Tamper-resistant controller using neural network and time-varying quantization
- PDF / 1,409,040 Bytes
- 7 Pages / 595.276 x 790.866 pts Page_size
- 23 Downloads / 178 Views
ORIGINAL ARTICLE
Tamper‑resistant controller using neural network and time‑varying quantization Fangyuan Xu1 · Ryo Ariizumi1 · Shun‑ichi Azuma1 · Toru Asai1 Received: 20 May 2020 / Accepted: 14 September 2020 / Published online: 13 October 2020 © International Society of Artificial Life and Robotics (ISAROB) 2020
Abstract In this paper, we consider a tamper-resistant control system aiming at protecting the knowledge of the controller from attackers. In this control system, the controller operates normally only for a limited number of time-varying specific states; otherwise, it outputs an incorrect value. We propose to realize the tamper-resistant controller by employing a neural network and time-varying quantization. Furthermore, we make it possible for only one trained neural network to be used for all quantization based on the local approximation linearity of the trained neural network. Without this approach, the neural network needs to be trained for every possible quantization, which leads to huge computation. We provide simulations to demonstrate the security and feasibility of the proposed method. Keywords Neural network · Tamper-resistant controller · Time-varying quantization · Linear state-feedback
1 Introduction The security of the controller is an important topic nowadays because control systems are widely applied in many fields [1]. There are many kinds of attacks aiming at stealing information of the controller, such as physical attacks and logical attacks [2]. Such attacks may result in leakage of commercial or military secrets, incurring enormous economic losses and even national security issues. With the continuous evolution of the attack technique, the information security problem of the controller demands prompt solutions. One way to address the information security problem of the controller is to encrypt the related information. In recent years, many studies have focused on this. In [3], the authors proposed to encrypt the important information using the modified homomorphic encryption schemes based on public-key RSA [4] and ElGamal [5] encryption systems. In this method, not only the signals over the communication This work was presented in part at the 3rd International Symposium on Swarm Behavior and Bio-Inspired Robotics (Okinawa, Japan, November 20–22, 2019). * Fangyuan Xu [email protected]‑u.ac.jp 1
Graduate School of Engineering, Nagoya University, Nagoya, Japan
13
Vol:.(1234567890)
links but also parameters of the controller are encrypted. In [6], the authors introduced a method using the fully homomorphic encryption. It has enabled arithmetic operations directly on the encrypted variables without decryption. However, such encryption has a finite lifespan, which decreases as an arithmetic operation is performed on the encrypted variables. Recently, another security solution for the control system is proposed. In [7], the researchers proposed a tamper-resistant controller, which was achieved using a neural network and time-invariant quantization. The idea is that with the v
Data Loading...
