Xing Gao
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
Xing Gao has a total of 31 co-authors. Here are the top co-authors by number of shared publications:
- Cong Wang
Zhejiang University
3 shared publications - Zhenkai Zhang
Clemson University, Clemson, SC, USA
3 shared publications - Xiaofan Li
The University of Delaware, Newark, DE, USA
2 shared publications - Yanru Xiao
Old Dominion University
2 shared publications - Chu Qiao
The University of Delaware, Newark, DE, USA
1 shared publication - Chu Qiao
University of Delaware Zhejiang University Clemson University University of Texas at Arlington University of Delaware
1 shared publication
Latest Publications
- 24 April 2026
- Published by Springer Nature
- p. 541-564
When VR is Not Just for Gaming: Unmasking GPU Resource Misuse on VR Headset
- 24 April 2026
- Published by Springer Nature
- p. 541-564
When VR is Not Just for Gaming: Unmasking GPU Resource Misuse on VR Headset
- Open Access
Investigating Security Implications of Automatically Generated Code on the Software Supply Chain
In recent years, various software supply chain (SSC) attacks have posed significant risks to the global community. Severe consequences may arise if developers integrate insecure code snippets that are vulnerable to SSC attacks into their products. Particularly, code generation techniques, such as large language models (LLMs), have been widely utilized in the developer community. However, LLMs are known to suffer from inherent issues when generating code, including fabrication, misinformation, and reliance on outdated training data, all of which can result in serious software supply chain threats. In this paper, we investigate the security threats to the SSC that arise from these inherent issues. We examine three categories of threats, including eleven potential SSC-related threats, related to external components in source code, and continuous integration configuration files. We find some threats in LLM-generated code could enable attackers to hijack software and workflows, while some others might cause potential hidden threats that compromise the security of the software over time. To understand these security impacts and severity, we design a tool, SSCGuard, to generate 439,138 prompts based on SSC-related questions collected online, and analyze the responses of four popular LLMs from GPT and Llama. Our results show that all identified SSC-related threats persistently exist. To mitigate these risks, we propose a novel prompt-based defense mechanism, namely Chain-of-Confirmation, to reduce fabrication, and a middleware-based defense that informs users of various SSC threats.
Open AccessInvestigating Security Implications of Automatically Generated Code on the Software Supply Chain
In recent years, various software supply chain (SSC) attacks have posed significant risks to the global community. Severe consequences may arise if developers integrate insecure code snippets that are vulnerable to SSC attacks into their products. Particularly, code generation techniques, such as large language models (LLMs), have been widely utilized in the developer community. However, LLMs are known to suffer from inherent issues when generating code, including fabrication, misinformation, and reliance on outdated training data, all of which can result in serious software supply chain threats. In this paper, we investigate the security threats to the SSC that arise from these inherent issues. We examine three categories of threats, including eleven potential SSC-related threats, related to external components in source code, and continuous integration configuration files. We find some threats in LLM-generated code could enable attackers to hijack software and workflows, while some others might cause potential hidden threats that compromise the security of the software over time. To understand these security impacts and severity, we design a tool, SSCGuard, to generate 439,138 prompts based on SSC-related questions collected online, and analyze the responses of four popular LLMs from GPT and Llama. Our results show that all identified SSC-related threats persistently exist. To mitigate these risks, we propose a novel prompt-based defense mechanism, namely Chain-of-Confirmation, to reduce fabrication, and a middleware-based defense that informs users of various SSC threats.
- Open Access
- 2 December 2024
- Published by Association for Computing Machinery (ACM)
- p. 482-496
Toward Understanding the Security of Plugins in Continuous Integration Services
Mainstream Continuous Integration (CI) platforms have provided the plugin functionality to accelerate the development of CI pipelines. Unfortunately, CI plugins, which are essentially reusable code snippets, also expose new attack surfaces as plugins might be developed by less trusted users. In this paper, we present an in-depth study to understand potential security risks in existing CI plugins. We conduct a comprehensive analysis of plugin implementations on four mainstream CI platforms (GitHub Actions, GitLab CI, CircleCI, and Azure Pipelines), and investigate several weak links in existing plugin distributions and isolation mechanisms. We investigate seven attack vectors that can enable attackers to hijack plugins and distribute malicious code without plugins users being aware, and further exploit hijacked plugins to manipulate the workflow execution. Additionally, we find that plugin dependency (a plugin references other plugins) might further amplify the attack impact of our disclosed attacks. To evaluate the potential impact, we conduct a large-scale measurement on GitHub and GitLab, covering a total of 1,328,912 repositories using the aforementioned CI platforms. Our measurement results show that a large number of repositories and existing plugins, including many widely used ones, are potentially vulnerable to the proposed attacks. We have duly reported the identified vulnerabilities and received positive responses.
