Encrypted databases have been extensively studied with the increasing concern of data privacy in cloud services. For practical efficiency, most encrypted database systems are built under Dynamic Searchable Symmetric Encryption (DSSE) schemes to support fast query and update over encrypted data. However, DSSE schemes allow leakages in their security frameworks, especially access-pattern leakages (i.e., the search results corresponding to queried keywords), which lead to various attacks to infer sensitive information of queries and databases. Existing oblivious-access techniques, such as Oblivious RAM and differential privacy, suffer from excessive communication overhead and loss of query accuracy. In this paper, we propose a new DSSE scheme that enables access-pattern hiding keyword search and update operations. Servers can obliviously query and update databases within only a single communication round. Our building block is based on the Distributed Point Function (DPF), an advanced secret sharing technique that provides provable security guarantees against adversaries with arbitrary background knowledge. Moreover, we devise a novel update protocol that integrates DPF and Somewhat Homomorphic Encryption (SHE) such that servers can obliviously update their local data. We formally analyze the security and implement the prototype. The comprehensive experimental results demonstrate the security and efficiency of our scheme.

IOTA has emerged as a promising blockchain platform specially designed for the Internet of Things (IoT). Its distributed ledger, called tangle, adopts a directed acyclic graph (DAG) structure to achieve fast transaction confirmation and high scalability. While the tangle tremendously mitigates blockchain performance concerns relative to a traditional single chain, it simultaneously increases the potential risk of double-spending attacks. Utilizing constructing illegal tangle branches to substitute for legitimate ones, attackers inside IOTA can launch double-spending attacks and seriously compromise the tangle security. In this work, we take the first step toward investigating the problem of tangle vulnerability by leveraging the large deviation theory. The proposed scheme, called SecTangle, can assist IOTA in effectively reducing the tangle vulnerability to resist double-spending attacks. The core idea is to explore the security threshold defined and deduced to affect the robustness of the tangle by evaluating the probability of tangle vulnerability. By adjusting the critical factors of the security threshold, fake tangle branches can be found by IOTA efficiently to prevent double-spending attacks. Besides, we further devise a transaction recovery algorithm to recover time-sensitive legitimate transaction branches. This paper validates that the proposed scheme is efficient with comprehensive theoretical analysis and simulation experiments.

Pre-trained encoders in computer vision have recently received great attention from both research and industry communities. Among others, a promising paradigm is to utilize self-supervised learning (SSL) to train image encoders with massive unlabeled samples, thereby endowing encoders with the capability to embed abundant knowledge into the feature representations. Backdoor attacks on SSL disrupt the encoder's feature extraction capabilities, causing downstream classifiers to inherit backdoor behavior and leading to misclassification. Existing backdoor defense methods primarily focus on supervised learning scenarios and cannot be effectively migrated to SSL pre-trained encoders. In this article, we present a backdoor defense scheme based on self-supervised knowledge distillation. Our approach aims to eliminate backdoors while preserving the feature extraction capability using the downstream dataset. We incorporate the benefits of contrastive and non-contrastive SSL methods for knowledge distillation, ensuring differentiation between the representations of various classes and the consistency of representations within the same class. Consequently, the extraction capability of pre-trained encoders is preserved. Extensive experiments against multiple attacks demonstrate that the proposed scheme outperforms the state-of-the-art solutions.

Named-Data Networking (NDN) is a novel network that secures network communication by fetching semantically named and secured data. All data packets in NDN are signed by producers and verified by data consumers. Therefore, it is vital to have producers' certificates available all the time. In this paper, we describe the design of CLedger, a secure distributed certificate ledger, to ensure certificate availability in NDN. CLedger logs certificate records in an immutable Directed Acyclic Graph (DAG) structure and replicates the DAG among a set of distributed loggers. We implemented CLedger using NDN's pub/sub API, and evaluated our design through an emulated deployment setting. Our initial evaluation results show that CLedger is effective, efficient, and resilient to failures.

Federated learning has recently emerged as a popular learning paradigm that enables multiple clients to jointly train a high-quality model without sharing their local training datasets. Each client will train its local model by its local dataset, and the global model will be aggregated by local models. However, training datasets from all clients are usually heterogeneous, because they are chosen by the clients themselves. This leads to over-fitting in the local models, thus affecting the performance of the global model. There are existing methods such as regularization in local optimization and improving model aggregation. However, they introduce additional computing or storage overhead.In this paper, we present a novel system design FedIR to eliminate the impact of data heterogeneity. FedIR introduces the idea of learning invariant features in domain adaptation so that the aggregated global model can handle the data heterogeneity well. We refer to the Optimal Path Search to assist model training in obtaining better invariant representations. The modifications to local model structures are very small, with little impact on local training and server aggregation. Extensive experiments demonstrate that FedIR achieves state-of-the-art performance on popular federated learning benchmarks including CIFAR-10 and CIFAR-100, with less computation cost and communication rounds.

