In this paper, we present a three-layer (i.e., device, field, and factory layers) deterministic federated learning (FL) framework, named DetFed, which accelerates collaborative learning process for ultra-reliable and low-latency industrial Internet of Things (IoT) via integrating 6G-oriented Time-sensitive Networks (TSN). Utilizing dispersive local data, industrial IoT devices distributively train a deep neural network (DNN) model, and the updated model parameters are aggregated at their associated field servers every round or at a centralized factory server every a few rounds. Aiming at optimizing the learning accuracy of FL without affecting the co-transmission of burst traffic (e.g., safety-critical traffic), an integrated TSN is considered to establish connections among the three layers, where a cyclic queuing and forwarding mechanism is deployed in each switch to support deterministic model parameter transmission with microsecond-level delay and near-zero packet loss requirements. To improve the FL performance, we formulate a multi-objective stochastic optimization problem to simultaneously maximize the scheduling success ratio and learning accuracy while satisfying the deterministic requirements of delay, jitter, and packet loss. Since the objective function is implicit and the available time slots of the considered TSN in each FL round are temporally correlated, the problem is difficult to solve in real time. Therefore, we transform the problem into a Markov decision process formulation and propose a dynamic resource scheduling algorithm, based on deep reinforcement learning, to make optimal resource scheduling decisions while adapting to device heterogeneity and network dynamics. Experimental results based on real-world dataset demonstrate that the proposed DetFed significantly accelerates FL convergence and improves learning accuracy as compared to state-of-the-art benchmarks.

Truth discovery has received considerable attention in mobile crowdsensing systems. In real practice, it is vital to resolve conflicts among a large amount of sensory data and estimate the truthful information. Although truth discovery has been widely explored to improve aggregation accuracy, numerous security and privacy issues still need to be addressed. Existing schemes either do not guarantee the privacy of each participating user, or fail to consider practical needs in crowdsensing systems. In this paper, we present two reliable and privacy-preserving truth discovery schemes for different scenarios. Our first design is fit for applications where users are relatively stable. By employing the homomorphic Paillier encryption, one-way hash chain, and super-increasing sequence techniques, this approach not only guarantees strong privacy, but also is highly efficient and practical. Our second design suits applications where users are frequently moving. In such an application, we explore data perturbation and homomorphic Paillier encryption to shift all user workloads to the server side, without compromising users' privacy. Through detailed security analysis, we demonstrate that both schemes are secure, practical, and privacy-preserving. Moreover, extensive experiments based on real world and simulated mobile crowdsensing systems, we demonstrate the efficiency of our proposed schemes.

The integration of wearable wireless devices and cloud computing in e-health systems has significantly improved their effectiveness and availability. Patients can upload their personal health information (PHI) files to the cloud, from where the health service providers (HSPs) can obtain appropriate information to determine the health state. This system not only reduces the costs associated to healthcare, but also provides timely diagnosis to save lives. However, a number of privacy concerns arise while sharing sensitive information. In this work, we propose a novel privacy-preserving patient health information sharing scheme, which allows HSPs to access and search PHI files in a secure yet efficient manner. We make use of the searchable encryption technique with keyword range search and multi-keyword search. The proposed privacy-preserving equality test protocol allows different types of numeric comparison searches on encrypted data. We also use a variant of Bloom Filter and message authentication code to classify PHI files, filter false data, and check integrity of search results. Simulations on real world and synthetic data show the feasibility and efficiency of the system, and security analysis proves the privacy-preservation properties.

With the proliferation of machine learning, the cloud server has been employed to collect massive data and train machine learning models. Several privacy-preserving machine learning schemes have been suggested recently to guarantee data and model privacy in the cloud. However, these schemes either mandate the involvement of the data owner in model training or utilize high-cost cryptographic techniques, resulting in excessive computational and communication overheads. Furthermore, none of the existing work considers the malicious behavior of the cloud server during model training. In this paper, we propose the first privacy-preserving and verifiable support vector machine training scheme by employing a two-cloud platform. Specifically, based on the homomorphic verification tag, we design a verification mechanism to enable verifiable machine learning training. Meanwhile, to improve the efficiency of model training, we combine homomorphic encryption and data perturbation to design an efficient multiplication operation for the encryption domain. A rigorous theoretical analysis demonstrates the security and reliability of our scheme. The experimental results indicate that our scheme can reduce computational and communication overheads by at least 43.94% and 99.58%, respectively, compared to state-of-the-art SVM training methods.

Incentive plays an important role in knowledge discovery, as it impels users to provide high-quality knowledge. To promise incentive schemes with transparency, blockchain technology has been widely used in incentive schemes. Currently, privacy, reliability, streamlined processing, and quality awareness are major challenges in designing blockchain-based incentive schemes. In this paper, we design a blockchain-based eFficient and pRivacy-preserving qUality-aware IncenTive scheme called FRUIT. With well-designed smart contracts, FRUIT achieves privacy, reliability, streamlined processing, and quality awareness during the whole procedure. Specifically, we design a novel lightweight encryption method by combining matrix decomposition with proxy re-encryption and a privacy-preserving task allocation based on the polynomial fitting function and hash function. Then, we leverage our proposed lightweight encryption and task allocation to build an efficient and privacy-preserving knowledge discovery protocol in order to securely calculate the data quality and truthful knowledge. To promise user reliability in the incentive scheme, we utilize the Dirichlet distribution to realize the automatic reputation prediction based on the data quality by deploying the reputation management on the blockchain. Moreover, we also deploy the payment management on the blockchain, endowing the incentive scheme to reward participants based on the data quality automatically. Through a detailed security analysis, we demonstrate that data privacy and task privacy are well preserved during the whole process. Theoretical analysis and extensive experiments on real-world datasets demonstrate that FRUIT has acceptable efficiency and affordable performance in terms of computation cost, communication overhead, and gas consumption.

The neural network has been widely used to train predictive models for applications such as image processing, disease prediction, and face recognition. To produce more accurate models, powerful third parties (e.g., clouds) are usually employed to collect data from a large number of users, which however may raise concerns about user privacy. In this paper, we propose an Efficient and Privacy-preserving Neural Network scheme, named EPNN, to deal with the privacy issues in cloud-based neural networks. EPNN is designed based on a two-cloud model and techniques of data perturbation and additively homomorphic cryptosystem. This scheme enables two clouds to cooperatively perform neural network training and prediction in a privacy-preserving manner and significantly reduces the computation and communication overhead among participating entities. Through a detailed analysis, we demonstrate the security of EPNN. Extensive experiments based on real-world datasets show EPNN is more efficient than existing schemes in terms of computational costs and communication overhead.