Workshops

Session AidTSP-TS1: AidTSP Technical Session I Session AidTSP-TS2: AidTSP Technical Session II Session AidTSP-TS3: AidTSP Technical Session III Session AidTSP-TS4: AidTSP Technical Session IV
Session AoI-Open: Opening Session Session AoI-Key: Keynote Session AoI-SCH: Scheduling and Optimization for AoI Session AoI-CBrk: Virtual Coffee Break Session AoI-BEY: Beyond AoI Metrics
Session CNERT-O: Opening Session Session CNERT-K: Keynote Session CNERT-P: Panel Session CNERT-S1: Session 1, Experimentation Enabling Capabilities Session CNERT-S2: Session 2, Services and network design Session CNERT-S3: Session 3, Wireless Session CNERT-D: Demo Session
Session DeepWireless-OS: Opening Session Session DeepWireless-KS1: Keynote Session 1 Session DeepWireless-TS1: Technical Session 1 - Deep Learning for Wireless Security Session DeepWireless-TS2: Technical Session 2 - Deep Learning for Wireless Applications Session DeepWireless-KS2: Keynote Session 2 Session DeepWireless-TS3: Technical Session 3 - Deep Learning for 5G Communications Session DeepWireless-CS: Closing Session
Session DroneCom-TS1: Technical Session I Session DroneCom-KS1: Keynote Session I Session DroneCom-TS2: Technical Session II Session DroneCom-TS3: Technical Session III Session DroneCom-KS2: Keynote Session II Session DroneCom-TS4: Technical Session IV
Session FOGML-S1: Opening Remarks Session FOGML-S2: Federated Learning Session FOGML-S3: Distributed Learning and Edge Computing
Session GI-00: Session#0 - Opening Session with Invited Keynote Session GI-01: Session#1 - Applying Machine Learning to Next Generation Internet Session GI-02: Session#2 - Performance Optimization of Next Generation Internet
Session ICCN-KS1: Security in Wireless Body Area Networks Session ICCN-B1: Virtual Coffee Break Session ICCN-S1: Cloud and Edge Computing Session ICCN-S2: Cloud and Edge Security Session ICCN-B2: Virtual Lunch Break Session ICCN-S3: Cloud and Edge Computing Session ICCN-S4: Cloud and Edge Computing
Session MobiSec-S1: AI & Security (I) Session MobiSec-K: Keynote: AI for Cybersecurity and Security of AI Session MobiSec-S2: AI & Security (II) Session MobiSec-S3: Mobile & Edge Security Session MobiSec-S4: Communication Security
Session NG-OPERA-OS: Opening Session Session NG-OPERA-KS: Keynote Session Session NG-OPERA-PS: Panel Session Session NG-OPERA-Break-1: Break Session NG-OPERA-TS1: Technical Session 1: Machine Learning and Resource Optimization for 6G RANs Session NG-OPERA-Lunch: Lunch Session NG-OPERA-TS2: Technical Session 2: System-level Solutions for Open RANs Session NG-OPERA-Break-2: Break Session NG-OPERA-TS3: Technical Session 3: Non-terrestrial Integration and Posters Session NG-OPERA-ACS: Awards and Closing
Session NetSciQCom-S1: Keynote Talk Session NetSciQCom-S2: Contributed Talks
Session PerAI-6G-Opening: Opening Session Session PerAI-6G-Keynote-1: Keynote Session 1 Session PerAI-6G-Keynote-2: Keynote Session 2 Session PerAI-6G-Session-1: Session 1: AI applications in 6G Session PerAI-6G-Session-2: Session 2: Distributed Learning in 6G Session PerAI-6G-Session-3: Session 3: Digital Twin in 6G Session PerAI-6G-Session-4: Session 4: Resource Management in 6G Session PerAI-6G-Closing: Closing Session
Session WISARN-O: Opening Session WISARN-K1: Keynote Session WISARN-CB: Virtual Coffee Break Session WISARN-S1: Technical Session
Session Wireless-Sec-I: Wireless Security I Session Wireless-Sec-II: Wireless Security II

