Multilevel Programming-Based Coordinated Cyber Physical Attacks and Countermeasures in Smart Grid

Abstract

Since the Ukraine blackout in 2015, coordinated cyber-physical attacks (CCPAs) have been emerging and are used to mask line outages in the smart grid. In this paper, we investigate the features of CCPAs and constitute the mathematic formulation with respect to topologies and electric parameters of a power grid before and after attacks. With the objective of maximizing the number of overloaded lines, a bilevel programming model is developed to describe the interaction between the adversary and the control center. The most damaging CCPA can be determined by transforming the developed bilevel model to a single mixed-integer linear programming problem using the Karush-Kuhn-Tucker conditions. Based on the features of the bilevel model, the countermeasure is expressed as a trilevel model with one leader and multiple followers. The implicit enumeration-based searching strategy is proposed to solve the trilevel model to identify the protected meters. Both the implementation of CCPAs and the effectiveness of the developed countermeasure are verified on the modified IEEE 14-bus system. INDEX TERMS Bilevel programming, coordinated cyber physical attacks (CCPAs), trilevel programming, smart grid. NOMENCLATURE Sets and Indices: O/d Bus set / bus index. G/g Generation set / generation index. L/l Transmission line set / transmission line index.

Cyber-physical risk modeling with imperfect cyber-attackers

“…The same concept was exploited in [6] by an attacker seeking to maximize the loading of a transmission line. Zhang and Sankar developed in [7] an elaborate bilevel formulation for an attacker using load redistribution in order to hide a physical change in the grid topology, while a model of an attacker seeking to maximize the number of overloads induced through load redistribution has been presented in [8].…”

“…Expression (3) imposes a limit on the maximum number of load meters that can be manipulated by the attacker, while (4) enforces that the false load measurement data injection is balanced across the grid and ( 5) sets the maximum relative amount of false data that can be injected by the attacker at any node 1 . Equalities (7,8) model the power flow of the grid as perceived by the cyber-attacker only. Notice that the power balance constraint (7) includes the optimal values of the generation redispatch variables (p g ) as decided in the gridoperator's lower-level problem (20 -26).…”

“…We adopt the single-area version (24 bus) of the IEEE-RTS96 benchmark 4 . Following the practice of relevant studies (e.g., [5], [6], [8]) we simulate a stressed operational condition by reducing all branch transmission capacities to 65% of the original values. We further model a malicious cyber-attacker seeking to overload at least U = 2 transmission elements to at least ρ = 5% of the respective capacities.…”

Risk Assessment of Cyber Attacks on Power Grids Considering the Characteristics of Attack Behaviors

“…Analysis based on topological models or simulation platforms always focus on implementation scheme rather than the decision strategies of the attack and so the intention of the attacker and other subjective factors are generally omitted. Some studies concerning with attack intention employed game theory and multilevel programming to model attack-and-defense competition with optimal strategies, such as Static game for attack probability modeling [19][20][21], Colonel Blotto game and Stackelberg game for optimal attack resource allocation [22,29], Markov game for dynamic attack modeling [23][24][25], Tri-level programming model for vulnerability assessment and component protection [26][27][28]. However, the factors affecting attack decision making and the diversity of attack intention have not been fully coved in existing studies, primarily due to the difficulty of data acquisition of necessary information for characterizing the attack behaviors in detail.…”

Cascading Outage Analyses by Integrating Distribution Factor Method With AC Power Flow

“…Large-scale blackouts have happened worldwide during the past decades, resulting in economic loss, social inconvenience, and cyber physical-attacks [1]- [5]. For the sake of ensuring security and reliability of the power system operation, the principle of N − 1 security is widely employed in industry [6], where the whole system is required to withstand the failure of any single component.…”