Modern cloud computing has witnessed a growing trend of leveraging hardware accelerators, such as FPGAs, to boost the performance of computation-intensive workloads. Further, FPGA virtualization has been adopted to improve the efficiency and utilization of FPGA resources in the cloud. However, the security of user data and computation becomes a major concern of the virtualized FPGA cloud services, especially considering the state-of-the-art hardware-based system security technique, namely trusted execution environment (TEE)-enabled confidential computing, does not support FPGAs in the cloud. In this paper, we develop FPGA-CC, an end-to-end confidential container framework designed for virtualized cloud FPGAs. FPGA-CC establishes a secure path between the virtualized FPGA resources and the TEE-based CPU container to accomplish the security objectives concerning virtual FPGA instances. Our experiments on real hardware and various benchmark applications demonstrate that FPGA-CC achieves high security with acceptable performance overhead.

Trusted execution environments (TEEs) can provide hardware and system-level protection for sensitive data and computations. However, the security perimeter of existing TEEs is limited to a single centralized machine, which contradicts with the growing trend of employing disaggregated computing resources (e.g., disaggregated memory) to achieve high performance and resource utilization. To address this limitation, we develop DM-TEE, a customized trusted execution environment supporting the emerging disaggregated memory architecture. DM-TEE extends the traditional TEEs from local memory to remote disaggregated memory, which is achieved by a newly designed secure memory allocation and access workflow to ensure the data confidentiality and integrity in the disaggregated memory. We implement DM-TEE on real hardware using Intel SGX and a state-of-the-art memory disaggregation system. Our evaluations on memory allocation, read/write operations, and benchmark program executions indicate that DM-TEE achieves the desired disaggregated memory security with minimal performance overhead.

Trusted execution environments (TEEs), such as Intel SGX, have become a popular security primitive with minimum trusted computing base (TCB) and attack surface. However, the existing CPU-based TEEs do not support FPGAs, even though FPGA-based cloud computing services have been rapidly deployed with security vulnerabilities that are expected to be eliminated by TEEs. To fill the gap, we present SGX-FPGA, a trusted hardware isolation path enabling the first FPGA TEE by bridging SGX enclaves and FPGAs in the heterogeneous CPU-FPGA architecture. Our experiments on real CPU-FPGA hardware justify the high security and low performance overhead achieved by SGX-FPGA.

Deep neural networks (DNNs) have been deployed on FPGAs to achieve improved performance, power efficiency, and design flexibility. However, the FPGA-based DNNs are vulnerable to fault injection attacks that aim to compromise the original functionality. The existing defense methods either duplicate the models and check the consistency of the results at runtime, or strengthen the robustness of the models by adding additional neurons. However, these existing methods could introduce huge overhead or require retraining the models. In this paper, we develop a runtime verification method, namely Siamese path verification (SPV), to detect fault injection attacks for FPGA-based DNN execution. By leveraging the computing features of the DNN and designing the weight parameters, SPV adds neurons to check the integrity of the model without impacting the original functionality and, therefore, model retraining is not required. We evaluate the proposed SPV approach on Xilinx Virtex-7 FPGA using the MNIST dataset. The evaluation results show that SPV achieves the security goal with low overhead.

In recent years, there has been an explosive increase in the amount of existing visual data. Hashing techniques have been successfully applied to deal with the large-scale nearest neighbor search problem among data on this massive scale. However, existing hashing methods usually learn a single hash code for each data point, and only by taking the content correlations among them into account. In practice, however, when handling complex visual data such as video, strong temporal relations exist among the successive frames. Moreover, if the preferred performance for large-scale video search is to be delivered, multiple hash codes are required for each data point in order to build multiple hash table indices. To address these problems, in this paper, we first study the multi-table learning problem for video search and attempt to learn binary codes by capturing the intrinsic video similarities from both the visual and the temporal aspects. By regarding the search over multiple tables as an ensemble prediction, the whole multi-table learning problem can be solved in a boosting learning manner to complementarily cover the nearest neighbors. For each table, a temporal binary coding solution is devised that thinks over the intrinsic relations among the visual content and the temporal consistency among the successive frames simultaneously. More specifically, we approximate the intrinsic visual similarities using a low-rank matrix based on sparse, non-negative feature expression. Furthermore, to essentially preserve the temporal consistency, we introduce a subspace rotation to model the variation among the successive frames. Under the boosting learning framework, the binary codes, hash functions and temporal variation of each table can be efficiently and jointly optimized. Extensive experiments on three large video datasets demonstrate that the proposed approach significantly outperforms a number of state-of-the-art hashing methods.

Subspaces frequently offer powerful representation in many tasks including recognition, retrieval, and optimization. In these tasks, the nearest subspaces (i.e., subspace-to-subspace search) often inevitably arise. Several studies in the literature have attempted to address this hard problem using techniques such as locality-sensitive hashing. Unfortunately, these subspace hashing methods are severely affected by poor scaling, with consequently high computational cost or unsatisfying accuracy, when the subspaces originally distribute with arbitrary dimensions. Accordingly, in this paper, we propose random angular hashing, a new and efficient type of locality-sensitive hashing, for linear subspaces of arbitrary dimension. The method we proposed preserves the angular distances among subspaces by randomly projecting their orthonormal basis and then encoding them with binary codes, meanwhile not only achieving fast computation but also maintaining a powerful collision probability. Moreover, its flexibility to easily get a balance between efficiency and accuracy in terms of performance. The extensive experimental results on tasks of face recognition, video de-duplication, and gesture recognition demonstrate that the proposed approach performs better than the state-of-the-art methods heavily, in terms of both accuracy and efficiency (up to 16× speedup).

Recent years have witnessed the success of the emerging hash-based approximate nearest neighbor search techniques in large-scale image retrieval. However, for large-scale video search, most of the existing hashing methods mainly focus on the visual content contained in the still frames, without considering their temporal relations. Therefore, they usually suffer greatly from the insufficient capability of capturing the intrinsic video similarities, from both the visual and the temporal aspects. To address the problem, we propose a temporal binary coding solution in an unsupervised manner, which simultaneously considers the intrinsic relations among the visual content and the temporal consistency among the successive frames. To capture the inherent data similarities among videos, we adopt the sparse, nonnegative feature to characterize the common local visual content and approximate their intrinsic similarities using a low-rank matrix. Then a standard graph-based loss is adopted to guarantee that the learnt hash codes can well preserve the similarities. Furthermore, we introduce a subspace rotation to model the small variation among the successive frames, and thus essentially preserve the temporal consistency in Hamming space. Finally, we formulate the video hashing problem as a joint learning of the binary codes, the hash functions and the temporal variation, and devise an alternating optimization algorithm that enjoys fast training and discriminative hash functions. Extensive experiments on three large video datasets demonstrate the proposed method significantly outperforms a number of state-of-the-art hashing methods.