Session MobiSec-S1

AI & Security (I)

Conference
8:00 AM — 9:10 AM EDT
Local
May 20 Sat, 8:00 AM — 9:10 AM EDT

Bewitching the Battlefield: Repurposing the MouseJack Attack for Crazyflie Drones

Narayana Murari Gowrishetty (University of Maryland Baltimore County, USA); Sai Sree Laya Chukkapalli (University of Maryland, Baltimore County, USA); Anupam Joshi (University of Maryland, Baltimore County, USA)

1
The Internet of Battlefield Things (IoBT) connects autonomous, uncrewed air and ground vehicles with wireless networks. It is a promising technology that has the potential to significantly enhance the operational effectiveness of military units by providing them with real-time information about the battlefield and enhancing their situational awareness. Autonomous assets such as drones and sensors can gather data from various sources and transmit it back to the command center, where it can be used to make informed decisions. However, the trustworthiness of the data provided by these autonomous assets depends on the systems' security and resilience to adversary attacks. Adversaries can potentially exploit vulnerabilities in the system to disrupt or manipulate the data. For example, adversaries can potentially hijack drones or disrupt the network bandwidth, causing Denial of Service (DoS) attacks or injecting fake data. In this paper, we demonstrate one such attack by showing how certain drone systems can be compromised by sniffing unencrypted wireless network channels to extract key information, create packets with chosen payloads, and send it to an unsuspecting host to affect its behavior. Specifically, we demonstrate that the initial steps of the MouseJack attack can be combined with packet reverse engineering to take control of Crazyflie drones.
Speaker Narayana Murari (University of Maryland Baltimore County)

Narayana Murari received Masters in Computer Science from University of Maryland Baltimore Country in Dec 2022 and worked as a Research Assistant under Dr Anupam Joshi. He is currently working as a Software Engineer at Walmart Global Tech.


Artificial Neural Network and Game Theory for Secure Optimal Charging Station Selection for Evs

Riya Kakkar (Nirma University, India); Aparna Kumari (Nirma University, India); Rajesh Gupta (Nirma University, India); Smita Agrawal (Nirma University, India); Sudeep Tanwar (Nirma University, India)

0
The penetration of electric vehicles (EVs) entails the deployment of more charging station (CS) infrastructure to realize the charging requirement issues of the EVs. But, limited installation of charging infrastructure and data security issues require a secure and efficient CS selection mechanism for EVs. Towards this goal, we proposed an Artificial Intelligence (AI) and game theory-based secure CS selection scheme for EVs using blockchain. Blockchain and AI-based proposed scheme provide security and privacy during the communication between partic- ipants, i.e., EVs and CSs, for optimal CS selection. Moreover, an incorporated blockchain network with Interplanetary File System (IPFS) strengthens the reliability and cost-efficiency of CS selection by using a 6G network and its ultra-intelligent features. Furthermore, the blockchain and AI-based proposed scheme utilizes coalition game theory approach to recommend the optimal CS for EV and balance the fair payoff between the participants in the network. Finally, experimental results show that the proposed scheme yields better results than the conventional approaches considering the performance evaluation metrics such as State of Charge (SoC), profit analysis, and latency comparison.
Speaker Riya Kakkar (Nirma University)

Riya Kakkar is a Full-Time Ph.D. Research Scholar in the Computer science and Engineering Department at Nirma University, Ahmedabad, Gujarat, India. She received her Bachelor as well as Master of Technology from the Banasthali Vidyapith, Jaipur, India in 2018 and 2021, respectively. She has authored or co-authored some publications (including papers in SCI Indexed Journal and IEEE ComSoc sponsored International Conference). Some of her research findings are published in top-cited journals and conferences such as IEEE Systems Journal, IEEE IoT Journal, JISA Journal, Wiley IJER, IEEE CITS, IEEE ICC, IEEE INFOCOM, and many more. Her research interest includes, Electric Vehicles, Blockchain Technology, 5G Communication Network, and Machine Learning. She is also an active member of ST Research Laboratory (www.sudeeptanwar.in).