Encrypted range query schemes that enable range-based searches over encrypted data have become an effective solution for secure data outsourcing services. However, existing schemes are still inadequate on desired functionality and security. Specifically, supporting efficient multi-range queries while hiding the data ordering leakage remains a challenging research problem. Existing works on order-hiding query schemes only work for encrypted single-range search and incur significant computational overhead due to the protection of the ordering information. In this paper, we present a privacy-preserving multi-range query scheme that can address the above problems simultaneously. It not only enables efficient multi-range queries over encrypted data but also guarantees the privacy of ordering information. To protect the ordering leakage, our design adopts an Order-hiding Encoding (OHE) scheme to support multi-range obfuscation. In addition, a novel order-hiding KD-tree index structure is designed as the core technique underlying our scheme, which accelerates efficient multi-range queries and obfuscates ordering information of encrypted values. Finally, the formal security analysis confirms that our proposed multi-range query scheme is secure in the random oracle model. The extensive experimental results conducted on real-world datasets demonstrate the practicality of our design.

Fog computing has emerged as a promising framework with the rapid growth of the Internet of Things (IoT). In fog computing, the new entity, named fog device, can help the cloud process the large amount of data generated by IoT devices. Along with this trend, a location-based query scheme that collects IoT devices’ data from specific areas is an important application, especially in fog-enhanced sensor networks. However, in this application, the cloud and fog devices require the user’s query, sensors’ locations, and sensor data so that it raises critical privacy and security concerns. In this article, we devise a privacy-preserving-location-based data query scheme in fog-enhanced sensor networks, which allows the cloud and fog devices to collect sensor data from a query area without learning the three kinds of information. Specifically, we resort to a cryptographic primitive, named somewhat homomorphic encryption (SHE), with ciphertext packing to encrypt query, locations, and sensor data and efficiently calculate the distances between the user’s query and sensors. Then, we show how to build a hardware-assisted data query scheme to extract the matched data based on the distances. We formally analyze the security strengths and implement the system prototype. In order to implement secure processing within software guard extension (SGX), we make an effort to adapt the existing mathematical libraries to the advanced SGX trusted environment. Evaluation results demonstrate that our proposed design is secure and efficient.

The explosive growth of the Internet of Things (IoT) is pushing forward the paradigm of fog computing services today. An important service for most fog-enhanced applications is geographic range-match, which means the fog-nodes can accurately collect sensed data from IoT devices based on their location distances. However, due to the increasing attacking surfaces, outsourcing range query operations to untrusted fog-nodes inevitably raises new privacy concerns about query content and device location exposure. In this article, we devise a new geographic range-match scheme for fog-enhanced services, which allows fog-nodes to securely collect range-based sensed data while protecting the location privacy of IoT devices. Our main idea is to formulate the problem of encrypted geographic queries as range-based pattern matching and carefully craft security schemes to enable efficient range queries in the ciphertext domain. The proposed range-match scheme is provably secure and can reduce accessible information during distance comparisons. We formally analyze the security strengths and complete the prototype implementation. The comprehensive experimental results demonstrate the practicality of our designs.

Crowdsourcing has attracted widespread attention in recent years and developed into various applications. An indispensable service of crowdsourcing systems is task recommendation, which means tasks should be accurately recommended to the workers with aligned interests. However, existing systems rely on their separate servers to conduct recommendation services, resulting in computing resources locked inside each isolated system. Moreover, due to the wide attacking surfaces of traditional centralized servers setting, existing systems are subject to single points of failure or malicious data breaches. Therefore, failure to address these inherent limitations properly will hinder the wide adoption of crowdsourcing. In this article, we propose and implement FedCrowd, the first federated and privacy-preserving crowdsourcing platform by using blockchain technology. Our main idea is to employ the smart contract as a trusted platform for systems to release encrypted tasks, and carefully craft matching protocols to enable efficient task recommendations in the ciphertext domain. Our task-matching protocols are highly customized for the decentralized settings, where users can securely perform keyword and range-based queries over federated task indexes without sharing secret keys. We formally analyze the security strengths and complete the prototype implementation on Ethereum. Experiment results demonstrate the feasibility and usability of the FedCrowd platform.

The increasing popularity of Online Ride-hailing (ORH) services has greatly facilitated our daily travel. It enables a rider to easily request the nearest driver through mobile devices in a short time. However, existing ORH systems require the collection of users' location information and thus raise critical privacy concerns. While several privacy-preserving solutions for ORH service have been proposed, most of existing schemes rely on an additional trusted party to compute the distance between a rider and a driver. Such a security assumption cannot fully address the privacy concerns for practical deployment. In this paper, we present a new ride-matching scheme for ORH systems, which allows privacy-preserving and effective distance calculation without involving a third-party server. Our proposed scheme enables ORH systems to securely compute the user distance while protecting the location privacy of both riders and drivers. Specifically, we resort to state-of-the-art distance calculation techniques based on Road Network Embedding (RNE), and show how to uniquely bridge cryptographic primitives like Property-preserving Hash (PPH) with RNE in depth to support privacy-preserving ride-matching services. Moreover, we also propose an optimized design to improve the matching efficiency. We formally analyze the security strengths and implement the system prototype. Evaluation results demonstrate that our design is secure and efficient for ORH systems.