Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Exploring Adversarial Threat Models in Cyber Physical Battery Systems

Shanthan Kumar Padisala, Shashank Dhananjay Vyas, Satadru Dey

TL;DR

The paper tackles cyber-physical threats to CPBS by formulating a multi-objective optimal-control framework to generate adversarial input currents that drive overcharge or over-discharge while maintaining output stealth. It derives a first-order ECM, casts the attack problem in state-space form, and solves for optimal input attacks via Riccati-based equations, complemented by open-loop and feedback-enabled output attacks to mask effects. Experimental validation on a commercial Li-ion cell demonstrates stealthy injections that closely track nominal voltages under attack and yield the intended SOC trajectories, with sensitivity analysis guiding attack parameter choices. The work provides a principled basis for designing detection and mitigation strategies in BMS/CPS contexts and points to future integration with active cyber-defense and ML-based defense schemes.

Abstract

Technological advancements like the Internet of Things (IoT) have facilitated data exchange across various platforms. This data exchange across various platforms has transformed the traditional battery system into a cyber physical system. Such connectivity makes modern cyber physical battery systems vulnerable to cyber threats where a cyber attacker can manipulate sensing and actuation signals to bring the battery system into an unsafe operating condition. Hence, it is essential to build resilience in modern cyber physical battery systems (CPBS) under cyber attacks. The first step of building such resilience is to analyze potential adversarial behavior, that is, how the adversaries can inject attacks into the battery systems. However, it has been found that in this under-explored area of battery cyber physical security, such an adversarial threat model has not been studied in a systematic manner. In this study, we address this gap and explore adversarial attack generation policies based on optimal control framework. The framework is developed by performing theoretical analysis, which is subsequently supported by evaluation with experimental data generated from a commercial battery cell.

Exploring Adversarial Threat Models in Cyber Physical Battery Systems

TL;DR

The paper tackles cyber-physical threats to CPBS by formulating a multi-objective optimal-control framework to generate adversarial input currents that drive overcharge or over-discharge while maintaining output stealth. It derives a first-order ECM, casts the attack problem in state-space form, and solves for optimal input attacks via Riccati-based equations, complemented by open-loop and feedback-enabled output attacks to mask effects. Experimental validation on a commercial Li-ion cell demonstrates stealthy injections that closely track nominal voltages under attack and yield the intended SOC trajectories, with sensitivity analysis guiding attack parameter choices. The work provides a principled basis for designing detection and mitigation strategies in BMS/CPS contexts and points to future integration with active cyber-defense and ML-based defense schemes.

Abstract

Technological advancements like the Internet of Things (IoT) have facilitated data exchange across various platforms. This data exchange across various platforms has transformed the traditional battery system into a cyber physical system. Such connectivity makes modern cyber physical battery systems vulnerable to cyber threats where a cyber attacker can manipulate sensing and actuation signals to bring the battery system into an unsafe operating condition. Hence, it is essential to build resilience in modern cyber physical battery systems (CPBS) under cyber attacks. The first step of building such resilience is to analyze potential adversarial behavior, that is, how the adversaries can inject attacks into the battery systems. However, it has been found that in this under-explored area of battery cyber physical security, such an adversarial threat model has not been studied in a systematic manner. In this study, we address this gap and explore adversarial attack generation policies based on optimal control framework. The framework is developed by performing theoretical analysis, which is subsequently supported by evaluation with experimental data generated from a commercial battery cell.
Paper Structure (13 sections, 13 equations, 12 figures, 1 table)

This paper contains 13 sections, 13 equations, 12 figures, 1 table.

Table of Contents

  1. Introduction
  2. Motivation
  3. Need for Adversarial Threat Models
  4. Literature Review
  5. Contribution
  6. Problem Statement: Adversarial attacks in Battery Systems
  7. Adversarial threat model
  8. Description of battery equivalent circuit model (ECM)
  9. State-space representation of battery systems
  10. Optimal control-based input attack generation
  11. Open-loop model and real-time feedback-based output attack generation
  12. Results and Discussion
  13. Conclusion and future work

Figures (12)

  • Figure 1: A simplified schematic of cyber physical BESS, adopted and modified from kim2018cloud.
  • Figure 2: Adversarial attack injection into battery systems.
  • Figure 3: First order equivalent circuit battery model.
  • Figure 4: Experimental setup.
  • Figure 5: Dynamic current profile and comparison of experimental and model output voltages (RMSE 0.0483 V).
  • ...and 7 more figures

Theorems & Definitions (3)

  • Remark 1
  • Remark 2
  • Remark 3