Shengjie Xu

Shengjie Xu

Auto-generated profile, based on publicly available information

Shengjie Xu

College of Information Science, The University of Arizona, Tucson, AZ, USA

Research Focus & Metrics

Academic Career Length

8Years (2019-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
8 October 2020 in IEEE Communications Surveys & Tutorials
Verified
Journal Article
Closed access
SGSohan Gyawali
Sohan Gyawali
SXShengjie Xu
Shengjie Xu
YQYi Qian
Yi Qian
RHRose Hu
Rose Hu
A wide variety of works have been conducted in vehicle-to-everything (V2X) communications to enable a variety of applications for road safety, traffic efficiency and passenger infotainment. Although dedicated short-range communications (DSRC) based V2X is already in the deployment phase, cellular based V2X is gaining more interest in academia and industry most recently. This paper surveys the existing work and challenges on LTE and 5G to support efficient V2X communications. First, we present the motivations for cellular based V2X communications. Second, we summarize the LTE V2X architecture and operating scenarios being considered. Third, we discuss the challenges in existing LTE for supporting V2X communications such as physical layer structure, synchronization, multimedia broadcast multicast services (MBMS), resource allocation, security and survey the recent solutions to these challenges. We further discuss the challenges and possible solutions for 5G based vehicular communications. Finally, we discuss the open research issues and possible research directions in cellular based vehicular communications.

20 August 2019 in IEEE Transactions on Network Science and Engineering
Verified
Journal Article
Closed access
SXShengjie Xu
Shengjie Xu
YQYi Qian
Yi Qian
RHRose Hu
Rose Hu
The advancement of networking platforms for assured online services requires robust and effective network intelligence systems against anomalous events and malicious threats. With the rapid development of modern communication technologies, artificial intelligence, and the revolution of computing devices, cloud computing empowered network intelligence will inevitably become a core platform for various smart applications. While cloud computing provides strong and powerful computation, storage, and networking services to detect and defend cyber threats, edge computing on the other hand will deliver more benefits in specific yet potential critical areas. In this paper, we present a study on the data-driven edge intelligence for robust network anomaly detection. We first highlight the main motivations for edge intelligence, and then propose an intelligence system empowered by edge computing for network anomaly detection. We further propose a scheme on the data-driven robust network anomaly detection. In the proposed scheme, four phases are designed to incorporate with data-driven approaches to train a learning model which is able to detect and identify a network anomaly in a robust way. In the performance evaluations with data experiments, we demonstrate that the proposed scheme achieves the robustness of trained model and the efficiency on the detection of specific anomalies.

27 June 2025 in IEEE Access
Verified
Journal Article
Open access
BCBryce Cordill
Bryce Cordill
DFDongfeng Fang
Dongfeng Fang
SXShengjie Xu
Shengjie Xu
The proliferation of unmanned aerial vehicles (UAVs) across diverse applications, from photography to critical infrastructure inspection, has raised significant security and privacy challenges. This comprehensive survey presents a systematic analysis of UAV security vulnerabilities, privacy concerns, and practical solutions, including concerning implementation feasibility and regulatory compliance. We begin with a detailed UAV system model encompassing hardware, software, and communication protocols, followed by a novel categorization of security vulnerabilities that clearly distinguishes between hardware attack, software attack, and communication technology attack surfaces. Our analysis of privacy issues extends beyond traditional concerns, including ethical considerations in UAV surveillance and compliance with global privacy frameworks such as GDPR. We present state-of-the-art technical solutions, including lightweight encryption protocols, privacy-preserving machine learning, and blockchain-based security mechanisms, with specific implementation metrics and resource utilization guidelines. The paper also examines policy-based approaches and regulatory frameworks across different jurisdictions, providing a balanced perspective on both technical and governance aspects. Our work concludes with an analysis of emerging technologies and future research directions, emphasizing the need for integrated solutions that balance security requirements with operational constraints. This survey serves as a comprehensive reference for researchers, practitioners, and policymakersworking to enhanceUAVsecurity and privacy while ensuring practical deployability.

10 May 2021
Verified
Conference Paper
Closed access
MBMatthew Bradley
Matthew Bradley
SXShengjie Xu
Shengjie Xu
Recent studies in the field of Adversarial Machine Learning (AML) have primarily focused on techniques for poisoning and manipulating the Machine Learning (ML) systems for operations such as malware identification and image recognition. While the offensive perspective of such systems is increasingly well documented, the work approaching the problem from the defensive standpoint is sparse. In this paper, we define a metric for quantizing the vulnerability or susceptibility of a given ML model to adversarial manipulation using only properties inherent to the model under examination. This metric will be shown to have several useful properties related to known features of classifier-based ML systems and is intended as a tool to broadly compare the security of various competing ML models based on their maximum potential susceptibility to adversarial manipulation.

30 April 2025 in IEEE Access
Verified
Journal Article
Open access
GMGary A. McCully
Gary A. McCully
JHJohn Hastings
John Hastings
SXShengjie Xu
Shengjie Xu
Ransomware and other malware inflict devastating financial and operational damage on organizations worldwide by exploiting deeply embedded, hard-to-detect vulnerabilities in their systems. Detecting these vulnerabilities in compiled code before malicious actors exploit them remains a critical challenge in cybersecurity. This research introduces TEDVIL (Transformer-based Embeddings for Discovering Vulnerabilities in Lifted Code), a novel framework which uses transformer-based embeddings to train neural networks to detect vulnerabilities in lifted code. The framework was implemented using bidirectional (BERT and RoBERTa) and unidirectional (GPT-1 and GPT-2) transformer-based models to generate embeddings for training Long Short-Term Memory (LSTM) neural networks to detect stack-based buffer overflows in Low-Level Virtual Machine (LLVM) intermediate representation code. For comparison, simpler word2vec models (Skip-Gram and Continuous Bag of Words) were also trained, and their embeddings were used to train LSTMs. The results show that the LSTMs using GPT-2 embeddings outperformed those using GPT-1, BERT, RoBERTa, and word2vec embeddings, achieving a top accuracy of 92.5% and an F1-score of 89.7%. Notably, these results are achieved when the embedding model is trained with a dataset of just 48,000 functions, demonstrating effectiveness in resource-constrained settings. The findings underscore the effectiveness of TEDVIL in identifying hard-to-detect vulnerabilities in compiled code, and lay the groundwork for future research in leveraging transformer-based models for vulnerability detection.

1 January 2026 in IEEE Access
Verified
Journal Article
Open access
RARobel Addis
Robel Addis
KPKiran Porter
Kiran Porter
JRJayden Ryan
Jayden Ryan
MSMicah Stull
Micah Stull
SXShengjie Xu
Shengjie Xu
SFSteven Forsyth
Steven Forsyth