Elsevier

ISA Transactions

Volume 104, September 2020, Pages 101-114
ISA Transactions

Research article
State estimation for cyber–physical systems with limited communication resources, sensor saturation and denial-of-service attacks

https://doi.org/10.1016/j.isatra.2018.12.032Get rights and content

Highlights

  • The estimator design is proposed for CPSs with limited communication resources, sensor saturation and DoS attacks.

  • The event-triggered scheme and quantization mechanism are adopted to relieve the effects of limited communication resources.

  • A novel mathematical model of estimating CPSs with limited communication resources, sensor saturation and DoS attacks is constructed.

  • A new criterion for stabilization analysis of CPSs is obtained and the estimator gains are derived.

Abstract

This paper addresses the issue of the state estimation for cyber–physical systems (CPSs) with limited communication resources, sensor saturation and denial-of-service (DoS) attacks. In order to conveniently handle nonlinear term in CPSs, a Takagi–Sugeno (T–S) fuzzy model is borrowed to approximate it. The event-triggered scheme and quantization mechanism are introduced to relieve the effects brought by limited communication resources. By taking the influence of sensor saturation and DoS attacks into account, a novel mathematical model of state estimation for CPSs is constructed with limited communication resources. By using the Lyapunov stability theory, the sufficient conditions, which can ensure the system exponentially stable, are derived. Moreover, the explicit expressions of the event-based estimator gains are obtained in the form of linear matrix inequalities (LMIs). At last, a simulated example is provided for illustrating the effectiveness of the proposed method.

Introduction

Cyber–physical systems (CPSs) are regard as a class of the multi-dimensional complex systems which integrate the physical environment, network and computational capability. Recently, the modeling and analysis of CPSs have been paid widespread attentions because of the wide spectrum of applications in areas such as transportation networks, healthcare and power generation, which leads to numbers of publications. For example, the authors in [1] investigate the transportation-cyber–physical systems which are applied in the vehicle-to-infrastructure-based communications to avert traffic disasters and traffic jams. In [2], the nonlinear CPSs with multiple stochastic incomplete measurements are discussed where a T–S fuzzy model is used for approximating the physical process. T–S fuzzy model is firstly proposed in [3], which can be used to approximate nonlinear systems by a set of local linear systems smoothly connected by fuzzy membership functions. Since T–S fuzzy model is an effective method of modeling nonlinear systems, considerable attentions have been paid by scholars and lots of achievements have been acquired [4], [5], [6], [7], [8]. For instance, a class of T–S fuzzy systems is studied with the consideration of non-fragile distributed H filtering problem [4]. With the assistance of the T–S fuzzy model, the authors propose the algorithm of H filtering design for networked systems in [5]. The authors in [8] address the issue of continuous-time T–S fuzzy systems by using the time derivatives. Motivated by this work, T–S fuzzy model which can be utilized to approximate nonlinear systems by a series of local linear systems is borrowed to describe CPSs in this paper.

Due to the extensive attentions paid to CPSs, the problem of state estimation for CPSs has aroused considerable interests. Lots of scholars devote themselves into investigating the issue and numerous results are available. In [9], the state estimation of the wireless CPSs subjected to DoS attacks is investigated. The authors in [10] focus on the state estimation of CPSs with DoS attacks occurring in a wireless channel. A finite time event-triggered supervised estimator is designed to estimate linear CPSs under the event-driven scheme [11]. Apart from the above statement about estimating CPSs, the issue of sensor saturation has also fascinated plenty of academics for the reason that sensor saturation is likely to undermine the stability of CPSs to some degree. It need to be pointed out that the physical process is measured by a series of sensors where there always exists sensor saturation. In fact, many efforts have been devoted into researching sensor saturation for several years and massive achievements are obtained [12], [13]. By taking the sensor saturations into account, the distributed H filter is designed in [12] to estimate sensor networks. With the consideration of randomly occurring sensor saturation, the authors in [13] investigate the problem of the H estimator design of complex networks. It needs to be pointed out that the problem of the state estimation for CPSs has received wide attentions in the recent years, however, the research of estimating CPSs with sensor saturation has not been fully done yet, which initially motivates this paper.