- 2 December 2024
- Published by Association for Computing Machinery (ACM)
- p. 482-496
Open AccessToward Understanding the Security of Plugins in Continuous Integration Services
Mainstream Continuous Integration (CI) platforms have provided the plugin functionality to accelerate the development of CI pipelines. Unfortunately, CI plugins, which are essentially reusable code snippets, also expose new attack surfaces as plugins might be developed by less trusted users. In this paper, we present an in-depth study to understand potential security risks in existing CI plugins. We conduct a comprehensive analysis of plugin implementations on four mainstream CI platforms (GitHub Actions, GitLab CI, CircleCI, and Azure Pipelines), and investigate several weak links in existing plugin distributions and isolation mechanisms. We investigate seven attack vectors that can enable attackers to hijack plugins and distribute malicious code without plugins users being aware, and further exploit hijacked plugins to manipulate the workflow execution. Additionally, we find that plugin dependency (a plugin references other plugins) might further amplify the attack impact of our disclosed attacks. To evaluate the potential impact, we conduct a large-scale measurement on GitHub and GitLab, covering a total of 1,328,912 repositories using the aforementioned CI platforms. Our measurement results show that a large number of repositories and existing plugins, including many widely used ones, are potentially vulnerable to the proposed attacks. We have duly reported the identified vulnerabilities and received positive responses.
- 13 October 2024
- Published by Springer Nature
- p. 145-164
Unsupervised Multi-criteria Adversarial Detection in Deep Image Retrieval
- 13 October 2024
- Published by Springer Nature
- p. 145-164
Unsupervised Multi-criteria Adversarial Detection in Deep Image Retrieval
- 7 August 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Security & Privacy
- Vol. 23 (1) , 64-74
Game Theory in Distributed Systems Security: Foundations, Challenges, and Future Directions
Combining insights from distributed system security and game theory could effectively address security challenges. We present foundational concepts from both fields that can be integrated to better secure distributed systems and outline several research challenges for the community to tackle.
- 7 August 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Security & Privacy
- Vol. 23 (1) , 64-74
Game Theory in Distributed Systems Security: Foundations, Challenges, and Future Directions
Combining insights from distributed system security and game theory could effectively address security challenges. We present foundational concepts from both fields that can be integrated to better secure distributed systems and outline several research challenges for the community to tackle.
- Open Access
Caching Aided Multi-Tenant Serverless Computing
One key to enabling high-performance serverless computing is to mitigate cold-starts. Current solutions utilize a warm pool to keep function alive: a warm-start can be analogous to a CPU cache-hit. However, modern cache has multiple hierarchies and the last-level cache is shared among cores, whereas the warm pool is limited to a single tenant for security concerns. Also, the warm pool keep-alive policy can be further optimized using cache replacement algorithms. In this paper, we borrow practical optimizations from caching, and design FaasCamp, a caching-aided multi-tenant serverless computing framework. FaasCamp extends the single-tier warm pool into multi-tiers, with a reclaim pool introduced enabling secure function instance sharing among tenants. Also, FaasCamp leverages machine learning to approximate the optimal cache replacement policy to improve the warm rate. We have implemented a prototype and conducted extensive experiments under multiple scenarios. The results show that FaasCamp can outperform existing platforms with minimal overhead.
Open AccessCaching Aided Multi-Tenant Serverless Computing
One key to enabling high-performance serverless computing is to mitigate cold-starts. Current solutions utilize a warm pool to keep function alive: a warm-start can be analogous to a CPU cache-hit. However, modern cache has multiple hierarchies and the last-level cache is shared among cores, whereas the warm pool is limited to a single tenant for security concerns. Also, the warm pool keep-alive policy can be further optimized using cache replacement algorithms. In this paper, we borrow practical optimizations from caching, and design FaasCamp, a caching-aided multi-tenant serverless computing framework. FaasCamp extends the single-tier warm pool into multi-tiers, with a reclaim pool introduced enabling secure function instance sharing among tenants. Also, FaasCamp leverages machine learning to approximate the optimal cache replacement policy to improve the warm rate. We have implemented a prototype and conducted extensive experiments under multiple scenarios. The results show that FaasCamp can outperform existing platforms with minimal overhead.
