Self-triggered prescribed-time time varying formation control of second-order multi-agent systems with denial of service (DoS) attacks Authors: Manmohan Sharma, Harshavarthini Shanmugam, Soumya Ranjan Mahapatro Published: 2026 ABSTRACT A prescribed-time time varying formation control of second-order multi-agent systems with DoS attacks is proposed in this article. Firstly, a prescribed-time control law is proposed under time-triggered scenario when the agents are affected with DoS attacks. This is further extended to the self-triggered scenario under DoS attacks. An undirected graph and time varying formation is considered in developing the control law. Rigorous mathematical proof is given to validate the control law. Simulation results in both time-triggered and self-triggered scenario under DoS attacks is presented to highlight the effectiveness of the approach in this article. Comparison results are also shown the bandwidth efficiency of the proposed approach. Introduction Cooperative control of multi-agent systems has gained considerable interest due to its wide applications in tasks such as formation flying, coordinated robotics, and sensor networks. Among these, formation control stands out as a critical area, aiming to develop control algorithms that enable a group of agents to maintain a prescribed geometric configuration in space. A key performance metric in formation control is the fast and reliable convergence rate, which directly influences the responsiveness and effectiveness of the system. Early methods achieved only asymptotic convergence, resulting in infinite convergence time1–3, which is impractical for time-sensitive tasks. To overcome this, finite-time convergence strategies were developed4–7, though their convergence time depends on initial conditions. Subsequently, fixed-time formation control methods emerged to guarantee convergence within a uniform time bound, independent of initial states8,9. While fixed-time control offers uniform convergence, it still lacks flexibility since the convergence time cannot be preassigned. To overcome these limitations, prescribed-time control has emerged as a powerful alternative10–13. It allows the user to specify the exact settling time in advance, regardless of the initial conditions or system parameters. Furthermore, it ensures smooth control inputs, exhibits strong robustness to external disturbances, and does not require knowledge of bounds on non-vanishing perturbations. These favorable properties have motivated extensive research into prescribed-time control and its applications11–16. In recent years, increasing attention has been given to formation control of multi-agent systems under Denial-of-Service (DoS) attacks, particularly through event-triggered control to reduce communication load and enhance resilience. Various strategies have been proposed, including finite-time17 and dynamic event-triggered approaches for mobile robots18,19, intelligent vehicles, and unmanned surface or aerial vehicles20. Some studies also address additional challenges such as actuator faults, communication constraints, and system uncertainties using tools like neural networks and model predictive control21–25. A time-varying formation tracking (TVFT) control method is developed in23 for multi-uncrewed aerial vehicles (UAVs) with unknown nonlinearities, while24 addresses secure formation tracking with communication bandwidth optimization under DoS constraints. A fault-tolerant dynamic event-triggered control scheme is proposed in25 for unmanned surface vehicles (USVs) experiencing both actuator faults and DoS attacks. In26, event-triggered consensus control is investigated for deterministic nonlinear MASs under DoS scenarios. The problem of distributed formation tracking control for second-order multiagent systems with unknown inertias is studied in27 while formation control in presence of disturbances and input constraints is studied in28. Additionally,29 introduces a fault-tolerant control strategy for second-order MASs with prescribed-time convergence. Moreover, in30 solves time-varying formation control with finite-time prescribed performance for nonstrict feedback second-order multi-agent systems. Furthermore, other types of attacks affecting nonlinear systems in event-triggered scenario has been compensated in31–33. Although existing works on formation control under Denial-of-Service (DoS) attacks improve resilience and reduce communication overhead, most approaches guarantee only asymptotic or finite-time convergence, with performance often dependent on initial conditions and system parameters. In particular, user-defined convergence time is rarely achieved in the presence of DoS attacks and system uncertainties. Moreover, fault-tolerant designs that also consider communication bandwidth constraints remain limited in scope. These observations highlight the need for a robust and efficient prescribed-time time varying formation control framework that ensures reliable convergence within a time defined by the user, irrespective of the initial conditions, while also addressing actuator faults and preserving communication resources. Motivated by this gap, the present article proposes a prescribed-time time-varying formation control strategy for second-order multi-agent systems under DoS attacks, with an emphasis on bandwidth efficiency. The developed self-triggered strategy obtains the next triggering instant from the last triggering instant and hence avoids continuous monitoring of the states. This is the shortcoming of the existing event-triggered strategies where continuous monitoring of states are required. The proposed self-triggered strategy saves the communication bandwidth in both sensor-controller channel as well as