Intelligent TOR Onion Routing Framework for Improving Anonymity in IoT Military Applications

Sucheta Gupta (Gujarat Technological University, India); Sushant Mahajan (Gujarat Technological University, India)

0
This paper overcomes the security and privacy issues of the Internet of military vehicles (IoMVs), wherein the attackers intercept and manipulates the IoMV communication to disrupt military operations. Moreover, IoMV communication uses the public Internet to route their sensitive data, which is open to several network-related attacks, such as data injection, session hijacking, privilege escalation, etc. To overcome the aforesaid security and privacy concerns, we proposed a blockchain and artificial intelligence (AI)-assisted onion routing framework to strengthen the security and anonymity of IoMVs participating in military operations. AI algorithms are amended to classify the malicious and non-malicious IoMVs so that only non-malicious IoMVs can participate in the data exchange via onion routing network. In addition, each IoMV is uniquely identified by two identifiers, i.e., tag and verifying token, which are stored in the block of a blockchain. These identifiers authenticate and validate the participating IoMVs. Based on that, nonmalicious IoMVs can participate in the onion routing network for secure transmission of sensitive data. Further, each entity of the proposed framework utilizes the indispensable characteristics of a 6G network to minimize the data loss and validated using the packet drop ratio parameter. Lastly, the proposed framework is evaluated by considering different valuation metrics, such as statistical measures, and blockchain processing time.
Speaker
Speaker biography is not available.
Session Chair

Lei Chen (Georgia Southern University, United States); Wenjia Li (New York Institute of Technology, United States); Yun Lin (Harbin Engineering University, P.R. China)

Session MobiSec-K

Keynote: AI for Cybersecurity and Security of AI

Conference
9:15 AM — 10:00 AM EDT
Local
May 20 Sat, 9:15 AM — 10:00 AM EDT

Keynote: AI for Cybersecurity and Security of AI

Houbing Song, Ph.D., IEEE Fellow (University of Maryland, Baltimore County, USA)

0
The mutual needs and benefits of AI and cybersecurity have been widely recognized. AI techniques are expected to enhance cybersecurity by assisting human system managers with automated monitoring, analysis, and responses to adversarial attacks. Conversely, it is essential to guard AI technologies from unintended uses and hostile exploitation by leveraging cybersecurity practices. The interplay between AI/machine learning, and cybersecurity introduces new opportunities and challenges in the security of AI as well as AI for cybersecurity. In this talk, I will present my recent research on AI for cybersecurity and the security of AI. First, I will introduce my research on AI for cybersecurity, i.e., real-time machine learning for quickest event (threat/intrusion/vulnerability…) detection. Next, I will present my research on the security of AI, i.e., coverage-driven testing and evaluation of deep learning systems. I will conclude my presentation with my ongoing research on neurosymbolic AI.
Speaker Houbing Song (University of Maryland, Baltimore County)

Houbing Song (M’12–SM’14-F’23) received the Ph.D. degree in electrical engineering from the University of Virginia, Charlottesville, VA, in August 2012.


