Adaptive Security in Smart Grid

Raja Ben Abdessalem and Mohamed Hamdi

MEDIATRON Laboratory, Higher School of Communication of Tunis, University of Carthage, Tunis, Tunisia

1 STAGE OF THE RESEARCH

The power infrastructure is being replaced by

digital systems based on renewable resources, to

meet the growing electricity demand and the

gradual depletion of fossil fuels as well as to satisfy

the customer’s needs in the near future. Smart grid

lets users manage, control and predict their energy

use. Therefore, the smart grid provides energy

efficiency and reduces peak power demand while

maintaining resilience, reliability, flexibility and

security of the grid (Santacana et al., 2010).

Consequently, smart grid should respond rapidly to

load changing conditions, reduce the cost of

interruptions and power quality disturbances and

reduce the probability and impact of attacks.

Smart grid introduces two-way communications

where electricity and information can be exchanged

between utility and its customers in real-time

(NIST,

2010b). However, the expansion of data

communication over the power grid, the large

number of involved parties and the heavily

interconnected communications increase

vulnerabilities and raise new security challenges.

Potential network intrusion caused by intentional

attackers may lead to a variety of damages (Metke

and Ekl, 2010).

Moreover, an intrinsic feature of smart grids is

that it is at the heart of critical services. In addition

to the direct impact that might be generated by the

occurrence of an attack, indirect losses can also be

taken into account. Therefore, specific security

models have to be developed in order to cope with

such issues.

This current research aims to develop an

adaptive risk-based security framework for smart

grid that addresses these security problems taking

into account smart grid requirements. The

framework will manage data in real-time, estimate

risks using game theory approach and apply the

appropriate adaptive security decision-making. We

mainly focus on analyzing security, privacy risk and

requirement capture for adaptive security in smart

grid. We also study the existing adaptive security

frameworks and to describe their limitations and

benefits. We fellow the Design Science

methodology (Hevner et al., 2004) to develop a

risk-based adaptive framework for the protection of

smart grid as well as to develop underlying

analytical models.

2 OUTLINE OF OBJECTIVES

The purpose of this thesis is to develop risk-based

adaptive security framework for smart grid that will

monitor in real-time smart grid environment to

detect possible threats, analyze and predict risk

damages using game-theory approach. Our

framework will adapt security decisions for

identified changes using decision-making.

The key objectives of this thesis are:

• Developing the adaptive security monitoring

model that uses a continuous cycle of a monitoring

framework for smart grid environment, based on the

Genetic Message-Oriented Secure Middleware

(GEMOM).

• Building models for estimating and predicting

risks and benefits using game theory. The thesis

will also further improve the accuracy of estimation

and prediction mechanisms by applying optimized

algorithm for resource allocation.

• Building the adaptive decision-making model

that adapt to the dynamic changing conditions of

smart grid by selecting the best adaptive security

model for a given situation and applying the

identified changes.

3 RESEARCH PROBLEM

Smart grid infrastructure might be vulnerable to

different forms of threats such as malware, intrusion

and Denial of Service (DoS) attacks; The smart grid

includes several devices that have two-way

Abdessalem R. and Hamdi M..

Adaptive Security in Smart Grid.

 2014 SCITEPRESS (Science and Technology Publications, Lda.)

communication with the electric system. These

numerous end points, which are located in insecure

large scale environments, bring new vulnerabilities.

A single vulnerability in a device can be used and

exploited to compromise all devices connected to

the network. In a smart grid, energy consumption,

pricing data and customer information can be

vulnerable to attacks. Therefore these critical

informations need to be protected. Thus, some

security goals need to be achieved and many

security parameters should be ensured such as

availability, confidentiality and integrity. Many

requirements and constraining issues should be

considered while designing security techniques and

algorithms for the smart grid.

• Big-data Requirement. Advanced technol-

ogies and application are integrated in the smart

grid. Accordingly, a huge amount of privacy-

sensitive data will be generated for further

control, analysis and real-time pricing methods.