controller-sensor channel. Moreover, the prescribed-time convergence property not only enhances robustness but also allows for faster mitigation of DoS effects by guaranteeing precise timing in the recovery and formation processes. The main contributions of this article are summarized as follows:Development of a new prescribed-time time-varying formation control algorithm for a second-order multi-agent system while saving bandwidth. The second-order multi-agent systems are more practical and realistic and therefore, second-order multi-agent systems are used in this study.A self-triggered control mechanism is introduced to eliminate the need for continuous monitoring of neighbors’ states, significantly reducing bandwidth consumption.A comprehensive stability and convergence analysis is provided, with rigorous mathematical proofs confirming prescribed-time convergence of the closed-loop system under DoS attacks. Development of a new prescribed-time time-varying formation control algorithm for a second-order multi-agent system while saving bandwidth. The second-order multi-agent systems are more practical and realistic and therefore, second-order multi-agent systems are used in this study. Development of a new prescribed-time time-varying formation control algorithm for a second-order multi-agent system while saving bandwidth. The second-order multi-agent systems are more practical and realistic and therefore, second-order multi-agent systems are used in this study. A self-triggered control mechanism is introduced to eliminate the need for continuous monitoring of neighbors’ states, significantly reducing bandwidth consumption. A self-triggered control mechanism is introduced to eliminate the need for continuous monitoring of neighbors’ states, significantly reducing bandwidth consumption. A comprehensive stability and convergence analysis is provided, with rigorous mathematical proofs confirming prescribed-time convergence of the closed-loop system under DoS attacks. A comprehensive stability and convergence analysis is provided, with rigorous mathematical proofs confirming prescribed-time convergence of the closed-loop system under DoS attacks. The remainder of the paper is arranged as follows. Section II defines the preliminary concepts required and utilized in this article. Section III outlines the proposed controller design along with the stability analysis. Simulation and comparison results are given in Section IV. Lastly, conclusions and future works are illustrated in Section V. Preliminaries System description In this article, an N number of multi-agent systems with second-order dynamics are considered i.e.1\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\begin{aligned}&{\dot{x}}_i=v_i \nonumber \\&{\dot{v}}_i=u_i, \text { where, } i=1,2,\ldots ,N \end{aligned}$$\end{document}\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$x_i,v_i \in {\mathbb {R}}$$\end{document}. If they are in p>1 dimensions, the control law can be appropriately modified by addition of Kronecker product. In this article self-triggered formation control of above mentioned second-order dynamics with DoS attacks is considered. To be precise, a control algorithm is designed such that :2\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\begin{aligned}&||x_i(t)-x_j(t)-d_{ji}|| \rightarrow 0 ~~\forall t \in [T_p,\infty ), ~~\forall i,j \in 1,2, \cdots , N \nonumber \\&||v_i(t)-v_j(t)-{\dot{d}}_{ji}|| \rightarrow 0 ~~\forall t \in [T_p,\infty ), ~~\forall i,j \in 1,2, \cdots , N \end{aligned}$$\end{document} All of the above signals are bounded for \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$t \in [0,T_p)$$\end{document}. Definition 1 34 The system \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\dot{x}}=f(x,t)$$\end{document} is said to be prescribed-time globally asymptotically stable in time \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$T_p$$\end{document} if there exist a function \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\mu (t): [0,T_p) \rightarrow {\mathbb {R}}_+$$\end{document} with μ increasing to ∞ as \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$t \rightarrow T_p$$\end{document} and a class \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${{\mathcal {K}}}{{\mathcal {L}}}$$\end{document} function β such that,3\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\begin{aligned}&||x(t)|| \le \beta (||x(0)||,\mu (t)),~~ \forall ~~t \in [0,T_p), \end{aligned}$$\end{document}where \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$T_p$$\end{document} is a finite time number prescribed in the design. The monotonically increasing function μ (t) is given by :4\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\begin{aligned}&\mu (t)=\frac{T_p}{T_p-t},~~t \in [0,T_p) \end{aligned}$$\end{document}where \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$T_p > 0$$\end{document}, with the properties that μ (0)=1 and \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\mu (T_p)=+\infty$$\end{document}. Definition 2 11 For the system \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\dot{x}}=f(x,t)$$\end{document}, consider a continuously differential function positive definite \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$V_t(x(t),t)$$\end{document}. If the derivative with respect to time of \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$V_t(x(t),t)$$\end{document} follows5\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\begin{aligned}&{\dot{V}}_t \le -bV_t-2\frac{{\dot{\mu }}}{\mu }V_t \end{aligned}$$\end{document}for real constant b>0 on [0,∞ ), then the system is globally prescribed-time stable with prescribed-time \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$T_p$$\end{document} given