Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Detecting Adversarial Spectrum Attacks via Distance to Decision Boundary Statistics

Wenwei Zhao, Xiaowen Li, Shangqing Zhao, Jie Xu, Yao Liu, Zhuo Lu

TL;DR

This work tackles adversarial spectrum attacks in cooperative spectrum sensing by introducing a distance-to-decision-boundary (DDB) based detector. It derives an efficient, direction-guided method to compute DDBs tailored to spectrum data and uses a Kolmogorov–Smirnov test to compare training and testing DDB distributions for attack detection. Experiments on realistic RTL-SDR spectrum data with a 20-node network demonstrate near-$99\%$ detection rates and sub-$1\%$ false alarms, while achieving $54$–$64\%$ faster DDB computation than competing methods. The approach provides a practical, scalable defense against adversarial ML threats in wireless spectrum sensing.

Abstract

Machine learning has been adopted for efficient cooperative spectrum sensing. However, it incurs an additional security risk due to attacks leveraging adversarial machine learning to create malicious spectrum sensing values to deceive the fusion center, called adversarial spectrum attacks. In this paper, we propose an efficient framework for detecting adversarial spectrum attacks. Our design leverages the concept of the distance to the decision boundary (DDB) observed at the fusion center and compares the training and testing DDB distributions to identify adversarial spectrum attacks. We create a computationally efficient way to compute the DDB for machine learning based spectrum sensing systems. Experimental results based on realistic spectrum data show that our method, under typical settings, achieves a high detection rate of up to 99\% and maintains a low false alarm rate of less than 1\%. In addition, our method to compute the DDB based on spectrum data achieves 54\%--64\% improvements in computational efficiency over existing distance calculation methods. The proposed DDB-based detection framework offers a practical and efficient solution for identifying malicious sensing values created by adversarial spectrum attacks.

Detecting Adversarial Spectrum Attacks via Distance to Decision Boundary Statistics

TL;DR

This work tackles adversarial spectrum attacks in cooperative spectrum sensing by introducing a distance-to-decision-boundary (DDB) based detector. It derives an efficient, direction-guided method to compute DDBs tailored to spectrum data and uses a Kolmogorov–Smirnov test to compare training and testing DDB distributions for attack detection. Experiments on realistic RTL-SDR spectrum data with a 20-node network demonstrate near- detection rates and sub- false alarms, while achieving – faster DDB computation than competing methods. The approach provides a practical, scalable defense against adversarial ML threats in wireless spectrum sensing.

Abstract

Machine learning has been adopted for efficient cooperative spectrum sensing. However, it incurs an additional security risk due to attacks leveraging adversarial machine learning to create malicious spectrum sensing values to deceive the fusion center, called adversarial spectrum attacks. In this paper, we propose an efficient framework for detecting adversarial spectrum attacks. Our design leverages the concept of the distance to the decision boundary (DDB) observed at the fusion center and compares the training and testing DDB distributions to identify adversarial spectrum attacks. We create a computationally efficient way to compute the DDB for machine learning based spectrum sensing systems. Experimental results based on realistic spectrum data show that our method, under typical settings, achieves a high detection rate of up to 99\% and maintains a low false alarm rate of less than 1\%. In addition, our method to compute the DDB based on spectrum data achieves 54\%--64\% improvements in computational efficiency over existing distance calculation methods. The proposed DDB-based detection framework offers a practical and efficient solution for identifying malicious sensing values created by adversarial spectrum attacks.
Paper Structure (31 sections, 20 equations, 11 figures, 2 tables)

This paper contains 31 sections, 20 equations, 11 figures, 2 tables.

Table of Contents

  1. Introduction
  2. System and Threat Models
  3. System Model
  4. Threat Model
  5. Problem Statement
  6. Detection Leveraging DDB
  7. Decision Boundary and DDB
  8. Definitions
  9. Finding Decision Boundary and DDB
  10. Proposed Method to Compute DDB for Spectrum Sensing
  11. Motivation
  12. Finding the Direction to Compute DDB
  13. Analysis and Design
  14. Detection Method
  15. Can Attack Bypass DDB-based Detection?
  16. ...and 16 more sections

Figures (11)

  • Figure 1: Cooperative sensing scenario with malicious nodes controlled by an attacker.
  • Figure 2: Our idea to find the direction vs an iterative method to find the DDB.
  • Figure 3: Detection rates under the different test group sizes.
  • Figure 4: False alarms under the different sizes of test data.
  • Figure 5: Detection rates under different attack occurrence ratios.
  • ...and 6 more figures