He is currently a Tenured Associate Professor, the Director of the NSF Center for Aviation Big Data Analytics (Planning), the Associate Director for Leadership of the DOT Transportation Cybersecurity Center for Advanced Research and Education (Tier 1 Center), and the Director of the Security and Optimization for Networked Globe Laboratory (SONG Lab, www.SONGLab.us), University of Maryland, Baltimore County (UMBC), Baltimore, MD. Prior to joining UMBC, he was a Tenured Associate Professor of Electrical Engineering and Computer Science at Embry-Riddle Aeronautical University, Daytona Beach, FL. He serves as an Associate Editor for IEEE Transactions on Artificial Intelligence (TAI) (2023-present), IEEE Internet of Things Journal (2020-present), IEEE Transactions on Intelligent Transportation Systems (2021-present), and IEEE Journal on Miniaturization for Air and Space Systems (J-MASS) (2020-present). He was an Associate Technical Editor for IEEE Communications Magazine (2017-2020). He is the editor of eight books, the author of more than 100 articles and the inventor of 2 patents. His research interests include cyber-physical systems/internet of things, cybersecurity and privacy, and AI/machine learning/big data analytics. His research has been sponsored by federal agencies (including National Science Foundation, US Department of Transportation, and Federal Aviation Administration, among others) and industry. His research has been featured by popular news media outlets, including IEEE GlobalSpec's Engineering360, Association for Uncrewed Vehicle Systems International (AUVSI), Security Magazine, CXOTech Magazine, Fox News, U.S. News & World Report, The Washington Times, and New Atlas.


Dr. Song is an IEEE Fellow (for contributions to big data analytics and integration of AI with Internet of Things), and an ACM Distinguished Member (for outstanding scientific contributions to computing). He is an ACM Distinguished Speaker (2020-present) and an IEEE Vehicular Technology Society (VTS) Distinguished Lecturer (2023-present). Dr. Song has been a Highly Cited Researcher identified by Clarivate™ (2021, 2022) and a Top 1000 Computer Scientist identified by Research.com. He received Research.com Rising Star of Science Award in 2022 (World Ranking: 82; US Ranking: 16). In addition to 2021 Harry Rowe Mimno Award, Dr. Song was a recipient of 10+ Best Paper Awards from major international conferences, including IEEE CPSCom-2019, IEEE ICII 2019, IEEE/AIAA ICNS 2019, IEEE CBDCom 2020, WASA 2020, AIAA/ IEEE DASC 2021, IEEE GLOBECOM 2021 and IEEE INFOCOM 2022.

Session Chair

Lei Chen (Georgia Southern University, United States); Wenjia Li (New York Institute of Technology, United States); Yun Lin (Harbin Engineering University, P.R. China)

Session MobiSec-S2

AI & Security (II)

Conference
10:30 AM — 12:00 PM EDT
Local
May 20 Sat, 10:30 AM — 12:00 PM EDT

Attack Evaluations of Deep Learning Empowered WiFi Sensing in IoT Systems

Jianchao Song (Towson University, USA); Cheng Qian (Towson University, USA); Yifan Guo (Towson University, USA); Kun Hua (Lawrence Technological University, USA); Wei Yu (Towson University, USA)

1
Recently, the market has witnessed fabulous growths of WiFi sensing technologies to advance the deployment of IoT systems. Besides being an essential Internet of Things (IoT) enabler, WiFi devices with smart sensing techniques carry more helpful information to improve the performance of IoT systems. Nevertheless, little literature considers the potential security threats of smart WiFi sensing assisted IoT systems. In this paper, we investigate the vulnerability of existing deep learning empowered WiFi sensing systems under False Data Injection (FDI) attacks. First, we analyze the attack vendors in three different IoT layers and consider violating the data integrity in three dimensions, e.g., time, space, and value. Then, we design three attack schemes, respectively, to realize FDI attacks on smart WiFi sensing models. By measuring the performance of activity recognition on seven datasets under diverse WiFi sensings, experimental results have shown that the accuracy of activity recognition drastically decreases under our attacks.
Speaker Jianchao Song (Towson University)

Jianchao Song received the B.S degree from Jilin University, Changchun, China, in 2003, and the M.S degree in Electrical and Computer Engineering from Lawrence Technological University in 2021. Currently, he is pursuing the doctoral degree at Towson University. His research interests include machine learning, Internet of Things, and cybersecurity. 

