The integration of smart medical technology has revolutionized personal health management, with devices like smartwatches and smartphones facilitating efficient data collection and analysis. However, the exponential growth in medical data has heightened privacy concerns, underscoring the need for a robust and secure transmission system. Traditional encryption, while effective, is limited by its resource-intensive nature, hindering its application in IoT devices. Our research delves into the application of compressed sensing for medical image privacy, addressing challenges such as high resource consumption, security vulnerabilities, and the need for robust watermarking. We have designed the Lightweight Quantum-resistant Medical Image Transmission (LQRMIT) model, which combines a 3D chaotic system with the Learning With Errors (LWE) theory, offering a key space of up to$2^{700}$, significantly enhancing security. Additionally, we have designed a watermark embedding algorithm compatible with compressed sensing, which ensures high invisibility and high-quality watermark extraction. The model also employs a meaningful image hiding strategy within encrypted transmissions to improve stealth. Extensive experimental and theoretical analysis has validated the effectiveness of our solution, providing a promising approach for secure and efficient medical image transmission in the digital medical era.

In the context of connected vehicles, encrypted image search technologies have gained significant importance. However, existing techniques are plagued by several limitations, including high resource consumption, lack of flexibility, insufficient security evaluation, and absence of hierarchical security mechanisms. These constraints render them inadequate for meeting the diverse resource requirements and multi-level security needs of different devices within the connected vehicle ecosystem. Therefore, improvements are urgently needed to enhance efficiency and security. In light of these issues, we introduce a novel framework designed for the secure retrieval of k-nearest Neighbor (kNN) images, leveraging cloud-based storage. This framework is underpinned by a Self-Adaptive Asymmetric Scalar Product Homomorphic Encryption algorithm (SA-ASPE). It incorporates an adaptive block mechanism to generate a suite of encryption keys tailored for images of diverse dimensions. Additionally, a security hierarchy, facilitated by a recursive split tree, is established to provide a selection of multi-tiered security options suitable for a variety of IoT contexts. To further augment the security of image feature vectors, we have integrated chaotic encryption techniques, which serve to thoroughly randomize these vectors, thereby significantly enhancing their unpredictability. To rigorously assess the robustness of the ASPE protocol against potential security threats, we have devised a comprehensive four-tier attack model coupled with an Amplified Attack (AA) strategy. Ultimately, through an extensive comparative analysis of existing and proposed schemes, we demonstrate that our framework adeptly satisfies the efficiency and security demands of a multitude of IoT device privacy protection scenarios.

The Internet of Things (IoT) technology, through the deployment of sensors and intelligent traffic cameras, facilitates the collection, processing, and analysis of traffic flow, vehicle density, and road conditions. However, traffic image data contains a substantial amount of personal sensitive information. Therefore, implementing hierarchical encryption on images is crucial for preventing the leakage of personal privacy and ensuring data is used in compliance with regulations. Moreover, the rise of quantum computing poses a substantial threat to existing key distribution systems, especially for resource-constrained IoT devices. To address these challenges, this article presents a lightweight hierarchical privacy protection scheme for traffic images that addresses key adaptability, quantum threats, and hierarchical privacy protection. The scheme includes a lattice-based public-key algorithm with updatable public and private keys to ensure quantum-resistant and forward security. It also introduces, for the first time, a hierarchical tree-structured encryption scheme based on the SHA3 hash function and the homomorphic property of compressed sensing, allowing high-permission users to generate keys for low-permission users without compromising the quality of image decryption. Additionally, the scheme includes a compressed sensing public-key encryption algorithm based on chaotic systems and the difficulty of matrix factorization, which supports sampling of images of any size, enhancing adaptability and reducing resource consumption. Tests indicate that this image encryption algorithm can resist various statistical analyses, achieving a high number of pixel change rate of 99.6277%. In summary, our research significantly contributes to the effective safeguarding of personal data security in resource-constrained IoT environments, particularly in the face of potential quantum computing threats.

With the development of content generative large models, spoofed speech can be synthesized more easily. However, the generalization of current cloud voice anti-spoofing detection model is insufficient, especially when facing attacks from unknown speech synthesis algorithms. Meanwhile, voice data is also susceptible to attacks during transmission. Therefore, we design an end-to-end voice anti-spoofing scheme, which can be applied in IoT cloud service. The scheme consists of an encryption transmission module and an anti-spoofing model. The encryption module is designed with post-quantum cryptography and chaos to protect transmission security. In the modle, we propose a new higher-order two-dimensional attentive statistics pooling (H2D-ASP) module to extract and aggregate more attention representations in spectral domain and temporal domain; And we propose a new channel-dependent self attention based graph aggregation (CSA-GA) module, which squeezes and aggregates spectral graphs and temporal graphs. Finally, we conduct experiments on the ASVspoof 5 Challenge deepfake database under the closed condition. The experiments show that the model in our scheme is a better single model which performs minDCF 18.28% better than the baseline model on the evaluation set. Without data augmentation, the model achieves minDCF of 0.2994, 0.3604, 0.581 on the development set, evaluationprog set and evaluation set, respectively. The last two proposed modules improve the vanilla model by 29.75% on evaluationprog set and 18.97% on evaluation set.