A Lightweight Modulation Classification Network Resisting White Box Gradient Attacks

Sicheng Zhang,Zhida Bao,Yun Lin,Jiangzhi Fu,Xin Liu

doi:10.1155/2021/8921485

Sicheng Zhang, Zhida Bao + Show 3 more

Open Access

https://doi.org/10.1155/2021/8921485

Copy DOI

Abstract

Improving the attack resistance of the modulation classification model is an important means to improve the security of the physical layer of the Internet of Things (IoT). In this paper, a binary modulation classification defense network (BMCDN) was proposed, which has the advantages of small model scale and strong immunity to white box gradient attacks. Specifically, an end-to-end modulation signal recognition network that directly recognizes the form of the signal sequence is constructed, and its parameters are quantized to 1 bit to obtain the advantages of low memory usage and fast calculation speed. The gradient of the quantized parameter is directly transferred to the original parameter to realize the gradient concealment and achieve the effect of effectively defending against the white box gradient attack. Experimental results show that BMCDN obtains significant immune performance against white box gradient attacks while achieving a scale reduction of 6 times.

Highlights

Introduction eInternet of ings (IoT) is an open and comprehensive network of intelligent objects
In the two networks with the same architecture, the parameters of float modulation classification defense network (FMCDN) are all floating-point data, and the fitting ability of the network is obviously better than that of binary modulation classification defense network (BMCDN), so the difference in accuracy is within expectations and acceptable
Aiming at the problem of high computational complexity and vulnerability to adversarial sample attacks when artificial intelligence is applied to the issue of IoT security, this paper designs a modulation classification defense model BMCDN that combines fast inference and antiwhite box gradient attacks to detect the malicious attacks and interference in the IoT

Summary

Introduction

Internet of ings (IoT) is an open and comprehensive network of intelligent objects. It is deployed in different environments through various sensor devices to realize realtime collection and interaction of different monitored, connected, and interactive objects or processes [1, 2]. Erefore, the physical layer security of the IoT has been brought to an unprecedentedly important position. As the basis of software wireless, cognitive radio, and spectrum detection, automatic modulation recognition has become an effective means to deal with physical layer security issues [8, 9]. Automatic modulation classification technology is used in various civil and military fields, such as user legitimacy detection, spectrum detection and management, interference and identification, electronic exhibition, and threat analysis [10,11,12,13]

Methods

Results

Conclusion