These data can be extremely vulnerable to

attacks and can be extracted using data mining

techniques by different agents such as criminals

and insurance companies that determining

health care premiums based on customers’ life

styles. Therefore, it is very critical to define the

communication infrastructure requirements to

provide a secure and reliable service (Gungor et

al., 2011).

• Resource Limitations. The smart grid

operating systems have unique performance

and reliability requirements. The limited

availability of computation and communication

capabilities makes the security of smart grid

more complex. Some smart grid devices have

limited bandwidth. However, a significant

communication channel is required to ensure

the integrity using authentication algorithms.

Thus, authentication algorithms should be

designed to operate on lower bandwidth

channels. Moreover, smart grid devices have

limited computational power, memory and

storage. This can lead to using weaker security

controls. Consequently, the security algorithms

and mechanisms should be adapted to these

constrained devices.

• Real-time Requirement. The smart grid uses a

wide area measurement and control to measure

the electrical network parameters, to supervise

data acquisition and to take decision in real-

time. In addition, customers require a real-time

transmission of critical information, such as the

power consumption, to adjust their

consumption and to control peak demand.

Customers’ personal information (such as their

activities, the type of devices used and the

energy consumption) can be exposed to various

kinds of attacks which can damage customer

privacy. Therefore, security algorithms with

minimum computational cost should be

deployed to ensure the real-time requirement.

Many threats manifests in real-time scenarios.

Hence, adaptive security is used to increase

awareness and to provide real-time detection and

prevention. Adaptive risk management provides

automated risk analysis and dynamically adapt to

changing conditions. However, most current risk

management frameworks that address the security

and the privacy problems do not consider these

smart grid requirements. Therefore, this current

research focuses on the development of adaptive

risk management that takes into account the smart

grid requirements. This framework will learn, and

adapt to changing environment dynamically and

will anticipate unknown threats.

4 STATE OF THE ART

In this section, we motivate the need for the risk-

based adaptive security, we give brief state-of-the-

art in existing privacy and security risk models and

we underline their limitations and advantages.

4.1 Adaptive Security

Security and privacy threats manifest in real-time

scenarios. Therefore, different vulnerabilities may

be introduced to the system. Possible changes might

expose the system to new security threats (Salehie

et al., 2012). To address these threats, adaptive

software models promise to increase awareness,

automating monitoring, detection and prevention.

Therefore, security controls can be adapted

according to the security and privacy requirements’

failures at runtime in order to protect critical assets.

Adaptive security will check integrity of data if a

theft of energy is detected (for example if the

energy usage does not much with the reported meter

data). In addition, to ensure security controls at

runtime, access to the grid’s data need adjustment.

Adaptive security and privacy aims also to monitor

information exchanged to detect possible threats

especially in case of mobility where privacy

concerns for example user behavior can be revealed.

Thus, privacy policies, in such case, need to be

adjusted to mitigate possible threats.

ICETE2014-DoctoralConsortium

4.2 Existing Adaptive Risk

Management Frameworks

Traditional security measures such as firewall,

Intrusion Detection Systems (IDS) or Intrusion

Prevention Systems (IPS) usually lack the

correlated domain and situational awareness needed

to analyze events and inputs. Consequently, they

cannot adapt their security postures to evolving

situations and transitions.

Risk management provides automated risk

analysis by discovering critical assets; realizing

their weaknesses and the suitable risk mitigation

approaches.

To achieve the vision of a secure smart grid

infrastructure, a combined framework for risk

management has been proposed in (Riadh W. Y.

Habash and Burr, 2013). The risk management

framework is used to identify threats and

vulnerabilities, to analyze risks and to recommend

new controls in order to reduce the risks. This

process is periodically repeated to account for

changes in the threat situation; and set up policies

that address human behavior, which is the basis for

all security risks.

Authors in (Ray, 2011) proposed an adaptive

risk management framework that exploits near real-

time security state and events of power and

information systems to adapt security postures of

the grid to the situational awareness. Security risks

is evaluated using a systematic methodology that

consist of identifying all assets (such as power and

information systems), assessing vulnerabilities of

each asset, analyzing all potential threats that can

exploit the identified vulnerabilities, identifying the

appropriate security control mechanisms and finally

determining the optimum security postures for each

asset.