In CPSs, owing to the limitation of bandwidth and communication resources, it may bring some problems which can degrade the system performance. How to handle the problems of limited communication resources has been widely discussed. During the past years, the periodic sampling (time-triggered scheme) is a frequently-used signal processing method, where the signals are sampled and transmitted at a fixed interval. However, the periodic sampling brings the issue of redundant data transmission, which results in the great network load. For the purpose of relieving the burden of communication network, various event-triggered schemes are proposed [14], [15], [16]. One of the famous event-triggered schemes is an event-triggered scheme relying on discrete supervision of the system states, which is firstly proposed in [16]. This event-triggered scheme only supervises the difference between the states sampled in discrete instants and the data can be transmitted only when they satisfy the given condition. Based on [16], numerous researches with event-triggered scheme are carried out during the last decade [17], [18], [19], [20], [21], [22], [23], [24], [25], [26], [27], [28]. For instance, the networked systems with multiple sensor distortions is investigated in [22] by utilizing the event-triggered scheme mentioned [16]. Inspired by [16], the authors propose an adaptive event-triggered communication mechanism where the assumption of the prior work is relaxed in [23]. On the basis of the event-triggered scheme presented in [16], the authors in [25] propose a new discrete event-triggered transmission protocol for networked control systems (NCSs) with delays and noise. Motivated by the work in [16], the resilient event-triggering communication scheme is proposed to improve the performance of NCSs with periodic DoS attacks in [27]. Based on the event-triggered scheme proposed in [16], the hybrid-driven scheme is proposed in [28], which combines the advantages of the time-triggered scheme and the event-triggered scheme. The authors in [29] address the problem of H filter design for neural networks with the hybrid-driven scheme. By adopting the hybrid-driven scheme, the issue of H control for networked cascade control systems is addressed in [30]. Besides, the hybrid-driven scheme proposed in [28] is utilized to study the T–S fuzzy systems [5], [31], [32]. In addition to the event-triggered mechanism as stated above, the quantization mechanism is also adopted to relieve the influence of limited communication resources [33], [34], [35], [36], [37], [38], [39], [40]. In [33], the authors use quantization mechanism to alleviate network transmission load of NCSs. A quantization mechanism is utilized to relieve the network bandwidth burden of Markovian jump systems in [36]. By using the quantization mechanism, the authors in [40] design the robust fuzzy-model-based filter for networked systems. Motivated by the above research, the event-triggered scheme and quantization mechanism are introduced in this paper to alleviate the effects of limited communication resources.

It is well known that the introduction of the network has brought many advantages to CPSs, however, it also may lead to the common communication constraints such as packet dropouts, limited bandwidth and time delay [41]. In addition, the network may be susceptible to cyber attacks due to the openness of communication networks [42]. In general, there are three kinds of cyber attacks including injection attacks, deception attacks and DoS attacks [29]. The injection attacks aim to replace the transmitted signal with the false data when the data is transmitted via the network. The deception attacks replace normal data by pretending to be trusted parties to degrade system performance. Different from the above attacks, the purpose of the DoS attacks is to make efforts to block data transmission. Recently, the cyber attacks have been paid considerable attentions by researchers and a great many results have been available in the literature. For instance, considering the influence of false data injection attacks, the issue of estimating the NCSs is addressed in [43]. From an attacker’s point of view, the optimal attacking region is determined while a false data injection attack takes place in a power grid [44]. The problem of H filter design for neural networks is investigated under the consideration of deception attacks [29]. By considering the effects of DoS attacks, the robust stabilization of NCSs with event-triggered scheme and quantization is studied in [26]. In [45], CPSs with DoS attacks are studied where the data is transmitted over the network from sensors to remote estimators. The event-triggered control model for CPSs subjected to DoS attacks is constructed in [46]. To the best of our knowledge, with the consideration of DoS attacks, the issue of state estimation for CPSs with limited communication resources and sensor saturation has been paid few attentions, which is the second motivation of this paper.

The rest of this paper is organized as follows. In Section 2, the mathematical models of considered problems are presented, respectively and system modeling is given. In Section 3, the sufficient conditions are acquired which can guarantee the exponential stability of the system and the desired estimator gains are accurately derived. In Section 4, a simulated example is supplied to illustrate the usefulness of designed estimator.

Notation

R>0, Rm and Rm×n represent the set of positive real scalars, the m-dimensional Euclidean space and the set of m×n real matrices, respectively; the superscript T denotes matrix transposition; I is the identity matrix with appropriate dimension; for XRm×m, the notation X>0 represents that the matrix X is real symmetric positive definite. For a matrix B and two symmetric matrices A and C, ABC stands for a symmetric matrix, where refers to the entries implied by symmetry. L2[0,) denotes the space of square-integrable vector functions defined on [0,). N denotes the set of non-negative integers. For a scalar ϵ>0, xtϵsupϵθ0{ẋ(t+θ),x(t+θ)}, where x(t) is a continuous function tk2,n denotes that tk2,n belongs to time interval [tk1,n,tk2,n).

Section snippets

System description

The framework of event-triggered state estimation for CPSs under sensor saturations, quantization and DoS attacks is shown in Fig. 1, where the data transmission between the quantizer and estimator is realized over the network with DoS attacks. In the following, we will give the mathematical model of the physical plant, event-triggered scheme, sensor saturation, quantizer and DoS attacks, respectively.

Consider the physical process which can be described as the following continuous-time

Main results

The performance of estimator error system (30) is discussed and the main results are exhibited in the following three theorems with their corresponding proofs. In Theorem 1, the sufficient conditions guaranteeing the system (30) exponentially stable are derived. Then the system performance of CPSs subjected to the external disturbance w(t) is discussed in Theorem 2. Based on the analysis in Theorem 1, Theorem 2, the gains of state estimator are obtained in Theorem 3.

Theorem 1

Given the sequence {nT}nN,

Simulation examples

In this section, a simulation example related to CPSs is presented to illustrate the feasibility of estimator design for CPSs, where the sensor saturation and DoS attacks are considered and the event-triggered scheme and quantization are introduced to relieve the influence brought by limited communication resources.

