Search

Deep convolutional neural networks have recently achieved dramatic success in super-resolution (SR) performance in the past few years. However, the parameters of the mapping functions of these networks require an external dataset for training. In this paper, we propose a convolutional network for image super-resolution reconstruction that can be trained using an internal dataset constructed using a single image. The proposed single image convolutional neural network (SICNN) is designed with two branches. First, a large scale-feature branch trains the feature mappings that are from the low resolution (LR) image patches to the high- resolution image (HR) patches. The LR image patches are the enlarged image patches via bicubic interpolation. Second, the small scale-feature branch trains the feature mappings that are from the down-sampling image patches to the enlarged image patches. In contrast to the existing SR networks, the SICNN enjoys two desirable properties: 1) it does not require external datasets to conduct training, and 2) it enlarges an SR image at an arbitrary scale while restoring the clear edges and textures. The results of evaluations on a wide variety of images show that the proposed SICNN achieves advantages over the state-of-the-art methods in terms of both numerical results and visual quality.

As a result of the popularity of smart mobile devices and the low cost of surveillance systems, visual data are increasingly being used in digital forensic investigation. Digital videos have been widely used as key evidence sources in evidence identification, analysis, presentation, and report. The main goal of this paper is to develop advanced forensic video analysis techniques to assist the forensic investigation. We first propose a forensic video analysis framework that employs an efficient video/image enhancing algorithm for the low quality of footage analysis. An adaptive video enhancement algorithm based on contrast limited adaptive histogram equalization (CLAHE) is introduced to improve the closed-circuit television (CCTV) footage quality for the use of digital forensic investigation. To assist the video-based forensic analysis, a deep-learning-based object detection and tracking algorithm are proposed that can detect and identify potential suspects and tools from footages.

With the continuous development of the Internet of Things (IoT) and communications technology, especially under the epoch of 5G, mobile tasks with big scales of data have a strong demand in deep learning such as virtual speech recognition and video classification. Considering the limited computing resource and battery consumption of mobile devices (MDs), these tasks are often offloaded to the remote infrastructure, like cloud platforms, which leads to the unavoidable offloading transmission delay. Edge computing (EC) is a novel computing paradigm, capable of offloading the computation tasks to the edge of networks, which reduces the transmission delay between the MDs and cloud. Therefore, combining deep learning and EC is expected to be a solution for these tasks. However, how to decide the offloading destination [cloud or deep learning-enabled edge computing nodes (ECNs)] for computation offloading is still a challenge. In this paper, a heuristic offloading method, named HOM, is proposed to minimize the total transmission delay. To be more specific, an offloading framework for deep learning edge services is built upon centralized unit (CU)-distributed unit (DU) architecture. Then, we acquire the appropriate offloading strategy by the origin-destination ECN distance estimation and heuristic searching of the destination virtual machines for accommodating the offloaded computation tasks. Finally, the effectiveness of the scheme is verified by detailed experimental evaluations.

Cyber security significantly relies on the dynamic communities in social networks. The location-based social network (LBSN) is a new type of social system that has sprung up recently that. It turns traditional social networks into heterogeneous networks by incorporating location information, which is used as the medium between the real world and the online social networks, thus bringing new challenges to the community discovery problems. This paper proposes a LBSN homogeneous network model (LSHNM) based on the user social relations and temp-spatial behaviors to calculate the user similarity relations in multi-dimensional features and construct LBSN isomorphism network topology, which can be used to improve cyber security practices. After that non-negative matrix decomposition (NMF) is used to find communities from above isomorphism network topology. The experimental results show that the LSHNM can find more satisfactory community structures.

Recent advances in both lightweight deep learning algorithms and edge computing increasingly enable multiple model inference tasks to be conducted concurrently on resource-constrained edge devices, allowing us to achieve one goal collaboratively rather than getting high quality in each standalone task. However, the high overall running latency for performing multi-model inferences always negatively affects the real-time applications. To combat latency, the algorithms should be optimized to minimize the latency for multi-model deployment without compromising the safety-critical situation. This work focuses on the real-time task scheduling strategy for multi-model deployment and investigating the model inference using an open neural network exchange (ONNX) runtime engine. Then, an application deployment strategy is proposed based on the container technology and inference tasks are scheduled to different containers based on the scheduling strategies. Experimental results show that the proposed solution is able to significantly reduce the overall running latency in real-time applications.

The advances in 5G, 6G, artificial intelligence (AI), edge computing, are expected to accelerate the future Internet of Things (FIoT) by providing intelligent, secure, and automatic connectivity for IoT services. However, there are still a number of challenges in FIoT needs to be addressed. This editorial introduces four key techniques solutions in 5G/6G communications, network model, network managements, and IoT applications in education.

