Strategic Remote Attestation: Testbed for Internet-of-Things Devices and Stackelberg Security Game for Optimal Strategies

Shanto Roy, Salah Uddin Kadir, Yevgeniy Vorobeychik, Aron Laszka

Abstract

Internet of Things (IoT) devices are increasingly targeted by memory-injection attacks due to their lack of robust security primitives. Remote Attestation (RA) offers a solution by verifying software integrity, but its deployment is constrained by the limited computational resources of IoT hardware. This paper presents a game-theoretic framework using a Stackelberg Security Game (SSG) to optimize RA strategies. By modeling the interaction between a defender and an attacker, we derive optimal attestation probabilities that balance security risks against performance overhead. Experimental validation on a physical IoT testbed demonstrates that strategic attestation significantly outperforms uniform or "always-on" defense mechanisms.

Introduction

The proliferation of IoT devices in critical infrastructure and smart homes has created a vast attack surface. These devices often run on low-power hardware and lack modern memory protections like Address Space Layout Randomization (ASLR), making them vulnerable to exploits. Remote Attestation (RA) is a promising defense where an external verifier checks the device's memory state. However, the computational cost of RA—consuming CPU cycles and battery—presents a significant hurdle. A naive approach of constant attestation is too costly, while infrequent attestation leaves windows of vulnerability. The core challenge lies in strategically deploying attestation to maximize security while minimizing performance costs.

System Architecture

The RA process involves a verifier challenging an IoT device to provide a proof of its memory state, typically via a checksum. This interaction is visualized in the system model below.

Figure 1: Overview of the Remote Attestation architecture involving an external verifier, multiple IoT servers, and a potential attacker.

Methodology

The researchers developed a physical testbed using Raspberry Pi 3 Model B+ devices to simulate real-world IoT applications, specifically an Irrigation Server and a Smart Home system. To optimize the defense, they modeled the scenario as a Stackelberg Security Game (SSG). In this model, the defender (leader) commits to a randomized attestation strategy, and the attacker (follower) chooses their target after observing this strategy.

Game-Theoretic Modeling

The game is defined by a set of constants, variables, and utility functions that capture the costs and gains for both parties. The defender aims to maximize their expected utility $U_D(\mathbf{p}, \mathbf{a})$ by choosing an optimal strategy vector $\mathbf{p}$, while the attacker seeks to maximize $U_A(\mathbf{p}, \mathbf{a})$.

Figure 2: Summary of symbols, variables, and functions used to define the Stackelberg Security Game for optimal remote attestation.

The model accounts for asymmetric costs, such as the defender's computational overhead and the attacker's exploit development costs. By deriving analytical solutions, the authors identified optimal strategies for single-device and multi-device scenarios.

Results and Key Insights

Experimental results confirmed a linear relationship between memory coverage, detection rate, and computational overhead. A key finding is the "Power of Deterrence": by strategically setting attestation probabilities, the defender can make the attacker's expected utility negative, effectively preventing attacks at a lower cost than continuous monitoring.

Numerical evaluations showed that the optimal strategy $p^*$ consistently resulted in significantly lower losses compared to baselines:

  • Never attest ($p=0$): High security loss due to unchecked compromises.
  • Always attest ($p=1$): Prohibitively high computational cost.
  • Uniform attestation: Inefficient allocation of resources across different device classes.

Conclusion

Strategic remote attestation provides a robust framework for securing resource-constrained IoT devices. By leveraging game theory, defenders can achieve high security levels while maintaining device performance. The physical testbed validation underscores the practical applicability of the mathematical model. Future work includes exploring multi-method attestation and applying reinforcement learning to handle dynamic, non-linear environments.

Summary updated

BibTeX
@inproceedings{roy2021strategic,
  author = {Shanto Roy and Salah Uddin Kadir and Yevgeniy Vorobeychik and Aron Laszka},
  title = {{Strategic Remote Attestation: Testbed for Internet-of-Things Devices and Stackelberg Security Game for Optimal Strategies}},
  booktitle = {GameSec},
  year = {2021},
  url = {https://openreview.net/forum?id=HJhBunRAb6R},
}
Figure