The application of the Internet of Things (IoT) in medical field has brought unprecedented convenience to human beings. However, attackers can use device configuration vulnerabilities to hijack devices, control services, steal medical data, or make devices operate illegally. These restrictions have led to huge security risks for IoT, and have challenged the management of critical infrastructure services. Based on these problems, this paper proposes an anomaly detection system for detecting illegal behavior (DIB) in medical IoT environment. The DIB system can analyze data packets transmitted by medical IoT devices, learn operation rules by itself, and remind management personnel that the device is in an abnormal operation state to ensure the safety of control service. We further propose a model which is based on rough set (RS) theory and fuzzy core vector machine (FCVM) to improve the accuracy of DIB classification anomalies. Experimental results show that the R-FCVM is effective.

SDN has provided significant convenience for network providers and operators in cloud computing. Such great advantage is extending to the IoT network. However, it also increases the risk if the security of a SDN network is compromised. For example, if the network operator’s permission is illegally obtained by a hacker, he/she can control the entry of the SDN network. Therefore, an effective authentication scheme is needed to fit various application scenarios with high-security requirements. In this paper, we design, implement and evaluate a new authentication scheme called The Hidden Pattern (THP), which combines graphics password and digital challenge value to prevent multiple types of authentication attacks at the same time. We examined THP in the perspectives of both security and usability, with a total number of 694 participants in 63 days. Our evaluation shows that THP can provide better performance than the existing schemes in terms of security and usability.

Access management of IoT devices is extremely important, and a secure login authentication scheme can effectively protect users’ privacy. However, traditional authentication schemes are threatened by shoulder-surfing attacks, and biometric-based schemes, such as fingerprint recognition and face recognition, that are commonly used today can also be cracked. Researchers have proposed some schemes for current attacks, but they are limited by usability. For example, the login authentication process requires additional device support. This method solves the problem of attacks, but it is unusable, which limits its application. At present, most authentication schemes for the Internet of Things and mobile platforms either focus on security, thus ignoring availability, or have excellent convenience but insufficient security. This is a symmetry problem worth exploring. Therefore, users need a new type of login authentication scheme that can balance security and usability to protect users’ private data or maintain device security. In this paper, we propose a login authentication scheme named PinWheel, which combines a textual password, a graphical password, and biometrics to prevent both shoulder-surfing attacks and smudge attacks and solves the current schemes’ lack of usability. We implemented PinWheel and evaluated it from the perspective of security and usability. The experiments required 262 days, and 573 subjects participated in our investigation. The evaluation results show that PinWheel can at least effectively resist both mainstream attacks and is superior to most existing schemes in terms of usability.

Deep neural networks (DNNs) are widely deployed across applications, but growing model sizes and performance demands on edge devices necessitate aggressive compiler-based acceleration, as enabled by frameworks like TVM. Compiling models into standalone executables, however, introduces new security risks, reverse-engineering these binaries can reveal core architectures and enable unauthorized model cloning or tampering. Existing defenses largely rely on trusted execution or encryption, which target weight confidentiality but fail to protect architectural details or operator attributes, often with high runtime costs. To address these gaps, we propose MACO, a multi-tier obfuscation framework built atop TVM, spanning high-level graph rewriting and low-level IR transformations. MACO integrates three transformation categories: operator attribute obfuscation, memory layout perturbation, and dummy branching, covering the full compilation stack. A tiered design enables flexible trade-offs between security and performance. We evaluate MACO on eight common models and three attack types: side-channel, symbolic execution, and deep learning–based extraction. Results show up to 85.8% reduction in attribute inference success and 99% reduction in topology reconstruction, with the lightest tier incurring only 1.036× overhead over the unobfuscated baseline.

Deep learning compilers such as TVM lower neural networks through intermediate representations (IRs) into optimized, hardware-specific binaries. While enabling high-performance deployment via optimizations like operator fusion and loop tiling, they leave stable execution signatures exploitable by reverse engineering. Prior attacks often rely on a single modality, symbolic lifting, instruction classification, or side channels, each struggles under at least one realistic condition, such as deep fusion, schedule diversity, or OS noise. We present SOFAN, a side-channel oriented fingerprinting and neutralization framework for TVM-compiled DNNs. On the attack side, TCScaptures timing and cache traces to recover operator boundaries via smoothing, non-maximum suppression (NMS), and dynamic time warping (DTW). A multimodal fusion network (MFN) then integrates these side-channel signals with instruction embeddings to classify deeply fused operators. On the defense side, LASR (Leakage-Aware Schedule Rewriting) selectively perturbs critical leakage via schedule diversification, access equalization, and memory remapping, under a fixed runtime budget. Evaluated across CNNs and fusion schedules, SOFANimproves segmentation and recognition over prior baselines. LASR reduces Top-1 attack accuracy by up to 24 points (16 on average) under 10-20% runtime overhead and minimal memory cost. By aligning both attack and defense with compiler boundaries, SOFANenables practical, budget-aware protection for real-world deployments.

Deep Neural Networks (DNNs) are gradually becoming indispensable in various technological domains. To cater to more deployment backends and increasingly complex model architectures, deep learning compiler-driven efficient compilation modes are becoming essential components of productivity. However, this deployment method exacerbates security risks. Recent studies have shown that attackers can reverse-engineer executable files to regenerate trainable deep learning models, leading to adversarial attacks and other security breaches. Previous research indicates that such attacks pose significant threats, yet progress in implementing cost-effective mitigation strategies remains limited. Existing defense mechanisms primarily focus on Trusted Execution Environments or partial encryption to protect critical model parameters, often at the expense of compiled execution efficiency. To address this gap, we propose a schedule search based operator obfuscation method (SOOM) to defend against model extraction attacks for models compiled and executed on standard CPU and GPU backends, where low latency on device inference is required. SOOM is built on TVM, a deep learning compiler, and constructs a comprehensive obfuscation space for deep learning operators. It leverages a security aware learned cost model based on XGBoost gradient boosted trees to balance security objectives and performance requirements, and ultimately generates obfuscated executable code for various deep learning operators. Extensive experiments covered over 105 operator configurations and more than 30,000 tensor computation test cases. Our method was tested against state-of-the-art model extraction attacks, raising the operator inference failure rate to as high as 89%. We also observe up to approximately 25.4% performance gains in selected cases, while the balanced setting keeps model-level latency overhead within a modest budget.