Authors in (Sridhar et al., 2012) proposed a

vulnerability assessment framework aiming to

quantify risk using Stochastic Petri Nets. The Petri

Nets are used to model the cyber network and to

obtain the probability of a successful attack. This

framework identifies and handles multiple events,

then analyzes the risk without providing response

strategies. However, Petri nets are facing

difficulties in application due to their computational

efficiency and scalability (Tang et al., 2004).

4.3 Gap Analysis

Table 1 describes the limitations and the benefits of

exiting adaptive risk management frameworks, and

the benefits of our proposed framework as

comparison as shown in the table below. In this

Table, limitations are indicated by ”(-)”, benefits by

”(+)”, ”Req.” (Requirements) indicates the

important smart grid requirements and ”Ref.”

(References) indicates some existing risk

management frameworks.

The risk management of the power grid control

information should exploit historical vulnerability

and threat data in order to enable domain specific

statistical analysis and characterization of attack

probabilities and risks.

Following this, we propose a risk-based adaptive

security framework using game theory that can

model the dynamic behavior of stakeholders. This

framework detects in real-time unknown security

and privacy threats, respond to them, and adapt to

the environment and changing degree of security.

The benefits of our proposed framework are

presented in Table 1.

4.4 Game Theory

The game theory is widely used to study a variety of

security problems (Bier and Azaiez, 2010)

(Manshaei et al., 2013). Game theory is a rich

mathematical tool that analyzes and models the

interaction between an attacker and a defender (the

players). The game theory methods can be very

useful in improving current risk analysis by

clarifying the risks of adversarial situations. The

game theory models, that can distinguish clearly

between strategic choices and random variable, can

make risk assessments more sensible and effective

for allocating defensive resources than current risk

scoring models (Cox, 2009).

5 METHODOLOGY

5.1 Introducing New Concepts

In this work, we develop a new methodology for

dynamic risk assessment based on the approach

proposed in (Poslad et al., 2013). The overall

adaptive security approach is illustrated in Figure 1.

The synthetic methods are concerned with

building models of phenomena related to the

development, operation and use of aggregated

services, mainly models of software artifacts. We

subject to the overall framework of Design Science

(Hevner et al., 2004), where information system

artifacts are built, and evaluated. Evaluation will be

performed using Action Research (Baskerville,

1999), case study research, and formal modelling of

AdaptiveSecurityinSmartGrid

Table 1: Gap analysis of existing risk management frameworks.

Req.

Ref.

Real-time Resources limitations Big data

(Riadh W. Y.

Habash and

Burr,

2013)

(-) The framework does not incorporate

attack prevention that accurately

predicts future events and adapts

accordingly in real-time.

(+) The framework is adapted to human

behavior by repeating periodically the

risk management process.

(-) The unified framework

is not optimized since it

contains several steps

(+) The risk strategy

depends on the available

resources.

(-) The framework cannot

manage a large volume of

data.

(Ray, 2011)

(-) The framework does not incorporate

attack prevention.

(+) The security control posture

changes in near real-time in response to

changes in the power and information

system state changes.

(+) This framework

depends on the available

resources.

(+) This framework can

include data centers and

Intelligent Electronic

Devices (IEDs) as well as

emergent infrastructure and

rocesses.

(Sridhar et

al., 2012)

(-) When the number of functional

alternative Services increases, many

transactions missed their deadlines

because they could not get enough

resources in time.

(+) Petri nets can model the

coordination algorithms of the real-

time grid transaction and verify their

correctness.

(-) Petri nets consume a lot

of resources when many

services are used.

(+) Petri nets can analyze a

large volume of statistics.

Our

proposed

framework

(+) The adaptive security monitoring

model collects data in real-time.

(+) SCAP enables sharing metrics and

measurements of smart grid in realtime.

(+) Our framework incorporates attack

prevention to estimate risk in real-time.

(+) Our framework enable security

control to coordinate with changes in

security state and events.

(+) Automated process

enables monitoring a larger

number of security metrics

with few resources and low

costs.

(+) The predictive model

will use an optimized

algorithm for resource

allocation to choose the

best strategy.