- Open Access
- 13 May 2024
- Published by Association for Computing Machinery (ACM)
- p. 1847-1858
Characterizing Ethereum Upgradable Smart Contracts and Their Security Implications
Upgradeable smart contracts (USCs) have been widely adopted to enable modifying deployed smart contracts. While USCs bring great flexibility to developers, improper usage might introduce new security issues, potentially allowing attackers to hijack USCs and their users. In this paper, we conduct a large-scale measurement study to characterize USCs and their security implications in the wild. We summarize six commonly used USC patterns and develop a tool, USCDetector, to identify USCs without needing source code. Particularly, USCDetector collects various information such as bytecode and transaction information to construct upgrade chains for USCs and disclose potentially vulnerable ones. We evaluate USCDetector using verified smart contracts (i.e., with source code) as ground truth and show that USCDetector can achieve high accuracy with a precision of 96.26%. We then use USCDetector to conduct a large-scale study on Ethereum, covering a total of 60,251,064 smart contracts. USCDetecor constructs 10,218 upgrade chains and discloses multiple real-world USCs with potential security issues.
- 13 May 2024
- Published by Association for Computing Machinery (ACM)
- p. 1847-1858
Open AccessCharacterizing Ethereum Upgradable Smart Contracts and Their Security Implications
Upgradeable smart contracts (USCs) have been widely adopted to enable modifying deployed smart contracts. While USCs bring great flexibility to developers, improper usage might introduce new security issues, potentially allowing attackers to hijack USCs and their users. In this paper, we conduct a large-scale measurement study to characterize USCs and their security implications in the wild. We summarize six commonly used USC patterns and develop a tool, USCDetector, to identify USCs without needing source code. Particularly, USCDetector collects various information such as bytecode and transaction information to construct upgrade chains for USCs and disclose potentially vulnerable ones. We evaluate USCDetector using verified smart contracts (i.e., with source code) as ground truth and show that USCDetector can achieve high accuracy with a precision of 96.26%. We then use USCDetector to conduct a large-scale study on Ethereum, covering a total of 60,251,064 smart contracts. USCDetecor constructs 10,218 upgrade chains and discloses multiple real-world USCs with potential security issues.
- Open Access
- 15 November 2023
- Published by Association for Computing Machinery (ACM)
- p. 960-974
T unne L s for B ootlegging: Fully Reverse-Engineering GPU TLBs for Challenging Isolation Guarantees of NVIDIA MIG
Recent studies have revealed much detailed information about the translation lookaside buffers (TLBs) of modern CPUs, but we find that many properties of such components in modern GPUs still remain unknown or unclear. To fill this knowledge gap, we develop a new GPU TLB reverse-engineering method and apply it to a variety of consumer- and server-grade GPUs in Turing and Ampere generations. Aside from learning significantly more comprehensive and accurate GPU TLB properties, we discover a design flaw of NVIDIA Multi-Instance GPU (MIG) feature. MIG claims full partitioning of the entire GPU memory system for secure GPU sharing in cloud computing. However, we surprisingly find that MIG does not partition the last-level TLB, which is shared by all the compute units in a GPU. Exploiting this design flaw and learned TLB properties, we are able to construct a covert channel for data exfiltration across MIG-enforced isolation. To the best of our knowledge, this is the first attack on MIG. We evaluate the proposed attack on a commercial cloud platform, and we successfully achieve reliable data exfiltration from a victim tenant at a speed of up to 31 kbps with a very high accuracy around 99.8%. Even when the victim is using the GPU for deep neural network training, the transmission can still reach more than 25 kbps with a more than 99.5% accuracy. We propose and implement a mitigation approach that can effectively thwart data exfiltration through this covert channel. Additionally, we present a preliminary study on exploiting the access patterns of the last-level TLB to infer the identity of applications running in other MIG-created GPU instances.
- 15 November 2023
- Published by Association for Computing Machinery (ACM)
- p. 960-974
Open AccessT unne L s for B ootlegging: Fully Reverse-Engineering GPU TLBs for Challenging Isolation Guarantees of NVIDIA MIG
Recent studies have revealed much detailed information about the translation lookaside buffers (TLBs) of modern CPUs, but we find that many properties of such components in modern GPUs still remain unknown or unclear. To fill this knowledge gap, we develop a new GPU TLB reverse-engineering method and apply it to a variety of consumer- and server-grade GPUs in Turing and Ampere generations. Aside from learning significantly more comprehensive and accurate GPU TLB properties, we discover a design flaw of NVIDIA Multi-Instance GPU (MIG) feature. MIG claims full partitioning of the entire GPU memory system for secure GPU sharing in cloud computing. However, we surprisingly find that MIG does not partition the last-level TLB, which is shared by all the compute units in a GPU. Exploiting this design flaw and learned TLB properties, we are able to construct a covert channel for data exfiltration across MIG-enforced isolation. To the best of our knowledge, this is the first attack on MIG. We evaluate the proposed attack on a commercial cloud platform, and we successfully achieve reliable data exfiltration from a victim tenant at a speed of up to 31 kbps with a very high accuracy around 99.8%. Even when the victim is using the GPU for deep neural network training, the transmission can still reach more than 25 kbps with a more than 99.5% accuracy. We propose and implement a mitigation approach that can effectively thwart data exfiltration through this covert channel. Additionally, we present a preliminary study on exploiting the access patterns of the last-level TLB to infer the identity of applications running in other MIG-created GPU instances.