Consider the system (3) with the following system matrices: A1=2.50.10.21.5,A2=2.10.20.52B1=1.20.4,B2=10.2C1=1.20.8,C2=0.80.6L1=0.250.28,L2=0.20.3

In the following, we present

Conclusions

In this paper, a problem of estimator design has been investigated for CPSs with limited communication resources, sensor saturation and periodic DoS attacks. For the considered CPSs, a T–S fuzzy model is borrowed to described it. In order to relieve the influence of limited communication resources, the event-triggered scheme and quantization mechanism are adopted to reduce the amount of data transmission. A new mathematical model of quantized state estimation for CPSs is established, which

Acknowledgments

This work was supported in part by the Open Fund of Key Laboratory of Grain Information Processing and Control of Hennan Province of China under Grant KFJJ-2018-203, and in part by the Natural Science Foundation of the Jiangsu Higher Education Institutions of China under Grant 18 kJB120002, and in part by the Natural Science Foundation of Jiangsu Province of China under Grant BK20171481, and in part by the Key R&D Program (Industry Prospect and Common Key Technologies) of Jiangsu Province, China

References (50)

  • GuZ. et al.

    Event-triggered reliable H filter design for networked systems with multiple sensor distortions: A probabilistic partition approach

    ISA Trans

    (2017)
  • PengC. et al.

    Network-based H control for T-S fuzzy systems with an adaptive event-triggered communication scheme

    Fuzzy Sets and Systems

    (2017)
  • HuS. et al.

    Event-triggered H stabilization for networked stochastic systems with multiplicative noise and network-induced delays

    Inform Sci

    (2015)
  • ChenX. et al.

    Event-based robust stabilization of uncertain networked control systems under quantization and denial-of-service attacks

    Inform Sci

    (2018)
  • LiuJ. et al.

    Hybrid-driven-based H filter design for neural networks subject to deception attacks

    Appl Math Comput

    (2018)
  • ZhangD. et al.

    Analysis and synthesis of networked control systems: A survey of recent advances and challenges

    ISA Trans

    (2017)
  • ChengJ. et al.

    Quantized H filtering for switched linear parameter-varying systems with sojourn probabilities and unreliable communication channels

    Inform Sci

    (2018)
  • ZhangD. et al.

    Robust H control for nonhomogeneous Markovian jump systems subject to quantized feedback and probabilistic measurements

    J Franklin Inst B

    (2018)
  • ZhaL. et al.

    Event-based finite-time state estimation for Markovian jump systems with quantizations and randomly occurring nonlinear perturbations

    ISA Trans

    (2017)
  • ZhangD. et al.

    Robust fuzzy-model-based filtering for nonlinear networked systems with energy constraints

    J Franklin Inst B

    (2017)
  • ShiY. et al.

    Robust mixed H2H control of networked control systems with random time dalays in both forward and backward communication links

    Automatica

    (2011)
  • HuL. et al.

    State estimation under false data injection attacks: Security analysis and system protection

    Automatica

    (2018)
  • YuanH. et al.

    Resilient strategy design for cyber-physical system under DoS attack over a multi-channel framework

    Inform Sci

    (2018)
  • SunY. et al.

    Periodic event-triggered resilient control for cyber-physical systems under denial-of-service attacks

    J Franklin Inst B

    (2018)
  • ZhaL. et al.

    Decentralized event-triggered H control for neural networks subject to cyber-attacks

    Inform Sci

    (2018)
  • Cited by (73)

    • Secure control design for nonlinear cyber–physical systems under DoS, replay, and deception cyber-attacks with multiple transmission channels

      2022, ISA Transactions
      Citation Excerpt :

      Hence, DoS, replay and deception cyber-attacks uncertainties are different as shown in Figs. 6 and 7. In almost all research studies in CPSs assume a single transmitting channel [11–22,25–29,31–39]. Only few results assume multiple channels [23,24,30,40] but with one kind of cyber-attacks.

    • Leader-following consensus for multi-agent systems subject to cyber attacks: Dynamic event-triggered control

      2022, ISA Transactions
      Citation Excerpt :

      The security consensus issue subject to different kinds of the DOS attacks was considered in [31]. Inspired by the work [31], the strategic cyber attacks were discussed in [32] by establishing a hybrid random security control framework. People do a lot of in-depth research on cyber attacks, but there are few research results about cyber attacks for MASs under the triggering strategy.

    • A holistic review on Cyber-Physical Power System (CPPS) testbeds for secure and sustainable electric power grid – Part – I: Background on CPPS and necessity of CPPS testbeds

      2022, International Journal of Electrical Power and Energy Systems
      Citation Excerpt :

      Table 2 show various cyberattack instances that occurred in the electric power sector around the world recently. The data is gathered from numerous research works and technical reports [1,32,46-48,58,63,65,66,94] and put it in a graphical style depending on its severity and history to provide a thorough and succinct review of large blackouts caused by technical faults and cyber-attacks. Fig. 4 depicts a summary of these results.

    View all citing articles on Scopus
    View full text