Presentations Evaluating the Effectiveness of Security Mechanisms - - PDF document

Trustworthy Cyber Infrastructure for the Power Grid

Presentations

University of Illinois • Dartmouth College • Cornell University • Washington State University

1

Evaluating the Effectiveness of

David Nicol

Security Mechanisms at Scale

University of Illinois • Dartmouth College • Cornell University • Washington State University

Vision/Drivers for Security Evaluation

A long history of research in evaluation of systems has led to significant capabilities – Reliability (e.g. in avionics, safety-critical systems, etc.) – Availability (e g computer systems offering services) Availability (e.g. computer systems offering services) – Capacity planning – Performance

• How fast?
• How much?

– Paths to deleterious states Lik lih d f d l t i t t

University of Illinois • Dartmouth College • Cornell University • Washington State University

– Likelihood of deleterious states – Compliance Testing

Vision : similar success in security evaluation

Trustworthy Cyber Infrastructure for the Power Grid

Presentations

University of Illinois • Dartmouth College • Cornell University • Washington State University

2

Vision/Drivers for Security Evaluation

Evaluation methodologies offer promise for – Assessing effectiveness of security mechanisms

• In context of identified threats

Assessing scalability of security mechanisms – Assessing scalability of security mechanisms

• Where are the bottlenecks?

– Quantifying the security of a system

• Domain-specific metrics

– A basis for cost/benefit analysis – Assessing regulatory compliance

University of Illinois • Dartmouth College • Cornell University • Washington State University

– Assessing impact of security mechanisms (and other applications) on overall system behavior

• e.g., real-time requirements

Drivers : economic, regulatory, competitive Problems for Security Evaluation

Metrics – What do we measure? Needs driven by emerging power technologies Needs driven by emerging power technologies – e.g., PMUs, smart switches, adaptive flow redistribution Threat characterization – Model known threats, account for unknown threats

University of Illinois • Dartmouth College • Cornell University • Washington State University

Issues of Scale (breadth and depth) – Security “flaws” often hinge on small technical details Model Verification / Validation

Trustworthy Cyber Infrastructure for the Power Grid

Presentations

University of Illinois • Dartmouth College • Cornell University • Washington State University

3

Approach to Evaluation

Research Thrusts – Power grid drivers for security research

• Distributed agent-based grid control
• Adaptive Power Flow Redistribution
• Adaptive Power Flow Redistribution

– Security Evaluation of Vulnerabilities and Solutions

• Line dependency analysis
• Automated security tools

University of Illinois • Dartmouth College • Cornell University • Washington State University

– Test-bed Development

• Wireless simulation / emulation models
• Wire-line MAC layer modeling
• Integration of hardware/software, Powerworld

Intelligent Agents

Problem : Emerging application would be distributed, automate control. Identify communication requirements, e.g. latency; identify security requirements, e.g. authentication y q , g Approach : Develop agent models on simulated network, develop control algorithms and induced communication Results : Simulation models on reference power architectures

University of Illinois • Dartmouth College • Cornell University • Washington State University

Trustworthy Cyber Infrastructure for the Power Grid

Presentations

University of Illinois • Dartmouth College • Cornell University • Washington State University

4

Line contingencies

Problem : Assess risk to grid of sets of lines going out Approach : Flow on each line depends on the flows on all other lines What metrics reveal ordered dependencies? – What metrics reveal ordered dependencies? – Explore ways of reducing computational cost through screening to find highest sensitivities Through contingency screening generate a structure that tracks lines impacts on each other

University of Illinois • Dartmouth College • Cornell University • Washington State University

Results : regions of grid most sensitive to tampering are identified

Automated Assessment Problem : Manual vulnerability assessment is difficult. Approach : Automate assessment through analysis of interacting – Automate assessment through analysis of interacting models Results : Design time analysis of SIP requirements and compliance

University of Illinois • Dartmouth College • Cornell University • Washington State University

Trustworthy Cyber Infrastructure for the Power Grid

Presentations

University of Illinois • Dartmouth College • Cornell University • Washington State University

5

Automated Assessment

University of Illinois • Dartmouth College • Cornell University • Washington State University

TCIP Testbed

Problem : Assessment of large-scale power systems Approach : Develop testbed comprised of Real RTUs control station systems historians etc – Real RTUs, control station systems, historians, etc. – Simulator of power generation and distribution – Simulator/emulator of communications, devices, networks – Experimental designs to evaluate technologies – Models and methodologies to deal with problems of scale

University of Illinois • Dartmouth College • Cornell University • Washington State University

Results : Demonstration of integrated simulators and power equipment

Trustworthy Cyber Infrastructure for the Power Grid

Presentations

University of Illinois • Dartmouth College • Cornell University • Washington State University

6

TCIP Testbed

University of Illinois • Dartmouth College • Cornell University • Washington State University

Access Policy Tool

Problem : – Firewall misconfigurations in process control systems – Expression/checking best practices in PCS Verifying compliance of implementation with global policy – Verifying compliance of implementation with global policy Approach : – Analyze all possible flows through network for compliance – Computation optimized – Statistical techniques for very large networks

University of Illinois • Dartmouth College • Cornell University • Washington State University

– Graphical interface Results : – Prototype tool in beta-testing with power industry partner

Trustworthy Cyber Infrastructure for the Power Grid

Presentations

University of Illinois • Dartmouth College • Cornell University • Washington State University

7

Access Policy Tool

University of Illinois • Dartmouth College • Cornell University • Washington State University