- Open Access
Unsupervised Multi-Criteria Adversarial Detection in Deep Image Retrieval
The vulnerability in the algorithm supply chain of deep learning has imposed new challenges to image retrieval systems in the downstream. Among a variety of techniques, deep hashing is gaining popularity. As it inherits the algorithmic backend from deep learning, a handful of attacks are recently proposed to disrupt normal image retrieval. Unfortunately, the defense strategies in softmax classification are not readily available to be applied in the image retrieval domain. In this paper, we propose an efficient and unsupervised scheme to identify unique adversarial behaviors in the hamming space. In particular, we design three criteria from the perspectives of hamming distance, quantization loss and denoising to defend against both untargeted and targeted attacks, which collectively limit the adversarial space. The extensive experiments on four datasets demonstrate 2-23% improvements of detection rates with minimum computational overhead for real-time image queries.
Open AccessUnsupervised Multi-Criteria Adversarial Detection in Deep Image Retrieval
The vulnerability in the algorithm supply chain of deep learning has imposed new challenges to image retrieval systems in the downstream. Among a variety of techniques, deep hashing is gaining popularity. As it inherits the algorithmic backend from deep learning, a handful of attacks are recently proposed to disrupt normal image retrieval. Unfortunately, the defense strategies in softmax classification are not readily available to be applied in the image retrieval domain. In this paper, we propose an efficient and unsupervised scheme to identify unique adversarial behaviors in the hamming space. In particular, we design three criteria from the perspectives of hamming distance, quantization loss and denoising to defend against both untargeted and targeted attacks, which collectively limit the adversarial space. The extensive experiments on four datasets demonstrate 2-23% improvements of detection rates with minimum computational overhead for real-time image queries.
- Open Access
A Framework for Behavioral Biometric Authentication using Deep Metric Learning on Mobile Devices
Mobile authentication using behavioral biometrics has been an active area of research. Existing research relies on building machine learning classifiers to recognize an individual's unique patterns. However, these classifiers are not powerful enough to learn the discriminative features. When implemented on the mobile devices, they face new challenges from the behavioral dynamics, data privacy and side-channel leaks. To address these challenges, we present a new framework to incorporate training on battery-powered mobile devices, so private data never leaves the device and training can be flexibly scheduled to adapt the behavioral patterns at runtime. We re-formulate the classification problem into deep metric learning to improve the discriminative power and design an effective countermeasure to thwart side-channel leaks by embedding a noise signature in the sensing signals without sacrificing too much usability. The experiments demonstrate authentication accuracy over 95% on three public datasets, a sheer 15% gain from multi-class classification with less data and robustness against brute-force and side-channel attacks with 99% and 90% success, respectively. We show the feasibility of training with mobile CPUs, where training 100 epochs takes less than 10 mins and can be boosted 3-5 times with feature transfer. Finally, we profile memory, energy and computational overhead. Our results indicate that training consumes lower energy than watching videos and slightly higher energy than playing games.
Open AccessA Framework for Behavioral Biometric Authentication using Deep Metric Learning on Mobile Devices
Mobile authentication using behavioral biometrics has been an active area of research. Existing research relies on building machine learning classifiers to recognize an individual's unique patterns. However, these classifiers are not powerful enough to learn the discriminative features. When implemented on the mobile devices, they face new challenges from the behavioral dynamics, data privacy and side-channel leaks. To address these challenges, we present a new framework to incorporate training on battery-powered mobile devices, so private data never leaves the device and training can be flexibly scheduled to adapt the behavioral patterns at runtime. We re-formulate the classification problem into deep metric learning to improve the discriminative power and design an effective countermeasure to thwart side-channel leaks by embedding a noise signature in the sensing signals without sacrificing too much usability. The experiments demonstrate authentication accuracy over 95% on three public datasets, a sheer 15% gain from multi-class classification with less data and robustness against brute-force and side-channel attacks with 99% and 90% success, respectively. We show the feasibility of training with mobile CPUs, where training 100 epochs takes less than 10 mins and can be boosted 3-5 times with feature transfer. Finally, we profile memory, energy and computational overhead. Our results indicate that training consumes lower energy than watching videos and slightly higher energy than playing games.