CPS Testbeds: Energizing a Cybersecure Smart Grid
By Manimaran Govindarasu, Adam Hahn, and Chen-Ching Liu
Current efforts to research cyber security technologies, risk assessment and system hardening techniques are constrained by the availability of realistic test environments. Creating a resilient smart grid requires the availability of cyber-physical system security testbeds for performance and security evaluations in order to achieve sufficient system resiliency.
Cyber-physical security (CPS) testbeds can accelerate the pace of security innovation and technology transfer in smart grid R&D by providing a bridge between theory and practice. This need has been identified as a national priority by the NIST Smart America Challenge, which focuses on the development of an integrated testbed framework across all CPS domains through the collaboration among researchers and industries.
In order to accurately model the smart grid, testbeds must provide “power-hardware-in-the-loop” and “cyber-in-the-loop” capabilities through a combination of real, simulated and emulated components or systems. Such a platform can realistically mimic the complex cyber-physical interactions in the power system that cannot be accurately modeled using traditional simulation tools or stand-alone cyber or power simulators.
A well-designed testbed can benefit four primary users: technology vendors, system owners (such as utilities and ISOs), regulatory agencies (like NERC and FERC in the United States), and R&D institutions. The potential benefits cover a range of applications including: environments to explore new vulnerabilities and assessment techniques; impact analysis evaluating system performance under attacks; risk assessment modeling real world environments; validating/evaluating the effectiveness of defense algorithms (including risk mitigation strategies); attack-defense exercise against various forms of cyber attacks such as data integrity attacks, timing and replay attacks, denial of service, intrusion-based attacks, and coordinated cyber attacks (like High Impact Low Frequency Events); education and training in cyber security technologies and operational system studies.
For R&D institutions like universities, national laboratories and companies, testbeds are crucial to understanding current security issues and developing next-generation technologies. System vulnerabilities can be explored in a trusted environment to ensure this sensitive information can be responsibly disclosed to vendors and mitigation measures can be taken. Additionally, new risk assessment techniques and security mechanisms can be designed and tested under real-world constraints.
Vendors need environments so their products can be tested and validated. New systems must be designed with the assurance that they can operate in the face of increased cyber threats. Additionally, these systems must be able to interoperate with many other products using many configurations. Testbeds provide an environment where products can be validated in a real-world environment and against realistic attacks to ensure proper interoperability and performance.
Utilities also benefit greatly from testbeds in the design and verification of their systems. Utilities currently design testbeds to explore how new technologies can integrate with and improve their services. For example, Southern California Edison has implemented its own tesbed with real-time digital simulators and phasor measurement units to evaluate benefits with wide-area situational awareness.
Utilities are responsible for ensuring their systems are deployed and maintained securely. Compliance requirements, such as NERC CIP (Critical Infrastructure Protection) require periodic audits to verify cyber security mechanisms are correctly introduced.
Unfortunately, many of the current methods used to perform cyber security testing are based on intrusive scanning techniques that can cause systems to crash or misoperate. Testbeds provide a safe environment in which newer, unobtrusive methods can be explored. For example, NIST is currently exploring the use of Security Content Automation Protocol (SCAP) technologies to evaluate systems in industrial control system environments. Testbeds play a key role in this evaluation as they can be used to verify correct operation on numerous system architectures and configurations.
Testbeds also provide a key role in education and training. At the university level, students can learn about current technologies and gain practical insights into the operation of modern power system. Additionally, utilities and independent system operators can leverage them to educate operators about best practices to prevent, identify and respond to cyber incidents. For example, NERC currently performs periodic GridEx exercises to explore how the utilities and regional entities would respond to critical security incidents. The ability to accurately simulate attacks in a real-world environment is crucial to understanding how key decision makers will respond. Testbeds provide an ideal environment in which mock cyber incidents can be simulated and response actions can be evaluated.
In recent years, several research efforts (for example at Iowa State University, University College Dublin, University of Illinois and Washington State University) have developed proof-of-concept implementations of CPS security testbeds with “power-hardware-in-the loop” and “cyber-in-the-loop” features demonstrating vulnerability analysis, impact analysis and attack-defense exercise capabilities. While testbeds provide substantial benefit to utilities, vendors and R&D institutions, many challenges remain to improve their scalability, accuracy and usability.
The efficacy of testbed-based evaluations and validations hinges on the accuracy of the models and datasets. Currently there is a lack of realistic CPS datasets and models that integrate the physical power system with the communication and control elements of the grid.
Another primary challenge is scaling up testbeds to mimic regional power grid and eventually the North American power grid to conduct GridEx-type experiments with dynamic scenarios. Currently, the scalability is constrained due to lack of adequate hardware and software resources and collaboration efforts. This problem calls for a testbed federation (such as NIST’s Smart America Challenge) with an ambitious vision of creating a CPS security cloud infrastructure by means of which a diverse pool of CPS resources from research institutions and industries can be pooled together and shared efficiently.
Collaborative efforts must be undertaken with participation from all stakeholders to realize the vision of a CPS security cloud and to create and manage realistic data sets and libraries to promote R&D and education in this field. In addition, testbeds must be made more remotely accessible to all researchers, not just those fortunate enough to have the large initial capital to fund their development. Just as the DETER testbed has provided a remotely accessible research environment for the experimentation of cyber security technologies, testbeds must promote a community access model where researchers can create, share and collaborate to accelerate innovation in smart grid security.
Manimaran Govindarasu, a senior member of IEEE, is Mehl Professor in the Department of Electrical and Computer Engineering at Iowa State University. His research expertise is in cyber-physical systems security for the smart grid, cyber security and real-time networks. He is the chairman of the IEEE Power & Energy Society’s Task Force on Cyber Security of Power Systems. He received his Ph.D. from the Indian Institute of Technology, Chennai, in 1998.
Adam Hahn, an IEEE member, is an information security engineer at the MITRE Corporation. His research expertise is in cyber vulnerability assessment and cyber security of critical infrastructure systems. He received his Ph.D. from Iowa State University in 2013.
Chen-Ching Liu, an IEEE fellow, is Boeing Distinguished Professor in the School of Electrical Engineering and Computer Science at Washington State University, Pullman, Wash. and a professor at University College Dublin, Ireland. His research expertise is in computational algorithms and cyber risk modeling for the smart grid. He is a former chairman of the IEEE Power & Energy Society’s Technical Committee on Power System Analysis, Computing and Economics. He received his Ph.D. from the University of California, Berkeley, in 1983.