Modeling cooperative behavior for resilience in cyber-physical systems using SDN and NFV
- PDF / 1,991,675 Bytes
- 13 Pages / 595.276 x 790.866 pts Page_size
- 72 Downloads / 162 Views
Modeling cooperative behavior for resilience in cyber‑physical systems using SDN and NFV Jose Moura1,2 · David Hutchison3 Received: 9 May 2020 / Accepted: 12 August 2020 © Springer Nature Switzerland AG 2020
Abstract Cyber-Physical Systems (CPSs) are increasingly important in everyday applications including the latest mobile devices, power grids and intelligent buildings. CPS functionality has intrinsic characteristics including considerable heterogeneity, variable dynamics, and complexity of operation. These systems also typically have insufficient resources to satisfy their full demand for specialized services such as data edge storage, data fusion, and reasoning. These novel CPS characteristics require new management strategies to support the resilient global operation of CPSs. To reach this goal, we propose a Software Defined Networking based solution scaled out by Network Function Virtualization modules implemented as distributed management agents. Considering the obvious need for orchestrating the distributed agents towards the satisfaction of a common set of global CPS functional goals, we analyze distinct incentive strategies to enact a cooperative behavior among the agents. The repeated operation of each agent’s local algorithm allows that agent to learn how to adjust its behavior following both its own experience and observed behavior in neighboring agents. Therefore, global CPS management can evolve iteratively to ensure a state of predictable and resilient operation. Keywords Cyber-physical systems · Internet of things · Software-defined networking · Game theory · Network function virtualization · Threats and cyber-attacks · Algorithms · Resilience · Resilient systems · Cooperation · Orchestration · Robustness
1 Introduction A Cyber-Physical System (CPS) is essentially a physical facility with embedded sensors and actuators that can be remotely monitored and controlled by computerized systems [1], which we assume here are distributed virtualized agents implemented by Virtual Network Functions (VNFs), most of them located at the network edge. The monitoring and control of CPS are made by logical control loops over physical communication channels. These channels are established between the sensors/actuators and the VNFs. The channels transfer data representing the facility status and control messages to change the operation mode. CPSs
are increasingly found in diverse applications areas such as power grids [2, 3], smart buildings [4, 5], next-generation mobile communication systems [6, 7], healthcare systems [8, 9], and also in precision farming systems [10, 11]. Recent years have shown the increasing relevance of CPSs in every day of our lives. In this way, the reliable acquisition and processing of the data originated at the physical part of each CPS become very important. To support the efficient extraction of useful knowledge from the processed CPS data, we argue that, in the current work, several research areas and techniques need to be combined. These are Software Defined Networking (SDN) [12]
Data Loading...