Asynchronous Federated Learning for Intrusion Detection in Vehicular Cyber-Physical Systems

Sunitha Safavat (Howard University, USA); Danda B. Rawat (Howard University, USA)

0
In recent years, development in IoV technologies has reached more promising progress. IoV technology helps vehicles interact and exchange information between public networks and the surrounding environment, which reduces road congestion. In order to protect the information from attack and to provide efficient data transmission, this paper proposes secure federated learning for a vehicular cyber-physical system using an Interpolated public key and private key-ROTation (IPP-ROT)-based Elliptic Curve Cryptography (ECC) and Fed Buff: Federated Learning with Buffered Asynchronous Aggregation based Log Sigmoid Multi-Layer Perceptron (FB-FL-BAA-LSMLP) techniques. Initially, the vehicles are registered with a cloud server by generating keys and cipher text using ECC and IPP-ROT algorithms. After that, vehicle parameters are sensed by the server. As a large number of vehicles cross the Road Side Units (RSU), hashing is performed to authenticate the vehicle crossing RSUs using the Digit Folding-based Hash of Variable Length (DF-HAVAL) algorithm to avoid data collisions and uneven delays. Further, the data classification performed using FB-FL-BAA-LSMLP, which classifies data, and attacked data will be detected. At last, the performance of the proposed method is verified by comparing it with the existing techniques, and the results show better performance than the other methods.
Speaker Sunitha Safavat (Howard University, USA)

Postdoc Fellow at Howard University

Securing PUFs against ML Modeling Attacks via an Efficient Challenge-Response Approach

Mieszko Ferens (Aalborg University, Denmark); Edlira Dushku (Aalborg University, Denmark); Sokol Kosta (Aalborg University, Denmark)

0
Physical Unclonable Functions (PUFs) are lightweight security primitives capable of providing functionalities such as device authentication and identification. Such lightweight solutions are particularly important for small resource-constrained devices that cannot support many of the standard security mechanisms like e.g., TPMs. Even though PUFs are constructed to be unpredictable and unclonable, they have been susceptible to Machine Learning (ML) modeling attacks. Mitigation of these attacks typically requires additional hardware, causing potential deviation from the lightweight nature of low-end embedded devices. In this paper, we analyze the technical details that lead to the success of the previous ML modeling attacks, and utilize these findings to devise a novel challenge-response approach that improves PUF's security, more specifically the 4-XOR and 5-XOR PUFs, without additional hardware requirements. Our experimental results show that the proposed approach reduces modeling accuracies of state-of-the-art ML attacks by 10−15%, lowering the success rate of attacks significantly while remaining practical in the implementation.
Speaker Mieszko Ferens (Aalborg University)

PhD fellow at the dept. of Electronic Systems in Aalborg University. His research focuses on hardware security of small resource-constrained devices with a focus on the applications of Machine Learning. Before joining Aalborg University, he participated in Vulcanus in Japan 2021/2022 where he worked on Urban Digital Twins with NTT Data Technology and Innovation Headquarters. MSc (2022) and BSc (2020) in Telecommunication Engineering at the University of Valladolid, he has a background in Electrical Engineering where he applied Reinforcement Learning techniques to Computation Offloading and Dynamic Routing problems.

Cyberbullying Detection on Social Media using Machine Learning

Biodoumoye George Bokolo (Sam Houston State University, USA); Qingzhong Liu (Sam Houston State University, USA)

1
Social media platforms have seen an increase in the prevalence of cyberbullying. Making social media platforms safer from cyberbullying is essential given the popularity and extensive use of social media among people of all ages. This is a comparison study using Support Vector Machines, Naive Bayes, and bidirectional long short-term memory. We evaluated a Twitter dataset, and the results ranged from 85% for Naive Bayes to 98% for the Bi-LSTM model, and SVM was at 97%. Our research offers numerous insightful conclusions regarding how to identify cyberbullying.
Speaker Biodoumoye George Bokolo(Sam Houston State University)

I am a third-year doctoral student at Sam Houston State University studying digital and cyber forensics science. My research area is social media forensics.