by (4). Moreover, for \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$t \in [0,T_p)$$\end{document}, it follows that \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$V_t(t) \le \mu ^{-2}e^{-bt}V_t(0)$$\end{document}. In addition, \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$V_t(t) \equiv 0$$\end{document} for \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$t \in [T_p,\infty )$$\end{document}. Lemma 1 For the system \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\dot{x}}=f(x,t)$$\end{document} , consider a continuously differential function positive definite \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$V_t(x(t),t)$$\end{document} . If the derivative with respect to time of \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$V_t(x(t),t)$$\end{document} follows 6\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\begin{aligned}&{\dot{V}}_t \le bV_t+\frac{{\dot{\mu }}}{\mu }V_t \end{aligned}$$\end{document} for real constant b>0 on [0,∞ ) , then \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$V_t(t) \le \mu e^{bt}V_t(0)$$\end{document} for \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$t \in [0,T_p).$$\end{document} Proof First observe that7\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\begin{aligned}&\frac{d}{dt}\left( \frac{V_t}{\mu }\right) =\frac{{\dot{V}}_t}{\mu }-\frac{{\dot{\mu }}}{\mu ^2}V_t \le b \left( \frac{V_t}{\mu }\right) \end{aligned}$$\end{document} Solving the above differential inequality gives :8\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\begin{aligned}&\frac{V_t}{\mu } \le e^{b(t-t_0)}\frac{V_t(t_0)}{\mu (t_0)}=e^{b(t-t_0)}V_t(t_0) \nonumber \\&\implies V_t(t) \le \mu e^{b(t-t_0)}V_t(t_0) \end{aligned}$$\end{document}□ Lemma 2 Assume the differential of a vector quantity, \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$Z_t(t)$$\end{document} satisfies the following inequality: 9\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\begin{aligned} {\dot{Z}}_t(t) \le -AZ_t(t)+l \end{aligned}$$\end{document} Given that the differential equation10\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\begin{aligned} {\dot{Z}}_t(t) \le -AZ_t(t) \end{aligned}$$\end{document}is prescribed-time stable, then the (9) is also prescribed-time stable if l>0 is a constant and bounded. Proof Solving the differential inequality (9) gives :11\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\begin{aligned} Z_t(t) \le e^{-A(t)}Z_t(0)+le^{-A(t)}\int _0^t e^{A(s)}ds \end{aligned}$$\end{document} From assumption, \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\text {lim}_{t \rightarrow T_p} A(t) = +\infty$$\end{document} and also \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\text {lim}_{t \rightarrow T_p} e^{-A(t)}\int _0^t e^{A(s)}ds = 0$$\end{document}. Hence \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\text {lim}_{t \rightarrow T_p} Z_t = 0$$\end{document}. □ Denial of service attacks DoS attacks affect the communication channel between the agents. In this article it is assumed that the attacks result in loss of transmission of information to the agents connected to the channel35,36. This means that the topology of the communication link is changed under DoS attacks. In this work it is assumed that the undirected graph without DoS attacks contains a spanning tree. A DoS attack is active for a time of \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\Delta _a$$\end{document} during the time interval [τ ,t]. That is DoS attack time is given by \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$[\tau ,\tau +\Delta _a]$$\end{document} while the inactive time is given by \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$[\tau +\Delta _a,t]$$\end{document} which is denoted by the sleep time \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\Delta _s$$\end{document}. Each attack duration results in different graph topology which may contain more than one connected components. In other words, the graph topology may not contain a spanning tree during the attack period. Proposed controller and stability analysis Let \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$X(t)=[x_1(t),x_2(t),\cdots ,x_N(t)]$$\end{document}, \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$V(t)=[v_1(t),v_2(t),\cdots ,v_N(t)]$$\end{document}, \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$U(t)=[u_1(t),u_2(t),\cdots ,u_N(t)]$$\end{document}, \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$D(t)=[d_1(t),d_2(t),\cdots ,d_N(t)]$$\end{document}, then dynamic equations (1) is simplified as :12\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\begin{aligned}&{\dot{X}}(t)=V(t) \nonumber \\&{\dot{V}}(t)=U(t), \end{aligned}$$\end{document} Before proceeding let us define following notation \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\tilde{X}}=[x_1-d_1,x_2-d_2,\cdots ,x_N-d_N]$$\end{document} where \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$d_i$$\end{document} is the desired position of agent i. \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\tilde{V}}=[v_1-{\dot{d}}_1,v_2-{\dot{d}}_2,\cdots ,v_N-{\dot{d}}_N]$$\end{document}. The following proposition defines the control input U for the closed loop system (12) to be prescribed-time stable. Proposition 1 The following control law U given in (13) results in prescribed-time stability of the closed-loop system (12) if the condition in (14) is satisfied.13\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\begin{aligned}&U(t)=\left( k+c\frac{{\dot{\mu }}}{\mu }\right) (-{\mathcal {L}}{\tilde{X}}-k_1{\mathcal {L}}{\tilde{V}})+{\ddot{D}} \end{aligned}$$\end{document}14\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\begin{aligned}&d\lambda _2({\mathcal {L}})\left( k_1\lambda _2({\mathcal {L}})-\frac{d}{\left( k+c\frac{{\dot{\mu }}}{\mu }\right) }\right) -\frac{N}{4\left( k+c\frac{{\dot{\mu }}}{\mu }\right) ^2}\nonumber \\&+0.5(1+dk_1) \sum d_{in}-\frac{(1+dk_1)^2}{4} \sum d_{in}^2 > 0 \end{aligned}$$\end{document}where \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$d_{in}$$\end{document} is the in-degree of the vertices. The condition (14) will be satisfied if the undirected graph has spanning tree. Proof Consider the following dynamics:15\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\begin{aligned}&\dot{{\tilde{X}}}={\tilde{V}} \end{aligned}$$\end{document}16\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\begin{aligned}&\dot{{\tilde{V}}}=U(t)-{\ddot{D}} \end{aligned}$$\end{document} Substituting from (13), one gets :17\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\begin{aligned}&\dot{{\tilde{X}}}={\tilde{V}} \end{aligned}$$\end{document}18\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\begin{aligned}&\dot{{\tilde{V}}}=\left( k+c\frac{{\dot{\mu }}}{\mu }\right) (-{\mathcal {L}}{\tilde{X}}-k_1{\mathcal {L}}{\tilde{V}}) \end{aligned}$$\end{document} To proceed with the proof let us consider the following Lyapunov function \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$V_f$$\end{document} as:19\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\begin{aligned} V_f&=\frac{1}{2}{\tilde{X}}^T{\tilde{X}}+\frac{1}{2}{\tilde{V}}^T{\tilde{V}}+d{\tilde{X}}^T{\tilde{V}} \end{aligned}$$\end{document}where d<1, a constant and \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$||\rho ||_F$$\end{document} is the Frobenius norm of ρ. The Lyapunov function \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$V_f$$\end{document} satisfies the following inequality :20\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\begin{aligned} \lambda _m(W_1)||z||^2 \le V_f \le \lambda _M(W_2)||z||^2 \end{aligned}$$\end{document}where\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\begin{aligned} W_1= \left[ \begin{array}{cc} \frac{1}{2} & -\frac{d}{2} \\ -\frac{d}{2} & \frac{1}{2} \end{array} \right] ~~ W_2= \left[ \begin{array}{cc} \frac{1}{2} & \frac{d}{2} \\ \frac{d}{2} & \frac{1}{2} \end{array} \right] \end{aligned}$$\end{document} Taking the time derivative of the above equation, we get :\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\begin{aligned}&{\dot{V}}_f={\tilde{X}}^T{\tilde{V}}+{\tilde{V}}^T(U-{\ddot{D}})+d{\tilde{X}}^T (U-{\ddot{D}})+d{\tilde{V}}^T{\tilde{V}} \nonumber \\&={\tilde{X}}^T{\tilde{V}}+{\tilde{V}}^T\left( \left( k+c\frac{{\dot{\mu }}}{\mu }\right) \left( -{\mathcal {L}}{\tilde{X}}-k_1{\mathcal {L}}{\tilde{V}}\right) \right) \nonumber \\&+d{\tilde{X}}^T \left( \left( k+c\frac{{\dot{\mu }}}{\mu }\right) \left( -{\mathcal {L}}{\tilde{X}}-k_1{\mathcal {L}}{\tilde{V}}\right) \right) +d{\tilde{V}}^T{\tilde{V}} \nonumber \\&=-d\left( k+c\frac{{\dot{\mu }}}{\mu }\right) {\tilde{X}}^T{\mathcal {L}}{\tilde{X}}-k_1 \left( k+c\frac{{\dot{\mu }}}{\mu }\right) {\tilde{V}}^T{\mathcal {L}}{\tilde{V}} +{\tilde{X}}^T{\tilde{V}} \nonumber \\&-\left( k+c\frac{{\dot{\mu }}}{\mu }\right) {\tilde{X}}^T{\mathcal {L}}V-dk_1 \left( k+c\frac{{\dot{\mu }}}{\mu }\right) {\tilde{X}}^T ({\mathcal {L}}){\tilde{V}}+d{\tilde{V}}^T{\tilde{V}} \nonumber \\&=-d\left( k+c\frac{{\dot{\mu }}}{\mu }\right) {\tilde{X}}^T{\mathcal {L}}{\tilde{X}}-k_1 \left( k+c\frac{{\dot{\mu }}}{\mu }\right) {\tilde{V}}^T{\mathcal {L}}{\tilde{V}} \nonumber \\&+{\tilde{X}}^T\left( I-\left( k+c\frac{{\dot{\mu }}}{\mu }\right) ({\mathcal {L}}+dk_1{\mathcal {L}})\right) {\tilde{V}}+d{\tilde{V}}^T{\tilde{V}} \nonumber \\&\le -d\left( k+c\frac{{\dot{\mu }}}{\mu }\right) \lambda _2({\mathcal {L}})||{\tilde{X}}||^2-\left( k_1\left( k+c\frac{{\dot{\mu }}}{\mu }\right) \lambda _2({\mathcal {L}})-d\right) \nonumber \\&||{\tilde{V}}||^2+||I-\left( k+c\frac{{\dot{\mu }}}{\mu }\right) ({\mathcal {L}}+dk_1{\mathcal {L}}||||{\tilde{X}}||||{\tilde{V}}|| \nonumber \end{aligned}$$\end{document}21\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\begin{aligned}&\le -\left( k+c\frac{{\dot{\mu }}}{\mu }\right) \lambda _m(W)||z||^2 \nonumber \\&\le -\left( k+c\frac{{\dot{\mu }}}{\mu }\right) \frac{\lambda _m(W)}{\lambda _M(W_2)}V_f \end{aligned}$$\end{document}where W is shown in (22).22\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\begin{aligned} W= \left[ \begin{array}{cc} d\lambda _2(\mathcal {L}) & -\frac{1}{2}||\frac{1}{\left( k+c\frac{\dot{\mu }}{\mu }\right) }I-(\mathcal {L}+dk_1\mathcal {L})|| \\ -\frac{1}{2}||\frac{1}{\left( k+c\frac{\dot{\mu }}{\mu }\right) }I-(\mathcal {L}+dk_1\mathcal {L})|| & k_1\lambda _2(\mathcal {L})-\frac{d}{\left( k+c\frac{\dot{\mu }}{\mu }\right) } \end{array} \right] \end{aligned}$$\end{document} From Definition 2, \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\text {lim}_{t \rightarrow T_p} V_f = 0$$\end{document}, if W is positive definite and hence the closed loop system will be prescribed-time stable. Moreover, from Definition 2, for \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$t \in [0,T_p),$$\end{document}23\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\begin{aligned} V_f(t) \le \mu ^{-2}e^{-b_1|\Delta _s|}V_f(0) \end{aligned}$$\end{document}where \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$b_1=k\frac{\lambda _m(W)}{\lambda _M(W_2)}$$\end{document}. The matrix W will be positive definite if24\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\begin{aligned}&d\lambda _2({\mathcal {L}})\left( k_1\lambda _2({\mathcal {L}})-\frac{d}{\left( k+c\frac{{\dot{\mu }}}{\mu }\right) }\right) \nonumber \\&-\frac{1}{4}||\frac{1}{\left( k+c\frac{{\dot{\mu }}}{\mu }\right) }I-({\mathcal {L}}+dk_1{\mathcal {L}})||^2 > 0 \end{aligned}$$\end{document} The last term can