Cyber-Physical Systems (CPSs) integrate the interdisciplinary fields of computing, networking and control to perform tasks in the real world. CPSs have recently found applications in many battery-powered devices with stringent energy consumption requirements. To ensure secure operation, CPS necessitates sufficient security mechanisms to be incorporated against cyber attacks. However, maximizing energy efficiency and improving security are desirable but contrasting requirements. Towards reducing energy consumption, the optimal strategy for CPS is to initialize the security mechanism dynamically, at the onset of cyberattacks. In the absence of attacks, CPS can deactivate the security mechanism to minimize energy consumption. In the case of CPS, this approach is novel and contrary to the traditional approach of long-term, continual operation of the security mechanism. Towards this goal, we use a decision-centric approach based on Markov Decision Process (MDP) to estimate a threshold upon which the system initiates its security mechanism. We evaluate our proposed mechanism using MATLAB based TrueTime simulator. Evaluation shows that our proposed MDP-based approach achieves maximum energy-savings of 8.26 and \(11.05\%\) in defending against Denial-of-Service and Deception attacks, respectively. Further, our approach can be used to develop sustainable CPS designs that balance the trade-off between energy-efficiency and security.

网络物理系统（CPS）集成了计算，网络和控制的跨学科领域，可以在现实世界中执行任务。CPS最近在许多对电池有严格能耗要求的设备中得到了应用。为了确保安全运行，CPS要求有足够的安全机制来合并以抵御网络攻击。然而，最大化能量效率和改善安全性是期望的，但是有相反的要求。为了降低能耗，CPS的最佳策略是在网络攻击发生时动态初始化安全机制。在没有攻击的情况下，CPS可以停用安全机制以最小化能耗。就CPS而言，这种方法是新颖的，与传统的长期方法相反，安全机制的持续运作。为了实现这一目标，我们使用基于马尔可夫决策过程（MDP）的以决策为中心的方法来估计系统启动其安全机制的阈值。我们使用基于MATLAB的评估我们提出的机制TrueTime模拟器。评估显示，我们提出的基于MDP的方法在抵御拒绝服务和欺骗攻击方面分别实现了8.26和\（11.05 \％\）的最大节能。此外，我们的方法可用于开发可持续的CPS设计，以平衡能源效率和安全性之间的权衡。