Session Chair

Lei Chen (Georgia Southern University, United States); Wenjia Li (New York Institute of Technology, United States); Yun Lin (Harbin Engineering University, P.R. China)

Session MobiSec-S3

Mobile & Edge Security

Conference
2:00 PM — 3:30 PM EDT
Local
May 20 Sat, 2:00 PM — 3:30 PM EDT

Information We Can Extract About a User From 'One Minute Mobile Application Usage'

Sarwan Ali (Georgia State University, USA)

1
Understanding human behavior is important and has applications in many domains, such as targeted advertisement, health analytics, security, entertainment, etc. For this purpose, designing a system for activity recognition (AR) is important. However, understanding and analyzing common patterns become challenging since every human can have different behaviors. Since smartphones are easily available to every human being in the modern world, using them to track human activities becomes possible. In this paper, we extracted different human activities using the accelerometer, magnetometer, and gyroscope sensors of android smartphones by building an android mobile application. Using different social media applications, such as Facebook, Instagram, Whatsapp, and Twitter, we extracted the raw sensor values along with the attributes of $29$ subjects along with their attributes (class labels) such as age, gender, and left/right/both hands application usage. We extract features from the raw signals and use them to perform classification using different machine learning (ML) algorithms. Using statistical analysis, we show the importance of different features in predicting class labels. Ultimately, we use the trained ML model on our data to extract unknown features from well-known activity recognition data from the UCI repository, highlighting the potential of privacy breaches using ML models. This security analysis could help researchers in the future to take appropriate steps to preserve the privacy of human subjects.
Speaker Sarwan Ali

Ph.D. student at Georgia State University

Trust-Aware Resource Management for Secure and Optimal Network Slicing in 5G Mobile Edge Networks

Chen Peng (New York University, USA); Quanyan Zhu (New York University, USA)

1
In 5G mobile edge networks, the increased attack surface promotes the need for trust management while maintaining optimal resource provisioning. This work addresses this challenge by proposing a trust management scheme that monitors tasks' end-to-end delay to secure the network against denial-of-service attacks. We formulate an optimal network slicing problem that the overall end-to-end delay. Leveraging techniques from distributed optimization and Bayesian methods, we develop iterative algorithms to achieve a trust-aware resource management scheme that can achieve an efficient and resilient network resource provisioning. We use the edge network in hierarchical tree topology and three kinds of 5G edge devices to evaluate the proposed scheme. Extensive simulation results show the effectiveness and scalability of the proposed scheme, which can quickly help services recover from cyber-attacks.
Speaker Chen Peng (New York University)

Chen Peng received her Bachelor’s degree in Electronic Information Science and Technology from Northeast Normal University (NNU) in Changchun, Jilin, in 2020 and her Master's in Computer Engineering from New York University (NYU) in 2023. Now she is a first-year Ph.D. student at Purdue University. Her research interests are communication networks, cyber security, and mobile edge computing.

Edge Oriented Redistribution of Computational Load for Authentication Systems

Zakaria El-Awadi (Louisiana Tech University, USA); Manki Min ( Louisiana Tech University, USA)

1
Having an authentication system that is secure can be rather computationally expensive on the server and even affect user experience. This can affect user experience due to high demand by many users during the authentication process. Furthermore, users may find an authentication system that requires more interaction to be inconvenient and may opt out of the current standard 2FA options. We propose a method of using hash chains that are computed on the client side, by a verified device, to alleviate some computational overhead of a server, while also providing high security during the transmission of secure information. This system hopes to be less inconvenient for a user by only requiring them to type in a username/password and scan a QR Code.
Speaker Zakaria El-Awadi (Louisiana Tech University)

My name is Zakaria El-Awadi, I am a PhD candidate at Louisiana Tech University and my current research focus is on Authentication Systems and more specifically multi-factor approaches.