be written as:25\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\begin{aligned}&||\frac{1}{\left( k+c\frac{{\dot{\mu }}}{\mu }\right) }I-({\mathcal {L}}+dk_1{\mathcal {L}})||^2 =tr\left( \left( \frac{1}{\left( k+c\frac{{\dot{\mu }}}{\mu }\right) }I\right. \right. -\nonumber \\&\left. \left. \left( {\mathcal {L}}+dk_1{\mathcal {L}}\right) \right) ^T \left( \frac{1}{\left( k+c\frac{{\dot{\mu }}}{\mu }\right) }I- ({\mathcal {L}}+dk_1{\mathcal {L}})\right) \right) \nonumber \\&=tr\left( \frac{1}{\left( k+c\frac{{\dot{\mu }}}{\mu }\right) ^2} I-(1+dk_1)\frac{{\mathcal {L}}}{\left( k+c\frac{{\dot{\mu }}}{\mu }\right) }\right. \nonumber \\&\left. -(1+dk_1)\frac{{\mathcal {L}}^T}{\left( k+c\frac{{\dot{\mu }}}{\mu }\right) } +(1+dk_1)^2{\mathcal {L}}^T{\mathcal {L}}\right) \end{aligned}$$\end{document} Therefore (25) becomes less than\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\begin{aligned}&\frac{N}{\left( k+c\frac{{\dot{\mu }}}{\mu }\right) ^2}-\frac{(1+dk_1)}{\left( k+c\frac{{\dot{\mu }}}{\mu }\right) }tr({\mathcal {L}}+{\mathcal {L}}^T) \nonumber \\&+(1+dk_1)^2 tr({\mathcal {L}}^T{\mathcal {L}}) \nonumber \\&\le \frac{N}{\left( k+c\frac{{\dot{\mu }}}{\mu }\right) ^2}-2(1+dk_1) \sum d_{in}+(1+dk_1)^2 \sum d_{in}^2 \nonumber \end{aligned}$$\end{document} Therefore the matrix W is positive definite if (14) is satisfied and the undirected graph has a spanning tree. Remark 1 At first it looks like one needs the information on desired positions to implement this formation controller. But the above controller can be decomposed in this form where only desired inter-agent distances are required.26\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\begin{aligned}&u_i(t)=\left( k+c\frac{{\dot{\mu }}}{\mu }\right) \sum _{j \in {\mathcal {N}}_i}a_{ij} \left( (x_j-x_i-d_{ji})\right. \nonumber \\&\left. +k_1(v_j-v_i-{\dot{d}}_{ji})\right) +\ddot{d}_i \end{aligned}$$\end{document}where \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\mathcal {N}}_i$$\end{document} are the set of neighbors of agent i and \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$a_{ij}$$\end{document} are the elements of the adjacency matrix. By replacing \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$x_j-x_i-d_{ji}$$\end{document} with \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$(x_j-d_j-x_i+d_i)$$\end{document} one obtains (13). Any set of \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$d_i$$\end{document} and \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$d_j$$\end{document} satisfying desired inter-agent distance \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$d_{ji}=d_j-d_i$$\end{document} will work. Remark 2 At first look it may seem that condition (14) is impossible to satisfy but one cursory look reveals the fact that the equation is satisfied for high values of \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$k_1$$\end{document}. □ Proposition 2 37 In the presence of DoS attack, the above control law results in formation control if the duration of the attacks satisfy:27\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\begin{aligned} |\Delta _a| \le \eta +\frac{t-t_0}{\theta } \end{aligned}$$\end{document}where \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\eta \ge 0$$\end{document} and θ > 1. The above assumption is required to limit the duration of DoS attacks otherwise the proposed system may operate in open-loop all the time resulting in loss of controllability. Proof Let the proposed multi-agent systems are operating under DoS attacks and the resulting topology has Laplacian matrix given by \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\mathcal {L}}_d$$\end{document}. Proceeding as in Proposition 1, we have:\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\begin{aligned}&{\dot{V}}_f=-d\left( k+c\frac{{\dot{\mu }}}{\mu }\right) {\tilde{X}}^T{\mathcal {L}}_d{\tilde{X}}-k_1\left( k+c\frac{{\dot{\mu }}}{\mu }\right) {\tilde{V}}^T{\mathcal {L}}_d{\tilde{V}} \nonumber \\&+{\tilde{X}}^T\bigg (I-\left( k+c\frac{{\dot{\mu }}}{\mu }\right) ({\mathcal {L}}_d+dk_1{\mathcal {L}}_d)\bigg ){\tilde{V}}+d{\tilde{V}}^T{\tilde{V}} \nonumber \end{aligned}$$\end{document} Taking \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\mathcal {L}}_d={\mathcal {L}}_d+{\mathcal {L}}-{\mathcal {L}}$$\end{document}, we have:28\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\begin{aligned}&{\dot{V}}_f=-d\left( k+c\frac{{\dot{\mu }}}{\mu }\right) {\tilde{X}}^T({\mathcal {L}}_d+{\mathcal {L}}){\tilde{X}}-k_1\left( k+c\frac{{\dot{\mu }}}{\mu }\right) {\tilde{V}}^T\nonumber \\&({\mathcal {L}}_d+{\mathcal {L}}){\tilde{V}} +{\tilde{X}}^T\bigg (I-\left( k+c\frac{{\dot{\mu }}}{\mu }\right) ({\mathcal {L}}_d+{\mathcal {L}}+dk_1({\mathcal {L}}_d+{\mathcal {L}}))\bigg ){\tilde{V}}\nonumber \\&+d{\tilde{V}}^T{\tilde{V}} +d\left( k+c\frac{{\dot{\mu }}}{\mu }\right) {\tilde{X}}^T({\mathcal {L}}_d+{\mathcal {L}}){\tilde{X}}+k_1\left( k+c\frac{{\dot{\mu }}}{\mu }\right) {\tilde{V}}^T\nonumber \\&({\mathcal {L}}_d+{\mathcal {L}}){\tilde{V}} +{\tilde{X}}^T\bigg (\left( k+c\frac{{\dot{\mu }}}{\mu }\right) ({\mathcal {L}}_d+{\mathcal {L}}+dk_1({\mathcal {L}}_d+{\mathcal {L}}))\bigg ){\tilde{V}} \nonumber \\&\le -d\left( k+c\frac{{\dot{\mu }}}{\mu }\right) \lambda _2({\mathcal {L}}_d+{\mathcal {L}})||{\tilde{X}}||^2-k_1\left( k+c\frac{{\dot{\mu }}}{\mu }\right) \lambda _2({\mathcal {L}}_d+{\mathcal {L}})\nonumber \\&||{\tilde{V}}||^2 +||{\tilde{X}}||||I-\left( k+c\frac{{\dot{\mu }}}{\mu }\right) ({\mathcal {L}}_d+{\mathcal {L}}+dk_1({\mathcal {L}}_d+{\mathcal {L}}))||||{\tilde{V}}|| \nonumber \\&+d||{\tilde{V}}||^2+d\left( k+c\frac{{\dot{\mu }}}{\mu }\right) \lambda _M({\mathcal {L}}_d+{\mathcal {L}})||{\tilde{X}}||^2+k_1\left( k+c\frac{{\dot{\mu }}}{\mu }\right) \nonumber \\&+\lambda _M({\mathcal {L}}_d+{\mathcal {L}})||{\tilde{V}}||^2 ||{\tilde{X}}||\left( k+c\frac{{\dot{\mu }}}{\mu }\right) \nonumber \\&||({\mathcal {L}}_d+{\mathcal {L}}+dk_1({\mathcal {L}}_d+{\mathcal {L}}))||||{\tilde{V}}|| \nonumber \\&\le -\left( k+c\frac{{\dot{\mu }}}{\mu }\right) \lambda _m(W_3)||z||^2+\left( k+c\frac{{\dot{\mu }}}{\mu }\right) \lambda _M(W_4)||z||^2 \nonumber \\&\le -\left( k+c\frac{{\dot{\mu }}}{\mu }\right) \frac{\lambda _m(W)}{\lambda _M(W_2)}V_f+\left( k+c\frac{{\dot{\mu }}}{\mu }\right) \frac{\lambda _M(W_4)}{\lambda _m(W_1)}V_f \nonumber \\&\le \left( k+c\frac{{\dot{\mu }}}{\mu }\right) \bigg (\frac{\lambda _M(W_4)}{\lambda _m(W_1)}-\frac{\lambda _m(W)}{\lambda _M(W_2)}\bigg )V_f \end{aligned}$$\end{document}where \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$W_3$$\end{document} and \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$W_4$$\end{document} are shown in (29) and (30)29\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\begin{aligned}&W_3= \left[ \begin{array}{cc} d\lambda _2(\mathcal {L}+\mathcal {L}_d) & -\frac{1}{2}||\frac{1}{(k+c\frac{\dot{\mu }}{\mu })}I-(1+dk_1)(\mathcal {L}+\mathcal {L}_d)|| \\ -\frac{1}{2}||\frac{1}{(k+c\frac{\dot{\mu }}{\mu })}I-(1+dk_1)(\mathcal {L}+\mathcal {L}_d)|| & k_1\lambda _2(\mathcal {L}+\mathcal {L}_d)-\frac{d}{(k+c\frac{\dot{\mu }}{\mu })} \end{array} \right] \end{aligned}$$\end{document}30\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\begin{aligned}&W_4= \left[ \begin{array}{cc} d\lambda _M({\mathcal {L}}) & \frac{1}{2}(1+dk_1)||{\mathcal {L}}|| \\ \frac{1}{2}(1+dk_1)||{\mathcal {L}}|| & k_1\lambda _M({\mathcal {L}}) \end{array} \right] \end{aligned}$$\end{document}□ Moreover, from Lemma 1, for \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$t \in [0,T_p),$$\end{document}31\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\begin{aligned} V_f(t) \le \mu e^{b_2|\Delta _a|}V_f(0) \end{aligned}$$\end{document}where \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$b_2=k\bigg (\frac{\lambda _M(W_4)}{\lambda _m(W_2)}-\frac{\lambda _m(W)}{\lambda _M(W_2)}\bigg )$$\end{document} and \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$c\bigg (\frac{\lambda _M(W_4)}{\lambda _m(W_2)}-\frac{\lambda _m(W)}{\lambda _M(W_2)}\bigg )=1$$\end{document}. From Proposition 1, for interval \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$[t_0,t]$$\end{document}, we have:32\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\begin{aligned} V_f(t) \le \mu ^{-1}e^{b_2|\Delta _a|-b_1|\Delta _s|}V_f(0) \end{aligned}$$\end{document} From Proposition 2, we have:33\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\begin{aligned} |\Delta _s| \ge \left( 1-\frac{1}{\theta }\right) (t-t_0)-\eta \end{aligned}$$\end{document} Therefore,34\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\begin{aligned} V_f(t) \le \mu ^{-1}e^{(b_1+b_2)\eta } e^{-\nu (t-t_0)} V_f(0)~, ~~t \ge t_0 \end{aligned}$$\end{document}where \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\nu =b_1(1-\frac{1}{\theta })-\frac{b_2}{\theta }$$\end{document}. Since \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\mu \rightarrow +\infty$$\end{document} as \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$t \rightarrow T_p$$\end{document}, we have \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\text {lim}_{t \rightarrow T_p} V_f = 0$$\end{document}. In the next proposition a self-triggered strategy is presented in the presence of DoS attack which results in good formation of the agents. Proposition 3 (Self-Triggered strategy in the presence of DoS attacks) In the presence of DoS attack, a good formation is obtained if, given \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\delta ,\epsilon _1 > 0$$\end{document}, the communication between the agents is triggered at the following instants:35\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\begin{aligned} t_{k+1}=t_k+\frac{\delta }{k\rho ({\mathcal {L}})\epsilon _1} \end{aligned}$$\end{document}where \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$t_k, k \in {\mathbb {N}}$$\end{document} is the \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$k^{th}$$\end{document} triggering instant and \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\delta ,\epsilon _1$$\end{document} are positive constants. Proof The closed-loop system with self-triggered control becomes :36\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\begin{aligned}&\dot{{\tilde{X}}}={\tilde{V}} \end{aligned}$$\end{document}37\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\begin{aligned}&\dot{{\tilde{V}}}=\left( k+c\frac{{\dot{\mu }}}{\mu }\right) (-{\mathcal {L}}{\tilde{X}}_k-k_1{\mathcal {L}}{\tilde{V}}_k) \end{aligned}$$\end{document}where38\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\begin{aligned}&{\tilde{X}}_k={\tilde{X}}(t_k)~~ \text {for} ~~t \in [t_k,t_{k+1}) \end{aligned}$$\end{document} Similarly,39\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\begin{aligned}&{\tilde{V}}_k={\tilde{V}}(t_k)~~ \text {for} ~~t \in [t_k,t_{k+1}) \end{aligned}$$\end{document} Let us define the triggering error as40\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\begin{aligned}&{\tilde{e}}_X={\tilde{X}}-{\tilde{X}}_k \end{aligned}$$\end{document}41\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\begin{aligned}&{\tilde{e}}_V={\tilde{V}}-{\tilde{V}}_k \end{aligned}$$\end{document} Then the closed loop (37) becomes :42\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\begin{aligned}&\dot{{\tilde{X}}}={\tilde{V}} \end{aligned}$$\end{document}43\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\begin{aligned}&\dot{{\tilde{V}}}=\left( k+c\frac{{\dot{\mu }}}{\mu }\right) (-{\mathcal {L}}{\tilde{X}}_k-k_1{\mathcal {L}}{\tilde{V}}_k)+\left( k+c\frac{{\dot{\mu }}}{\mu }\right) {\mathcal {L}}({\tilde{e}}_X+k_1{\tilde{e}}_V) \end{aligned}$$\end{document}which can be written as :44\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\begin{aligned} & \begin{bmatrix} \dot{{\tilde{X}}} \\ \dot{{\tilde{V}}} \end{bmatrix} =- \begin{bmatrix} 0_n & -I_n\\ \left( k+c\frac{{\dot{\mu }}}{\mu }\right) {\mathcal {L}} & k_1 \left( k+c\frac{{\dot{\mu }}}{\mu }\right) {\mathcal {L}} \end{bmatrix} \begin{bmatrix} {\tilde{X}} \\ {\tilde{V}} \end{bmatrix}+ \end{aligned}$$\end{document}45\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\begin{aligned} & \begin{bmatrix} 0 \\ \left( k+c\frac{{\dot{\mu }}}{\mu }\right) {\mathcal {L}}({\tilde{e}}_X+k_1{\tilde{e}}_V) \end{bmatrix} \end{aligned}$$\end{document} From Proposition 1 and Lemma 1, the above equation is prescribed-time stable if the second term is bounded. Therefore, let46\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\begin{aligned} ||\left( k+c\frac{{\dot{\mu }}}{\mu }\right) {\mathcal {L}}({\tilde{e}}_X+k_1{\tilde{e}}_V)|| < \delta \end{aligned}$$\end{document} Given \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$||({\tilde{e}}_X+k_1{\tilde{e}}_V)|| > \epsilon _1$$\end{document}, one can trigger at the following interval :47\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\begin{aligned} \Delta t \le \frac{\delta }{\left( k+c\frac{{\dot{\mu }}}{\mu }\right) ||{\mathcal {L}}||||({\tilde{e}}_X+k_1{\tilde{e}}_V)||} \le \frac{\delta }{\left( k+c\frac{{\dot{\mu }}}{\mu }\right) ||{\mathcal {L}}||\epsilon _1} \end{aligned}$$\end{document}from which (35) follows. Given \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\delta ,\epsilon _1 > 0$$\end{document}, one can trigger at the instants given by (35) to obtain good formation. This can be further simplified by observing that \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$k+c\frac{{\dot{\mu }}}{\mu } \ge k$$\end{document} and \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$||{\mathcal {L}}|| \ge \rho ({{\mathcal {L}}})$$\end{document}. Therefore, the triggering intervals are given by:48\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\begin{aligned} \Delta t = \frac{\delta }{k\rho ({{\mathcal {L}}})\epsilon _1} \end{aligned}$$\end{document} Remark 3 The spectral radius of the original graph without DoS attacks will be larger than the spectral radius of the graph with DoS attacks. Therefore, in (35), only spectral radius of the original topology is enough which is not difficult to obtain. Remark 4 The above self-triggered strategy obtains the next triggering instant from the last triggering instant and hence avoids continuous monitoring of the states. This is the shortcoming of the existing event-triggered strategies where continuous monitoring of states are required. The proposed self-triggered strategy saves the communication bandwidth in both sensor-controller channel as well as controller-sensor channel. □ Avoidance of Zeno behavior Since Δ t is strictly positive, and lower bounded by \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\frac{\delta }{k\rho ({{\mathcal {L}}})\epsilon _1}$$\end{document}, the phenomenon of infinite triggering i.e. Zeno behavior is completely avoided. A block diagram of the control framework is presented in Fig.1. Fig. 1Control framework. Control framework. Simulation result Performance of the proposed controller A simulation was performed with four agents in \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\mathbb {R}}^3$$\end{document} with undirected topologies shown in Figs. 2, 3 and 4 under DoS attacks. The initial topology is shown in Fig. 2 under no attacks. The DoS attacks are activated at t=10~s and 20 s for a duration of 2.5 s with the topology shown in Fig. 3. The DoS attacks are further activated at t=30~s and 40 s for a duration of 5 s with the topology shown in Fig. 4. Another DoS attacks are activated at t=50~s and 60 s for a duration of 5 s with the topology shown in Fig. 3. Last DoS attacks are activated at t=70~s and 80 s for a duration of 3 s with the topology shown in Fig. 4. The four agents were initialized at the following position\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\begin{aligned}&{\textbf {Agent 1}} ~: ~[0.8,0.7,0]^T;~~{\textbf {Agent 2}} ~: ~[2.2,1.1,0]^T \nonumber \\&{\textbf {Agent 3}} ~: ~[1.8,2.1,0]^T;~~{\textbf {Agent 4}} ~: ~[1.1,2.3,0]^T \nonumber \end{aligned}$$\end{document}Fig. 2Topology 1.Fig. 3Topology 2.Fig. 4Topology 3. Topology 1. Topology 2. Topology 3. All the agents were initialized with zero initial velocity. The various gains for simulation were taken to be\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\begin{aligned} k=2, k_1=0.5,c=0.5,T_p=80~s \end{aligned}$$\end{document} The desired distances used in the simulation for the agents are:\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\begin{aligned}&d_1=[1,1\cos (2\pi t/80),1\sin (2\pi t/80)]^T\\&d_2=[2,1\cos (2\pi t/80),1\sin (2\pi t/80)]^T\\&d_3=[2,2\cos (2\pi t /80),2\sin (2\pi t/80)]^T\\&d_4=[1;2\cos (2\pi t/80),2\sin (2\pi t/80)]^T \end{aligned}$$\end{document} The formation achieved for a simulation time of 80 s is shown in Fig. 5 where one can observe a good formation is achieved even in the presence of DoS attacks. To measure the goodness of formation, ratio of distances between each agents were found and plotted in Figs. 6 and 7 from which one can observe that the agents are maintaining a good square formation even in the presence of DoS attacks. As can be seen from Fig. 6, the ratio of distances between agents are almost 1 while the ratio of distances between diagonal agents are almost 1.4 as observed from Fig. 7. This indicates a good square