(+) Our framework uses

SCAP that can monitor a

large volume of data.

Figure 1: The overall adaptive security approach.

relevant information flows. The methods are also

concerned with validating the models using formal

criteria and with drawing conclusions from those

models. Examples on the use of such methods:

Make models based on the formal semantics of

programming languages, and analyse the properties

of those models, using for example flow analysis to

determine flow of information. This leads to sound

conclusion using well-established theoretical

methods. The security models will be realized as a

prototype specification, accompanied with brief

guidelines and profiles for use. Additional

validation of the design based on formal modelling,

will be carried out. The other methods, from

software and systems engineering, are used to

analyse existing smart grids systems and plan their

use.

5.2 Developing, Validating, and

Testing New Models

Our proposed framework aims to adapt dynamically

the security postures to the changing conditions by

monitoring smart grid environment in real-time,

ICETE2014-DoctoralConsortium

analyzing the collected information (to anticipate

unknown threats) and making adaptive decision.

Figure 2 illustrates our adaptive risk management

framework for smart grid. The framework is

composed of three models: the adaptive monitoring

model, the predictive model and the adaptive

decision model.

Figure 2: Adaptive risk management framework for

Smart Grid.

5.2.1 Adaptive Monitoring

The adaptive security monitoring model collects

data in real-time from different devices such as

meters and sensors. However, the interoperability

among them is a critical requirement to integrate

security and risk mitigation. In this context, the

National Institute of Standards and Technology

(NIST) developed a framework that includes

protocols and model standards for information

management to achieve interoperability of smart

grid devices and systems (NIST, 2010a). Authors in

(DHS, 2011) proposed three interdependent

building blocks to enhance cybersecurity:

interoperability, automation and authentication.

Interoperability defines cyber communities by

policies which permit to cyber participants to

collaborate dynamically in automated community

defense. Automated process enables monitoring a

greater number of security metrics with fewer

resources, lower cost and greater reliability and

efficiency than using manual process. Automated

monitoring provides measurable, relevant and

timely information. Many technical security

controls can be used for monitoring with automated

tools. Automated tools, such as Security Content

Automation Protocol (SCAP) (Barrett et al., 2009),

help monitoring large volumes of data. SCAP is

used to communicate data in a standardized format

for performing security automation capabilities and

promoting interoperability of security products.

SCAP integrates the Common Vulnerability

Scoring System (CVSS) and Common

Vulnerabilities and Exposures (CVE) in order to

provide quantitative and repeatable measurement

and scoring of software flaw vulnerabilities across

system. The standardization makes security metrics

and measurements easier to share. In order to ensure

an up-to-date threat and vulnerability knowledge in

the smart grid system, we propose to use shared

metrics and measurement repositories such as

SCAP.

Our adaptive security monitoring will use a

continuous cycle of a monitoring framework for

smart grid environment, based on the Genetic

MessageOriented Secure Middleware (GEMOM).

GEMOM, which was proposed in the European

Commission’s Framework Programme 7 project

GEMOM (2008-

2010), develops a messaging platform that is

resilient, evolutionary, self-organizing, self-healing,

scalable and secure (Abie et al., 2009).

5.2.2 Predictive Model

The predictive model uses the monitored

information to estimate and predict risks based on

game theory approach. The game theory is suitable

for modeling and analyzing the complexity of

interaction between stakeholders and adversaries as

well as their behaviors and the choice of their

strategies. However, the players’ strategies can

include allocations of resources to prepare for

possible attacks. The predictive model will use an

optimized algorithm for resource allocation to

choose the defender’s best allocation of defensive

resources, taking into account the attacker’s best

response.

5.2.3 Adaptive Decision

Adaptive decision model will enable security

control to coordinate with changes in security state

and events. The resulting predictive model helps

smart grid to take decisions on their security

strategies while considering their resource

constraints.

Adaptive security decision-making will adapt

security control according to the level of security

and the role of stakeholders.

