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.

Federated learning (FL) relies on a central authority to oversee and aggregate model updates contributed by multiple participating parties in the training process. This centralization of sensitive model updates naturally raises concerns about the trustworthiness of the central aggregation server, as well as the potential risks associated with server failures or breaches, which could result in loss and leaks of model updates. Moreover, recent attacks have demonstrated that, by obtaining the leaked model updates, malicious actors can even reconstruct substantial amounts of private data belonging to training participants. This underscores the critical necessity to rethink the existing FL system architecture to mitigate emerging attacks in the evolving threat landscape. One straightforward approach is to fortify the central aggregator with confidential computing (CC), which offers hardware-assisted protection for runtime computation and can be remotely verified for execution integrity. However, a growing number of security vulnerabilities have surfaced in tandem with the adoption of CC, indicating that depending solely on this singular defense may not provide the requisite resilience to thwart data leaks.

This paper presents Protected Execution Facility (PEF), a virtual machine-based Trusted Execution Environment (TEE) for confidential computing on Power ISA. PEF enables protected secure virtual machines (SVMs). Like other TEEs, PEF verifies the SVM prior to execution. PEF utilizes a Trusted Platform Module (TPM), secure boot, and trusted boot as well as newly introduced architectural changes for Power ISA systems. Exploiting these architectural changes requires new firmware, the Protected Execution Ultravisor. PEF is supported in the latest version of the POWER9 chip. PEF demonstrates that access control for isolation and cryptography for confidentiality is an effective approach to confidential computing. We particularly focus on how our design (i) balances between access control and cryptography, (ii) maximizes the use of existing security components, and (iii) simplifies the management of the SVM life cycle. Finally, we evaluate the performance of SVMs in comparison to normal virtual machines on OpenPOWER systems.

The ability to count and fingerprint devices (optionally, from a given type) that are in a network, and potentially behind a NAT is important not only for network management (e.g., inventory and asset management) but also for business analysis (e.g., product adoption) and security (e.g., to block traffic from a malicious device behind a NAT). As such, researchers have developed a number of solutions to address these questions. However, most existing solutions rely on incidental characteristics of end devices' behaviors. Software updates to end devices or middleboxes (e.g., firewall, NAT) could render existing solutions ineffective. As such, how effective are those solutions in today's settings, e.g., with IoT devices? We propose to answer this answer by evaluating three major approaches that rely on (1) the IP id field, (2) a device's clock skew, and (3) a combination of a device's boot time and the frequency of its TCP timestamp clock, on the network traffic of seventy IoT devices. We show that existing approaches are ineffective with recent IoT devices, and as such the problem of counting devices behind a NAT remains an open problem. Finally, we explore and discuss future potential directions.

In this work we present Scalable Attestation, a method which combines both secure boot and trusted boot technologies, and extends them up into the host, its programs, and up into the guest's operating system and workloads, to both detect and prevent integrity attacks. Anchored in hardware, this integrity appraisal and attestation protects persistent data (files) from remote attack, even if the attack is root privileged. As an added benefit of a hardware rooted attestation, we gain a simple hardware based geolocation attestation to help enforce regulatory requirements. This design is implemented in multiple cloud test beds based on the QEMU/KVM hypervisor, Open Stack, and Open Attestation, and is shown to provide significant additional integrity protection at negligible cost.

Server virtualization more readily enables the collocation of disparate workloads on a shared physical platform. When employed on systems across a data center, the result can be a dramatic increase in server utilization and a decrease in overall power, cooling and floor space requirements. However, in an environment where workloads share the underlying platforms, achieving other desirable workload goals, such as availability and security, becomes a challenge. In particular, enforcing isolation between workloads in a large, dynamic, and virtualized data center requires strong yet easily configurable controls on the sharing of resources at the virtualization layer. Commercial hypervisors usually offer reasonable isolation of individual virtual machines (VMs). However, on hypervisor-based platforms, one cannot currently define a single policy that automatically enforces restrictions on the sharing of resources between multiple VMs or request an air gap between workloads. In this paper, we describe the design and implementation of a Hypervisor-based Mandatory Access Control (MAC) that achieves policy-driven distributed workload isolation for the IBM Power Hypervisor (PHYP). We discuss our experiences and lessons learned and examine the implications and trade-offs involved in providing MAC on a production- level, commercially-available hypervisor. Our goal is to simplify the security management of data centers through centralized security management and policy- driven distributed access control and data protection.

We present the sHype hypervisor security architecture and examine in detail its mandatory access control facilities. While existing hypervisor security approaches aiming at high assurance have been proven useful for high-security environments that prioritize security over performance and code reuse, our approach aims at commercial security where near-zero performance overhead, non-intrusive implementation, and usability are of paramount importance. sHype enforces strong isolation at the granularity of a virtual machine, thus providing a robust foundation on which higher software layers can enact finer-grained controls. We provide the rationale behind the sHype design and describe and evaluate our implementation for the Xen open-source hypervisor