formation as desired. The control input for agent 1 is plotted in Fig. 8 which is smooth and bounded. The formation achieved in the presence of self-triggered control proposed in this paper is shown in Fig. 9. Again one observes a good measure of formation as indicated in Figs. 10 and 11. The self-triggered control was tested under the same scenario as the time triggered control that is same set of topologies and same set of duration of DoS attacks. One can observe a slight deviation in the formation around t=30~s under DoS attacks but that is immediately mitigated when the attacks end. The control input for agent 1 is plotted in Fig. 12. Similar plots are obtained for other agents and are omitted for sake of repetition. The various gains for simulation were taken to be\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\begin{aligned} k=2, k_1=0.5,c=0.5,T_p=80~s, \delta =0.5, \epsilon _1=0.5 \end{aligned}$$\end{document}Fig. 5Formation achieved by the agents in time triggered scenario.Fig. 6Ratio of relative distance between agents in time triggered scenario.Fig. 7Ratio of relative distance between diagonally placed agents in time triggered scenario.Fig. 8Control input for agent 1 in time triggered scenario.Fig. 9Formation achieved by the agents in self-triggered scenario.Fig. 10Ratio of relative distance between agents in self-triggered scenario.Fig. 11Ratio of relative distance between diagonally placed agents in self-triggered scenario.Fig. 12Control input for agent 1 in self-triggered scenario. Formation achieved by the agents in time triggered scenario. Ratio of relative distance between agents in time triggered scenario. Ratio of relative distance between diagonally placed agents in time triggered scenario. Control input for agent 1 in time triggered scenario. Formation achieved by the agents in self-triggered scenario. Ratio of relative distance between agents in self-triggered scenario. Ratio of relative distance between diagonally placed agents in self-triggered scenario. Control input for agent 1 in self-triggered scenario. There were a total of 97 triggers during this time interval of \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$T_p$$\end{document}. The self-triggered control proposed in this paper results in good formation even in the presence of DoS attacks. Comparison result A comparison was performed with38 where event-triggered fixed time consensus was developed for a second order systems with nonlinearity and disturbances. The comparison was done with the same set of parameters and leader trajectory as demonstrated in38 but with the topology of previous section. For fair comparison with the proposed approach in this paper, all the nonlinearities and disturbances are made zero. For first 10 s, the simulation was done with the topology of Fig. 2. For the next 20 seconds, topology of Fig. 3 was utilized. For the last 10 s, the simulation was done with the topology of Fig. 4. The results for first 30 s are plotted in Figs. 13 and 14, where the Fig. 13 is for38 while Fig. 14 is for the proposed method. The RMS error in consensus for both the cases is shown in Table 1 from which it is observed that the error is less in the proposed method and hence better consensus is obtained. The average number of triggering in38 is 1392, while in the proposed approach it is 1196. After 30 s of simulation, topology of Fig. 4 becomes active and the controller in38 blows up as shown in Fig. 15 but the proposed method maintains stability. This shows that the proposed method not only guarantees better consensus with less number of triggering but also works in the presence of DoS attacks. The controller in38 does not perform well when there are more than one connected components in the topology but the proposed method works in that scenario also.Table 1Comparison between RMS errors.AgentsRMS error in consensus in38 [m]RMS error in consensus in the proposed approach [m]11.44520.516721.78010.610630.86300.663641.85590.4143 Table 1Comparison between RMS errors.AgentsRMS error in consensus in38 [m]RMS error in consensus in the proposed approach [m]11.44520.516721.78010.610630.86300.663641.85590.4143 Comparison between RMS errors. Fig. 13Consensus performance in38. Fig. 14Consensus performance in the proposed approach. Fig. 15Consensus performance in38 after 30 s. Consensus performance in38. Consensus performance in the proposed approach. Consensus performance in38 after 30 s. Conclusion This article presented a prescribed-time time-varying formation control framework for second-order multi-agent systems operating under Denial-of-Service (DoS) attacks using both time-triggered and self-triggered communication strategies. Initially, a prescribed-time formation control protocol was developed to guarantee convergence of second-order multi-agent systems to the desired time-varying formation despite the presence of DoS-induced communication interruptions. Subsequently, the proposed framework was extended to a self-triggered control architecture, significantly reducing communication and computation overhead while still ensuring prescribed-time convergence under adversarial network conditions. Stability and convergence properties of the proposed methods were analytically established, and extensive simulation studies, along with comparative performance evaluations, validated the effectiveness, robustness, and superiority of the proposed approaches over conventional methods. One of the limitation of the proposed work is that it has been developed for a simple second order system only. Therefore, future work will focus on extending the proposed prescribed-time resilient formation control strategy to higher-order multi-agent systems and more complex networked dynamical environments.