6 EXPECTED OUTCOME

The proposed risk management framework ensures

an adaptive security for smart grid by monitoring

data in real-time, predicting risks and adapting

AdaptiveSecurityinSmartGrid

security decisions to changing circumstances. Our

framework is based on a continuous cycle of

adaptive monitoring, predictive analytics and

automated adaptive decisionmaking. In our future

work, we aim to develop the adaptive monitoring

model using automated process and based on the

GEMOM Middleware. Then, we plan to develop the

predictive model that will estimate risks and the

decision-making model.

REFERENCES

Abie, H., Savola, R., and Dattani, I. (2009). Robust,

secure, self-adaptive and resilient messaging

middleware for business critical systems. In Future

Computing, Service Computation, Cognitive,

Adaptive, Content, Patterns, 2009.

COMPUTATIONWORLD ’09. Computation World:,

pages 153–160.

Barrett, M., Johnson, C., Mell, P., Quinn, S., Scarfone, K.

Stephen Quinn, K. S. M. B., and Johnson, C. (2009).

Guide to adopting and using the Security Content

Automation Protocol (SCAP). NIST Special Publ.

800-117 (Draft), U.S. National Institute of Standards

and Technology.

Baskerville, R. L. (1999). Investigating information

systems with action research. Commun. AIS, 2(3es).

Bier, V. M. and Azaiez, M. N. (2010). Game

Theoretic Risk Analysis of Security Threats. Springer.

Cox, Jr., L. A. T. (2009). Game theory and risk analysis.

Risk Analysis, 29(8):1062–1068.

DHS (2011). Enabling Distributed Security in

Cyberspace: Building a Healthy and Resilient Cyber

Ecosystem with Automated Collective Action. DHS

National Protection and Programs Directorate.

Gungor, V., Sahin, D., Kocak, T., Ergut, S., Buccella, C.,

Cecati, C., and Hancke, G. (2011). Smart grid

technologies: Communication technologies and

standards. Industrial Informatics, IEEE Transactions

on, 7(4):529–539.

Hevner, A. R., March, S. T., Park, J., and Ram, S. (2004).

Design science in information systems research. MIS

Q., 28(1):75–105.

Manshaei, M. H., Zhu, Q., Alpcan, T., Bacs¸ar, T., and

Hubaux, J.-P. (2013). Game theory meets network

security and privacy. ACM Comput. Surv.,

45(3):25:1–25:39.

Metke, A. and Ekl, R. (2010). Smart grid security

technology. In Innovative Smart Grid Technologies

(ISGT), 2010, pages 1–7.

NIST (2010a). Nist framework and roadmap for smart

grid interoperability standards, release 1.0.

NIST (2010b). The smart grid interoperability panel-

cyber security working group: Smart grid cyber

security strategy and requirements. NIST IR-7628.

Poslad, S., Hamdi, M., and Abie, H. (2013). Adaptive

security and privacy management for the internet of

things (aspi 2013). In Proceedings of the 2013 ACM

Conference on Pervasive and Ubiquitous Computing

Adjunct Publication, UbiComp ’13 Adjunct, pages

373–378, New York, NY, USA. ACM.

Ray, P. D. (2011). Interoperating grid cyber security

systems: Adaptive risk management across unified ot

and it domains. Grid-InterOp.

Riadh W. Y. Habash, V. G. and Burr, K. (2013). Risk

management framework for the power grid cyber-

physical security. British Journal of Applied Science

and Technology.

Salehie, M., Pasquale, L., Omoronyia, I., and Nuseibeh,

B. (2012). Adaptive security and privacy in smart

grids: A software engineering vision. In Software

Engineering for the Smart Grid (SE4SG), 2012

International Workshop on, pages 46–49.

Santacana, E., Rackliffe, G., Tang, L., and Feng, X.

(2010). Getting smart. Power and Energy Magazine,

IEEE, 8(2):41–48.

Sridhar, S., Govindarasu, M., and Liu, C.-C. (2012). Risk

analysis of coordinated cyber attacks on power grid.

In Control and Optimization Methods for Electric

Smart Grids. Springer New York.

Tang, F., Li, M., and Huang, J. Z. (2004). Real-time

transaction processing for autonomic grid

applications. Eng. Appl. Artif. Intell., 17(7):799–807.

ICETE2014-DoctoralConsortium