State Estimation with Secrecy against Eavesdroppers

Anastasios Tsiamis, Konstantinos Gatsis, George J. Pappas

Abstract

We study the problem of remote state estimation in the presence of an eavesdropper. An authorized user estimates the state of a linear plant based on data received from a sensor, while the data may also be intercepted by the eavesdropper. To maintain confidentiality, we introduce a novel control-theoretic definition of perfect secrecy requiring that the user's expected error remains bounded while the eavesdropper's expected error grows unbounded. We propose a secrecy mechanism that guarantees perfect secrecy by randomly withholding sensor information, provided the user's packet reception rate exceeds the eavesdropper's interception rate. We explore the tradeoff between user utility and confidentiality via an optimization problem and provide insights through numerical examples.

Introduction

In the era of the Internet of Things (IoT), sensors frequently transmit critical data regarding physical systems—such as power grids or industrial plants—over wireless channels. The broadcast nature of wireless signals makes this data vulnerable to interception by eavesdroppers. While standard cryptographic solutions like AES are effective, they are often "application-agnostic" and do not leverage the underlying physical dynamics of the monitored system. This research explores whether the inherent instability of a physical system can be utilized to achieve secrecy. Specifically, the authors define "Perfect Expected Secrecy" as a state where a legitimate user maintains a low estimation error while the eavesdropper’s error becomes infinitely large.

System Model and Methodology

The problem is modeled using a linear dynamical system and an Intermittent Kalman Filter (IKF) framework. The architecture consists of a sensor monitoring a dynamical system and a secrecy mechanism that decides whether to transmit measurements.

System architecture for remote state estimation with a secrecy mechanism.

The core components of the methodology include:

  1. Secrecy Mechanism: The sensor uses a biased coin flip at each time step to decide whether to transmit the measurement $y(k)$.
  2. Channel Disparity: The model assumes the legitimate user has a higher packet reception rate ($p_1$) compared to the eavesdropper's interception rate ($p_2$).
  3. Exploiting Phase Transitions: IKF theory identifies a "critical probability" ($p_c$) for packet reception. If the reception rate is above $p_c$, the estimation error remains bounded; if it falls below $p_c$, the error diverges. The mechanism tunes the transmission probability $p$ such that the user’s effective rate ($p \cdot p_1$) is above $p_c$, while the eavesdropper’s rate ($p \cdot p_2$) is below it.
  4. Optimization: Since calculating exact IKF errors is computationally intensive, the authors utilize analytical upper and lower bounds to formulate a bisection-based optimization algorithm. This allows for minimizing user error while ensuring a minimum error threshold for the eavesdropper.

Key Results

The study establishes that perfect expected secrecy is achievable if and only if $p_1 > p_2$. By strategically withholding data, the transmitter can exploit the mathematical phase transition of Kalman filters.

Comparison of estimation errors over time for the user and eavesdropper, with and without the secrecy mechanism.

In numerical simulations of an unstable second-order system, the secrecy mechanism demonstrated significant efficacy. As shown in the figure above, without secrecy, both the user and the eavesdropper maintain relatively low, comparable errors. However, with the mechanism active, the user's time-averaged error is kept at approximately 6.2, while the eavesdropper's average error explodes to 2922.7. This confirms that instability can act as a shield: the faster a system "blows up," the more sensitive the estimation becomes to missing packets, making it easier to blind the eavesdropper.

Discussion and Future Work

A critical insight is that "perfect secrecy in expectation" does not guarantee the eavesdropper is always wrong; there may be brief periods where they intercept enough consecutive packets to gain an accurate estimate. Furthermore, the current model assumes the transmitter knows the eavesdropper's interception rate ($p_2$), which is challenging to estimate in practice.

Future research could address these limitations by:

  • Incorporating feedback (ACKs) from the user to enable non-random withholding strategies.
  • Accounting for "bursty" losses in real-world wireless channels (Markovian fading).
  • Developing stronger secrecy definitions that bound error at every time step rather than just in expectation.

Summary updated

Abstract

We study the problem of remote state estimation, in the presence of an eavesdropper. An authorized user estimates the state of a linear plant, based on the data received from a sensor, while the data may also be intercepted by the eavesdropper. To maintain confidentiality with respect to state, we introduce a novel control-theoretic definition of perfect secrecy requiring that the user's expected error remains bounded while the eavesdropper's expected error grows unbounded. We propose a secrecy mechanism which guarantees perfect secrecy by randomly withholding sensor information, under the condition that the user's packet reception rate is larger than the eavesdropper's interception rate. Given this mechanism, we also explore the tradeoff between user's utility and confidentiality with respect to the eavesdropper, via an optimization problem. Finally, some examples are studied to provide insights about this tradeoff.
BibTeX
@inproceedings{tsiamis2017state,
  author = {Anastasios Tsiamis and Konstantinos Gatsis and George J. Pappas},
  title = {{State Estimation with Secrecy against Eavesdroppers}},
  booktitle = {IFAC-PapersOnLine},
  year = {2017},
  doi = {10.1016/j.ifacol.2017.08.1563},
}
Figure