Resilient Control for Unbounded False Data Injection Attacks

This direction is merged into another direction: stealthy false data injection. Its last version stays readable and citable here.

Version 1 of 112 papers24 conceptsfrozen 2026-10-04

In traditional control theory, security is often treated as a boundary problem: we assume an attacker can only inject a certain amount of "noise" or that their interference won't change too quickly. But what happens when an attacker isn't polite? If a malicious actor injects a signal that grows indefinitely—accelerating at a polynomial or even exponential rate—standard controllers eventually hit a breaking point and the system diverges.

A growing body of research is shifting the paradigm from attack detection to inherent self-resilience. Instead of trying to identify and isolate a compromised node, these new controllers use adaptive compensation to "outgrow" the attack, maintaining stability even when the interference is theoretically infinite.

The Failure of Bounded Assumptions

Most networked systems, such as AC microgrids or swarms of autonomous robots, rely on distributed consensus. Agents share data with their neighbors to synchronize frequency, voltage, or position. A False Data Injection (FDI) attack exploits this trust by injecting a signal $a(t)$ into the communication channel.

If $a(t)$ is a constant bias, a robust controller can usually reject it. However, as noted in Secondary Defense Strategies of AC Microgrids Against Generally Unbounded Attacks, real-world attackers can deploy "generally unbounded" signals—functions like $t^2$ or $t^3$ that grow over time. Standard defenses often fail here because they assume the attack's derivative (its rate of change) is bounded. When the attack accelerates, the controller can no longer keep up, and the system's error grows without bound.

Adaptive Compensation: Matching Growth with Growth

The core intuition behind these new resilient strategies is adaptive matching. Rather than trying to "block" an attack, the controller estimates the magnitude of the interference in real-time and generates a cancellation signal.

To handle unbounded attacks, the "gain" or strength of this cancellation signal cannot be fixed. In Defense Strategies for Autonomous Multi-agent Systems: Ensuring Safety and Resilience Under Exponentially Unbounded FDI Attacks, Rui Liu demonstrates that by using time-varying tuning parameters, an observer can match the scale of an exponentially growing attack.

Think of it as an automated tug-of-war. If the attacker pulls harder and harder, the adaptive controller senses the increasing tension (the neighborhood error) and automatically increases its own pulling force. As long as the controller's "muscles" (the adaptive laws) can grow as fast as the attacker's, the system remains stable.

Relaxing the Derivative Constraint

A significant technical hurdle in this field is the "bounded derivative" requirement. Many previous methods required that while an attack could be large, it couldn't accelerate too quickly.

Recent work, such as Lyapunov-based Resilient Secondary Synchronization Strategy of AC Microgrids Under Exponentially Energy-Unbounded FDI Attacks, has moved toward relaxing these constraints. By formulating the control law to handle "energy-unbounded" signals, these systems can survive attacks that are both massive in magnitude and rapidly accelerating. This is achieved by incorporating dynamic compensational terms that update based on local neighborhood information, effectively absorbing the malicious data directly into the control loop.

Architectural Variations: Digital Twins and Privacy

While the mathematical mechanism often involves adaptive gains, the architectural implementation varies:

  • The Digital Twin Layer: Yukang Cui, Lingbo Cao, and colleagues propose a Digital Twin framework that decouples the defense. A secure "Twin Layer" handles the logic of where the agents should be, while the physical layer uses adaptive parameters to cancel out "cocktail attacks" (combinations of DoS and FDI).
  • Privacy-Preserving Defense: In Privacy-Preserving Distributed Defense Framework for DC Microgrids, Shan Zuo and Yi Zhang show that these adaptive laws can be combined with data-masking. This ensures that while the system is adapting to an attack, it isn't leaking sensitive operational data to potential eavesdroppers.
  • Event-Triggered Communication: To save bandwidth and handle intermittent network disruptions (DoS), Yi Zhang and Mohamadamin Rajabinezhad utilize data-driven event-triggered mechanisms. The system only communicates when necessary, but remains resilient to asymmetric attacks when it does.

Why It Works: Uniformly Ultimately Bounded (UUB) Stability

In the presence of an infinite attack, it is often impossible to achieve "perfect" zero-error convergence. Instead, these researchers aim for Uniformly Ultimately Bounded (UUB) stability.

This means that while the attack might push the system slightly off-course, the adaptive mechanism guarantees the error will stay within a small, predictable "neighborhood" of the target state. By matching the growth rate of the attack—whether it is polynomial (Transient-Safe and Attack-Resilient Secondary Control) or exponential—the controller ensures that the system's energy remains contained, preventing the catastrophic divergence that would otherwise lead to hardware failure in power grids or robot collisions in the field.

Papers in this version

History & citationVersion 1 · 2026-10-04

Merged into another direction · Permanent link

View frozen version 1

Cite this direction

Latest version (a living page):

@misc{lacuna_resilient_control_for_unbounded_false_da_d7b2e155,
  title = {Resilient Control for Unbounded False Data Injection Attacks},
  author = {{Lacuna}},
  howpublished = {Lacuna research direction},
  year = {2026},
  url = {https://lacuna.tiptreesystems.com/direction/resilient-control-for-unbounded-false-data-injection-attacks/rdr_d7b2e155cf6e4d32bb1fc20b2f56d14e},
  note = {Latest version of a living research direction; cite a version for a fixed text. Accessed 2026-11-19},
}

This exact version (version 1, frozen text):

@misc{lacuna_resilient_control_for_unbounded_false_da_d7b2e155_v1,
  title = {Resilient Control for Unbounded False Data Injection Attacks},
  author = {{Lacuna}},
  howpublished = {Lacuna research direction},
  year = {2026},
  url = {https://lacuna.tiptreesystems.com/direction/resilient-control-for-unbounded-false-data-injection-attacks/rdr_d7b2e155cf6e4d32bb1fc20b2f56d14e/v/1},
  note = {Version 1, frozen 2026-10-04. Accessed 2026-11-19},
}