The satellite-to-ground link is susceptible to eavesdropping from various open-space locations. In this paper, we investigate a new eavesdropping scenario in a multiconstellation satellite network, where very low earth orbit (VLEO) satellites eavesdrop on low earth orbit (LEO) constellations from legitimate positions. We model the VLEO eavesdropping risks in a theoretical way and propose a crossinterference scheme called LEO active cross-interference (LEOACI) to protect satellite-to-ground downlinks from eavesdropping attacks. The scheme prevents eavesdropping by scheduling spare satellites to transmit artificial noise (AN) over the protected link. Finally, we conduct simulations to evaluate the performance of the LEO-ACI scheme and it shows that VLEO eavesdroppers can possess high eavesdropping coverage and long duration. Meanwhile, the LEO network can ensure that 98.5 % of downlinks are free from eavesdropping at the cost of 12.3 % of the satellite-to-ground capacity degradation.

The use of smart contracts enhances the capabilities of blockchain-based botnets, allowing for greater information capacity, richer application scenarios, and the deployment of program functions directly on the blockchain. However, smart blockchains offer a better solution for the intelligence of IoT systems, but they also come with some security risks. Botnet is a highly insecure community because it is used to do hazardous things like Distributed Denial of Service (DDoS). It is extremely essential to detect botnets with some useful tools, such as artificial intelligence (AI) algorithms, because these algorithms can assist us to monitor the network automatically. We need to pay the utmost attention to some feature engineering work, as recognition rates of AI models are considerably improved with suitable features. In this article, we propose domain embedding (DE) models to generate low-dimensional features for domains with unsupervised learning algorithms. We also explore some key parameters of the DE model to obtain decent effects on domain features. A modified version of the$k$-means algorithm called extended$k$-means, is used to cluster these domains in certain hubs and botnets that can be found for smart blockchain-based IoT systems. In the experiments, some domain correlation scores can be computed during the DE model, and similar domains have higher correlation scores.

Memory-intensive applications, such as in-memory databases, caching systems, and key-value stores, are increasingly demanding larger main memory to fit their working sets. Conventional swapping can enlarge the memory capacity by paging out inactive pages to backend stores. However, existing swapping solutions suffer several performance and compatibility issues, making them unsuitable for high-concurrency and memory-intensive applications. In this article, we redesign the swapping system and propose Lightswap, a high-performance user-space swapping solution that supports paging with both local SSDs and remote memories. First, to avoid kernel involvement, we propose to leverage the extended Berkeley packet filter (eBPF) for handling page faults (PFs) in user space and further eliminate the heavy I/O stack with the help of user-space I/O drivers. Then, we co-design the PF handling with lightweight thread (LWT) scheduling to improve system throughput and reduce the end-to-end PF latency. Finally, we propose a try-catch framework in Lightswap to deal with swap-in errors which have been exacerbated by the scaling in process technology. We implement Lightswap in our production-level system and evaluate it with various benchmarks. Results show that Lightswap achieves scalable PF notification latency ($4 \mu \text{s}$under 128 LWTs), reduces the PF handling latency by 3–5 times, and improves the throughput of memcached by more than 40% compared with the state-of-art swapping systems.

Summary In the process of building a smart city, face recognition can be applied to the transformation of enterprises, communities, and parks. The combination of building security system and face recognition technology can improve the security experience of enterprises and citizens through the solution of hardware and software integration. Face recognition is still facing the challenges of illumination, occlusion, and attitude change in the actual application process. In addition, the end‐to‐end convolutional neural networks (CNN) seldom make use of the hierarchical feature of the network. So, we propose a hierarchy feature fusion method for face recognition, which uses supervisory information to learn shallow and deep facial features. The features are fused to enhance the recognition accuracy of face recognition against illumination and occlusion. The method is applied to the transformation of the visual geometry group network and Lightened CNN. The face recognition experiments are carried out using the hierarchy network. Our method has achieved good recognition results in the labeled faces in the wild (LFW) and AR face databases.

Deep learning is widely used in the medical field owing to its high accuracy in medical image classification and biological applications. However, under collaborative deep learning, there is a serious risk of information leakage based on the deep convolutional generation against the network's privacy protection method. Moreover, the risk of such information leakage is greater in the medical field. This paper proposes a deep convolution generative adversarial networks (DCGAN) based privacy protection method to protect the information of collaborative deep learning training and enhance its stability. The proposed method adopts encrypted transmission in the process of deep network parameter transmission. By setting the buried point to detect a generative adversarial network (GAN) attack in the network and adjusting the training parameters, training based on the GAN model attack is forced to be invalid, and the information is effectively protected.

The emergence of artificial intelligence technology has promoted the development of the Internet of Things. However, this promising cyber technology can encounter serious security problems while accessing the internet. A malicious website can disguise itself as a normal website, and obtain users' private information. Thus, it is very important to detect malicious websites using tools such as machine learning (ML) algorithms, as these algorithms can help us to identify abnormal information hidden in the mass traffic more easily. Accordingly, many feature engineering tasks must be performed from memory, as a strong machine learning model is greatly improved with good features. In this article, we propose an unsupervised learning algorithm that learns URL embedding. We also explore some key parameters regarding a domain embedding model to obtain a good effect on domain features.

While the industrial Internet of Things (IIoT) brings convenience to the industry, it also brings security problems. Due to the massive amount of data generated by the surge of IIoT devices, it is impossible to ensure whether these data contain an attack or untrustworthy data, therefore, how to ensure the security and trustworthiness of IIoT devices has become an urgent problem to solve. In this article, we design a new hinge classification algorithm based on mini-batch gradient descent with an adaptive learning rate and momentum (HCA-MBGDALRM) to minimize the effects of security attacks. The algorithm significantly improves the performance of deep network training compared with traditional neural networks, decision trees and logistic regression in terms of scale and speed. In addition, we have solved the data skew problem in the shuffle phase, and we implement a parallel framework for HCA-MBGDALRM to accelerate the processing speed of very large traffic data sets.