使用聲音驅動進行MEMS陀螺儀身份驗證之方法

Abstract

智慧工廠在監測環境變化方面使用大量傳感器，以取代傳統人工操作的方式並降低錯誤風險。然而，攻擊者可能竄改感測器數據，導致自動控制系統做出錯誤判斷，進而對工廠造成損失。因此，對傳感器數據進行驗證具有重要性。本研究提出了一種基於挑戰-響應的驗證系統，利用MEMS陀螺儀對聲波的獨特響應作為驗證基礎。我們透過超聲波挑戰MEMS陀螺儀，並從其響應中提取特徵，與驗證系統中的資料進行比對，以驗證接收到的數據是否具備正確特徵。為了克服環境干擾對驗證過程的影響，我們提出了一個策略式演算法，用於計算特徵之間的相似度。我們在實驗室環境中進行了多種測試，包括改變喇叭與傳感器之間的距離，以及引入噪音和多路徑干擾等環境條件。實驗結果顯示，我們的策略式演算法具有一定程度的抗環境干擾能力。此外，我們通過與目前最新的MEMS陀螺儀辨識系統進行比較，在蒙地卡羅方法模擬中，我們的研究能將特徵重複發生的機率降低為之前研究的五十分之一。綜上所述，我們的方法在面對干擾以及區分特徵方面都展現出良好的表現。

Smart factories employ a multitude of sensors to monitor environmental changes, replacing traditional manual operations and reducing the risk of errors. However, attackers can manipulate sensor data, leading to erroneous decisions by the automated control systems and resulting in losses for the factory. Therefore, it is crucial to verify the integrity of sensor data. In this study, we propose a challenge-response-based verification system that utilizes the unique response of MEMS gyroscopes to acoustic signals as a basis for authentication. We challenge the MEMS gyroscope using ultrasonic signals and extract features from its response, comparing them with the data in the verification system to confirm the presence of correct features in the received data. To overcome the impact of environmental interference on the verification process, we introduce a heuristic algorithm for computing the similarity between features. We conducted various tests in a laboratory environment, including varying the distance between the speaker and sensor, and introducing noise and multipath interference. The experimental results demonstrate that our heuristic algorithm exhibits sufficient robustness against environmental interference. Furthermore, compared to the latest MEMS gyroscope recognition systems, our research achieves a 50-fold reduction in the occurrence of feature duplication, as confirmed by Monte Carlo simulations. In summary, our approach demonstrates excellent performance in terms of resistance to interference and feature discrimination capabilities.