Android Malware Classification and Optimisation Based on BM25 Score of Android API

Rahul Yumlembam (Northumbria University, UK); Biju Issac (Northumbria University, UK); Longzhi Yang (Northumbria University, UK); Seibu Mary Jacob (Northumbria University, UK)

1
With the growth of Android devices, there is a rise in malware applications affecting these networked devices. Android malware classification is an important task in ensuring the security and privacy of Android devices. One promising approach to this problem is to capture the difference in the usage of API in benign and malware applications through the BM25 (Best Matching 25) scoring function by calculating the BM25 score of each API (Application Program Interface). A linear regression model is fitted using the BM25 score to select the 1000 most important APIs using the feature importance weight of the linear regression model. The selected APIs BM25 score and the Permission and Intents of an application are used to train Naïve Bayes, Random Forest, Decision Tree, Support Vector Machine, and CNN (Convolutional Neural Network) for classification. To illustrate the effectiveness of using the BM25 score of APIs for malware classification, we train the optimised Particle Swarm Optimisation (PSO) based Machine learning and Deep Learning algorithms using Permission and Intents features with and without the BM25 score. Experiments show that the BM25 score improves the result. Overall, this study demonstrates the potential of using the BM25 score of API calls, in combination with Permissions and Intents, as a valuable tool for Android malware classification.
Speaker Rahul Yumlembam

Rahul Yumlembam is currently pursuing a Ph.D. degree in Computer and Information Sciences at Northumbria University, United Kingdom. He received B. Tech degree in Computer Science and Engineering from Visvesvaraya Technological University and M.Tech in Computer Science and Engineering(Artificial Intelligence) from Assam Don Bosco University. Before joining the Ph.D. program in Northumbria, he was a Project Fellow at IIT, Guwahati, India working on various development and research project. His research interest includes Machine Learning, Deep Learning, Cyber Security using AI, Big Data, Brain Computer Interface, Text Mining, and Image Processing.

Session Chair

Lei Chen (Georgia Southern University, United States); Wenjia Li (New York Institute of Technology, United States); Yun Lin (Harbin Engineering University, P.R. China)

Session MobiSec-S4

Communication Security

Conference
4:00 PM — 5:30 PM EDT
Local
May 20 Sat, 4:00 PM — 5:30 PM EDT

MDPD: Mapping-based Data Processing Defense against Adversarial Attacks in Automatic Modulation Classification

Yan Zheng (Harbin Engineering University, China); Lin Qi (Harbin Engineering University, China); Zhida Bao (Harbin Engineering University, China); Jiangzhi Fu (Harbin Engineering University, China); Hang Jiang (Harbin Engineering University, China);

1
Deep learning (DL) based automatic modulation classification (AMC) has gained popularity for next-generation wireless communication systems. However, these DL-based AMC models are vulnerable to adversarial examples, which can cause false predictions with high confidence, leading to unreliable and non-robust communication networks. In this paper, we propose a data mapping-based adversarial defense scheme to address this issue. This scheme uses random split, time-domain flips, and phase rotations as three methods of data mapping on the input examples, effectively mitigating the impact of adversarial perturbations on the model's output and ensuring reliable model inference. Evaluation results on the RML2016.10a dataset demonstrate that the proposed defense scheme can effectively resist various white-box attacks and improve the robustness of the AMC model without requiring fine-tuning or incremental training. This scheme therefore offers a secure solution for intelligent communication networks.
Speaker Yan Zheng (Harbin Engineering University, China)


Homomorphic Filtering Adversarial Defense for Automatic Modulation Classification

Jiarun Yu (Harbin Engineering University, China); Zhongqiu He (Harbin Engineering University, China); Sicheng Zhang (Harbin Engineering University, China); Yu Han (Harbin Engineering University, China); Hang Jiang (Harbin Engineering University, China);