The IoT provides billions of smart devices connectivity to the Internet, which drives a number of promisingapplications range from smart home to the artificial intelligence (AI) enabled vehicle-to-everything (V2X).This editorial will introduce the emerging edge intelligence enabled IoT systems, including smart home, QRbased indoor navigation system, dyslexia monitoring system, and wearable device tracing against the COVID-19.

Zero trust is an emerging framework aims at enhancing the security of Internet of Things (IoT), which connects billions of smart devices. Specifically, the zero trust security shows great benefits for securing unmanaged devices that might be unable to run computational expensive cryptographic suits.This editorial will introduce the zero trust IoT (zIoT) and its application in new IoT era.

The vulnerability of deep neural network (DNN)-based systems makes them susceptible to adversarial perturbation and may cause classification task failure. In this work, we propose an adversarial attack model using the Artificial Bee Colony (ABC) algorithm to generate adversarial samples without the need for a further gradient evaluation and training of the substitute model, which can further improve the chance of task failure caused by adversarial perturbation. In untargeted attacks, the proposed method obtained 100%, 98.6%, and 90.00% success rates on the MNIST, CIFAR-10 and ImageNet datasets, respectively. The experimental results show that the proposed ABCAttack can not only obtain a high attack success rate with fewer queries in the black-box setting, but also break some existing defenses to a large extent, and is not limited by model structure or size, which provides further research directions for deep learning evasion attacks and defenses.

Secret image sharing (SIS) is an important application of the traditional secret sharing scheme, which has become popular in recent years. In an SIS scheme, a confidential image is encrypted into a group of shadows. Any set of shadows that reaches the threshold can reconstruct the image; otherwise, nothing can be recovered at all. In most existing SIS schemes, the threshold on shadows for image reconstruction is fixed. However, in this work, we consider more complicated cases of SIS, such that the threshold is changeable according to the security environment. In this paper, we construct a (k↔h,n) threshold-changeable SIS (TCSIS) scheme using a bivariate polynomial, which provides h−k+1 possible thresholds, k,k+1,…,h. During image reconstruction, each participant can update their shadow according to the current threshold T based only on their initial shadow. Unlike previous TCSIS schemes, the proposed scheme achieves unconditional security and can overcome the information disclosure problem caused by homomorphism.

Missing or incorrect data caused by improper operations can seriously compromise security investigation. Missing data can not only damage the integrity of the information but also lead to the deviation of the data mining and analysis. Therefore, it is necessary to implement the imputation of missing value in the phase of data preprocessing to reduce the possibility of data missing as a result of human error and operations. The performances of existing imputation approaches of missing value cannot satisfy the analysis requirements due to its low accuracy and poor stability, especially the rapid decreasing imputation accuracy with the increasing rate of missing data. In this paper, we propose a novel missing value imputation algorithm based on the evidence chain (MIAEC), which first mines all relevant evidence of missing values in each data tuple and then combines this relevant evidence to build the evidence chain for further estimation of missing values. To extend MIAEC for large-scale data processing, we apply the map-reduce programming model to realize the distribution and parallelization of MIAEC. Experimental results show that the proposed approach can provide higher imputation accuracy compared with the missing data imputation algorithm based on naive Bayes, the mode imputation algorithm, and the proposed missing data imputation algorithm based on K-nearest neighbor. MIAEC has higher imputation accuracy and its imputation accuracy is also assured with the increasing rate of missing value or the position change of missing value. MIAEC is also proved to be suitable for the distributed computing platform and can achieve an ideal speedup ratio.

The widespread use of social media, cloud computing, and Internet of Things generates massive behavior data recorded by system logs, and how to utilize these data to improve the stability and security of these systems becomes more and more difficult due to the increasing number of users and amount of data. In this paper, we propose a novel model named behavior rhythm (BR) to characterize and visualize the user’s behaviors from the massive logs and apply it to the system security management. Based on the BR model, we conduct the clustering analysis to mine the user clusters. Different management and access control policies can be applied to different clusters to improve the management efficiency. Then, we apply the non-negative matrix factorization method to analyze the BRs and perform abnormal detection, and employ the BR similarity calculation to perform fast potential anomaly tracking. The detection and tracing results can help the administrators to control the threats efficiently. Experimental results based on the datasets collected from the campus network center of Xi’an Jiaotong University verify the accuracy and efficiency of our method in user behavior profiling and security management, which lay a solid foundation for improving system stability and quality of service.

The IoT based smart technologies are significantly changing many aspects of our daily lives, including smart home assistant, smart helmet, security, location-based services, etc. Actually, the emerging technologies, such as 5G, artificial intelligence (AI), play key roles at the heart of the industry IoT transformation. This editorial summarised the most recent research results in different areas.

With the exponential growth of the Internet of Things (IoT) and cyber-physical systems (CPS), a wide range of IoT applications have been developed and deployed in recent years. To match the heterogeneous application requirements in IoT and CPS systems, many resource-constrained IoT devices are deployed, in which privacy and security have emerged as difficult challenges because the devices have not been designed to have effective security features.