Confidential collaborative machine learning (ML) enables multiple mutually distrusted data holders to jointly train an ML model while preserving the confidentiality of their private datasets due to regulatory or competitive reasons. However, existing works need frequent data and model exchanges during training via slower conventional links. They face increasing challenges due to the exponentially growing sizes of models and datasets in modern training workloads like large language models (LLMs), resulting in prohibitively high communication costs. In this paper, we propose a novel mechanism called GPU Travelling that leverages recently emerged confidential GPUs. With our rigorous design, the GPU can securely travel to the specific data holder to load the dataset directly into the GPU's protected memory and then return for training, eliminating the need for data transmission while ensuring confidentiality up to a data-centre level. We developed a prototype using Intel TDX and NVIDIA H100 and evaluated its performance on llm.c, a CUDA-based LLM training project, and demonstrated the performance and feasibility while maintaining strong security guarantees. The results showed at least 4x speed improvement when transmitting a 512 MiB dataset chunk versus conventional transmission.

The Confidential Containers (CoCo) project, as an open-source community initiative, inherits the system architecture of Kata Containers while integrating confidential computing to protect cloud-native container workloads. However, there exists a misalignment in the threat model and trusted computing base (TCB) between Kata Containers and confidential computing. The shifted trust boundaries could potentially expose a range of vulnerabilities, particularly in scenarios where a malicious actor on the host gains access to the CoCo's unprotected control interface. This paper conducts a thorough examination of CoCo's system architecture, exploring the attack surface resulting from the discord in trust boundaries. We have assessed all API endpoints of CoCo's control interface, categorizing them based on their security properties. Drawing from these insights, we have developed a bifurcation approach to splitting CoCo's control interface. This involves establishing an owner-side controller and minimizing the capabilities of the existing host-side controller. Under this framework, the host-side controller is exclusively responsible for allocating and recycling compute resources, while dedicated workload owners can directly manage their containers through alternative secure tunnels. This approach ensures seamless integration with cloud-native orchestration layers and aligns CoCo with the threat model of confidential computing. By doing so, it effectively prevents untrusted hosts from accessing confidential data and interfering with the execution of workloads within protected domains.

In this article, we conduct a measurement study to comprehensively compare the accuracy impacts of multiple embedding options in cryptographic API completion tasks. Embedding is the process of automatically learning vector representations of program elements. Our measurement focuses on design choices of three important aspects, program analysis preprocessing , token-level embedding , and sequence-level embedding . Our findings show that program analysis is necessary even under advanced embedding. The results show 36.20% accuracy improvement, on average, when program analysis preprocessing is applied to transfer bytecode sequences into API dependence paths. With program analysis and the token-level embedding training, the embedding dep2vec improves the task accuracy from 55.80% to 92.04%. Moreover, only a slight accuracy advantage (0.55%, on average) is observed by training the expensive sequence-level embedding compared with the token-level embedding. Our experiments also suggest the differences made by the data. In the cross-app learning setup and a data scarcity scenario, sequence-level embedding is more necessary and results in a more obvious accuracy improvement (5.10%).

Container-based clouds—in which containers are the basic unit of isolation—face security concerns because, unlike Virtual Machines, containers directly interface with the underlying highly privileged kernel through the wide and vulnerable system call interface. Regardless of whether a container itself requires dangerous system calls, a compromised or malicious container sharing the host (a bad neighbor) can compromise the host kernel using a vulnerable syscall, thereby compromising all other containers sharing the host.

Google and Apple jointly introduced a digital contact tracing technology and an API called "exposure notification,'' to help health organizations and governments with contact tracing. The technology and its interplay with security and privacy constraints require investigation. In this study, we examine and analyze the security, privacy, and reliability of the technology with actual and typical scenarios (and expected typical adversary in mind), and quite realistic use cases. We do it in the context of Virginia's COVIDWISE app. This experimental analysis validates the properties of the system under the above conditions, a result that seems crucial for the peace of mind of the exposure notification technology adopting authorities, and may also help with the system's transparency and overall user trust.

Data-oriented attacks manipulate non-control data to alter a program’s benign behavior without violating its control-flow integrity. It has been shown that such attacks can cause significant damage even in the presence of control-flow defense mechanisms. However, these threats have not been adequately addressed. In this survey article, we first map data-oriented exploits, including Data-Oriented Programming (DOP) and Block-Oriented Programming (BOP) attacks, to their assumptions/requirements and attack capabilities. Then, we compare known defenses against these attacks, in terms of approach, detection capabilities, overhead, and compatibility. It is generally believed that control flows may not be useful for data-oriented security. However, data-oriented attacks (especially DOP attacks) may generate side effects on control-flow behaviors in multiple dimensions (i.e., incompatible branch behaviors and frequency anomalies). We also characterize control-flow anomalies caused by data-oriented attacks. In the end, we discuss challenges for building deployable data-oriented defenses and open research questions.

Just-in-time return-oriented programming (JIT-ROP) allows one to dynamically discover instruction pages and launch code reuse attacks, effectively bypassing most fine-grained address space layout randomization (ASLR) protection. However, in-depth questions regarding the impact of code (re-)randomization on code reuse attacks have not been studied. For example, how would one compute the re-randomization interval effectively by considering the speed of gadget convergence to defeat JIT-ROP attacks? ; how do starting pointers in JIT-ROP impact gadget availability and gadget convergence time? ; what impact do fine-grained code randomizations have on the Turing-complete expressive power of JIT-ROP payloads? We conduct a comprehensive measurement study on the effectiveness of fine-grained code randomization schemes, with 5 tools, 20 applications including 6 browsers, 1 browser engine, and 25 dynamic libraries. We provide methodologies to measure JIT-ROP gadget availability, quality, and their Turing-complete expressiveness, as well as to empirically determine the upper bound of re-randomization intervals in re-randomization schemes using the Turing-complete (TC), priority, MOV TC, and payload gadget sets. Experiments show that the upper bound ranges from 1.5 to 3.5 seconds in our tested applications. Besides, our results show that locations of leaked pointers used in JIT-ROP attacks have no impacts on gadget availability but have an impact on how fast attackers find gadgets. Our results also show that instruction-level single-round randomization thwarts current gadget finding techniques under the JIT-ROP threat model.

Higher yield, climbing transistor counts and shrinking dimensions of a single Integrated Circuit (IC) have always been the demands in the fabrication market. However, this increased complexity and miniaturization of transistors present a challenge, due to high critical process variations combined with a ubiquitous presence of temperature and supply voltage variations, to achieve the required specification bounds on the desired performance of the circuits. Since optimization has become a very crucial task in IC design, the paper presents an efficient transistor sizing based optimization technique of the CMOS circuits to achieve low power, high performance and high yield design goals. The proposed memetic algorithm judiciously utilizes a threshold based local search procedure to improve convergence in its inherent genetic nature. The algorithm optimizes with the effect of temperature [-55 to 125]°C and supply voltage ±10% variations and in addition a number of statistically sampled sets generated as Gaussian, Latin Hypercube and Correlation Screened schemes of process variations. The proposed technique is applicable to any technology node and has been tested over several standard single-stage and some complex multi-stage digital circuits designed using a Multi-Gate high-K dielectric (MGK) 22nm CMOS model. The reduction in leakage power with propagation delay goes as high as 53% with 9% respectively, as observed across the various digital circuits.