1
Deep neural networks provide an intelligent means for automatic modulation classification (AMC) in the communication field. However, due to their interpretability flaws, neural networks are vulnerable to adversarial examples that lead to decision anomalies. In this paper, we propose a homomorphic filtering adversarial defense (HFAD) algorithm for filtering in the signal frequency domain to defend against adversarial examples and promote the safe and reliable application of AMC models. The algorithm attenuates the low frequency components of the signal by performing homomorphic filtering on it, effectively alleviates the error induction of the model output by the adversarial perturbation. Our experimental results demonstrate that the defense algorithm we propose can not only ensure a high recognition accuracy for the original signal, but also effectively resist a variety of white-box adversarial attacks and improve the robustness of the AMC model against adversarial examples.
Speaker Jiarun Yu (Harbin Engineering University)


Automatic Modulation Classification Based on Decentralized Learning and Model Averaging

Yunhe Xu (Nanjing University of Posts and Telecommunications, China); Min MA (Nanjing Vocational College of Information Technology, China); Xue Fu (Nanjing University of Posts and Telecommunications, China); Yu Wang (Nanjing University of Posts and Telecommunications, China); Guan Gui (Nanjing University of Posts and Telecommunications, China); Tomoaki Ohtsuki (Keio University, Japan)

0
Automatic modulation classification (AMC) is a promising technology to identify the modulation mode of the received signal in non-cooperative communication scenarios. Recently, benefitting from the outstanding classification performance of deep learning (DL), various deep neural networks (DNNs) have been introduced into AMC methods. Most current AMC methods are based on local framework (LocalAMC) where there is only one device or centralized framework (CentAMC) where multiple local devices (LDs) upload their data to the only one central server (CS). LocalAMC may not get ideal results due to insufficient data and finite computational power. CentAMC carries a significant risk of privacy leakage and the final data for training model in CS is quite massive. In this paper, we propose an AMC method based on decentralized learning with residual network (ResNet). Simulation results show that ResNet-based decentralized AMC (DecentAMC) method achieves similar classification performance to CentAMC while improving training efficiency and protecting data privacy.
Speaker Yunhe Xu (Nanjing University of Posts and Telecommunications)


Radio Frequency Signal Dataset Generation Based on LTE System and Variable Channels

Shupeng Zhang (Nanjing University of Posts and Telecommunications, China); Yibin Zhang (Nanjing University of Posts and Telecommunications, China); Xi-xi Zhang (Nanjing University of Posts and Telecommunications, China); Yang Peng (Nanjing University of Posts and Telecommunications, China); Jinlong Sun (Nanjing University of Posts and Telecommunications, China); Guan Gui (Nanjing University of Posts and Telecommunications, China); Tomoaki Ohtsuki (Keio University, Japan)

0
Deep learning-based radio frequency fingerprinting (RFF) identification has the potential to enhance the security performance of the physical layer. In recent years, a number of RFF datasets have been proposed to meet the large-scale data requirements for deep learning. However, these datasets are collected from similar channel environments and only contain receiver data. This paper employs different software radio peripherals to generate radio signals. Hence, it is able to adjust the signal's parameters, such as frequency band, modulation style, antenna gain, etc. In this paper, we propose a radio frequency signal dataset based on LTE system and variable channels to more properly characterize the generated signals in the real world. We collect signals at transmitters and receivers to construct the RFF dataset. Moreover, we confirm the dataset's dependability using various machine learning and deep learning methods. The dataset and reproducible code of this paper can be downloaded from GitHub.
Speaker Shupeng Zhang (Nanjing University of Posts and Telecommunications)


Session Chair

Lei Chen (Georgia Southern University, United States); Wenjia Li (New York Institute of Technology, United States); Yun Lin (Harbin Engineering University, P.R. China)


Gold Sponsor


Gold Sponsor


Bronze Sponsor


Student Travel Grants


Student Travel Grants


Local Organizer

Made with in Toronto · Privacy Policy · INFOCOM 2020 · INFOCOM 2021 · INFOCOM 2022 · © 2023 Duetone Corp.