the deactivation mechanism is switched off, our proposal proves to maintain the same robustness property with respect to matched uncertainties as conventional SMC. Moreover, the exponential stability of the origin of the controlled system state space can be proved, even if the actual system state is not always used to close the feedback, which implies a clear benefit in terms of bandwidth allocation. When the deactivation mechanism is on, then the robustness versus matched uncertainties is attenuated. Yet, it can be proved that the effect of such uncertainties is bounded. Note that, in this scheme, a traditional sensor, with no particular computational capability, is required. Moreover, the network mathematical model is not considered in the design. Finally, for the sake of simplicity, the network presence is assumed only between the sensor and the model based controller, since, the controlled system being single-input, the