Biblio

Deep neural networks (DNNs) are effective machine learning models to solve a large class of recognition problems, including the classification of nonlinearly separable patterns. The applications of DNNs are, however, limited by the large size and high energy consumption of the networks. Recently, stochastic computation (SC) has been considered to implement DNNs to reduce the hardware cost. However, it requires a large number of random number generators (RNGs) that lower the energy efficiency of the network. To overcome these limitations, we propose the design of an energy-efficient deep belief network (DBN) based on stochastic computation. An approximate SC activation unit (A-SCAU) is designed to implement different types of activation functions in the neurons. The A-SCAU is immune to signal correlations, so the RNGs can be shared among all neurons in the same layer with no accuracy loss. The area and energy of the proposed design are 5.27% and 3.31% (or 26.55% and 29.89%) of a 32-bit floating-point (or an 8-bit fixed-point) implementation. It is shown that the proposed SC-DBN design achieves a higher classification accuracy compared to the fixed-point implementation. The accuracy is only lower by 0.12% than the floating-point design at a similar computation speed, but with a significantly lower energy consumption.

This paper proposed an advanced round modification fault analysis (RMFA) at the theoretical level on AEGIS-128, which is one of seven finalists in CAESAR competition. First, we clarify our assumptions and simplifications on the attack model, focusing on the encryption security. Then, we emphasize the difficulty of applying vanilla RMFA to AEGIS-128 in the practical case. Finally we demonstrate our advanced fault analysis on AEGIS-128 using machine-solver based algebraic techniques. Our enhancement can be used to conquer the practical scenario which is difficult for vanilla RMFA. Simulation results show that when the fault is injected to the initialization phase and the number of rounds is reduced to one, two samples of injections can extract the whole 128 key bits within less than two hours. This work can also be extended to other versions such as AEGIS-256.

The engine is a significant and dynamic component of the aircraft. Because of the complicated structure and severe operating environment, the fault detection of the engine has always been the key and difficult issue in the field of reliability. Based on an engine and the acoustic emission technology, we propose a method of identifying fault types and determining different components in the engine by constructing the attenuation coefficient. There are several common faults of engines, and three different types of fault sources are generated experimentally in this work. Then the fault signal of the above fault sources propagating in different engine components are obtained. Finally, the acoustic emission characteristics of the fault signal are extracted and judged by the attenuation coefficient. The work effectively identifies different types of faults and studies the effects of different structural components on the propagation of fault acoustic emission signals, which provides a method for the use of acoustic emission technology to identify the faults types of the engine and to study the propagation characteristics of AE signals on the engine.*

Recently, the armed forces want to bring the Internet of Things technology to improve the effectiveness of military operations in battlefield. So the Internet of Battlefield Things (IoBT) has entered our view. And due to the high processing latency and low reliability of the “combat cloud” network for IoBT in the battlefield environment, in this paper , a novel “combat cloud-fog” network architecture for IoBT is proposed. The novel architecture adds a fog computing layer which consists of edge network equipment close to the users in the “combat-cloud” network to reduce latency and enhance reliability. Meanwhile, since the computing capability of the fog equipment are weak, it is necessary to implement distributed computing in the “combat cloud-fog” architecture. Therefore, the distributed computing load balancing problem of the fog computing layer is researched. Moreover, a distributed generalized diffusion strategy is proposed to decrease latency and enhance the stability and survivability of the “combat cloud-fog” network system. The simulation result indicates that the load balancing strategy based on generalized diffusion algorithm could decrease the task response latency and support the efficient processing of battlefield information effectively, which is suitable for the “combat cloud- fog” network architecture.

The broadcast nature of visible light beam has aroused great concerns about the privacy and confidentiality of visible light communication (VLC) systems.In this paper, in order to enhance the physical layer security, we propose a channel scrambling scheme, which realizes orthogonal factorized channel scrambling with location information embedded (OFCS-LIE) for the VLC systems. We firstly embed the location information of the legitimate user, including the transmission angle and the distance, into a location information embedded (LIE) matrix, then the LIE matrix is factorized orthogonally in order that the LIE matrix is approximately uncorrelated to the multiple-input, multiple-output (MIMO) channels by the iterative orthogonal factorization method, where the iteration number is determined based on the orthogonal error. The resultant OFCS-LIE matrix is approximately orthogonal and used to enhance both the reliability and the security of information transmission. Furthermore, we derive the information leakage at the eavesdropper and the secrecy capacity to analyze the system security. Simulations are performed, and the results demonstrate that with the aid of the OFCS-LIE scheme, MIMO-based VLC system has achieved higher security when compared with the counterpart scrambling scheme and the system without scrambling.

Transferring artistic styles onto everyday photographs has become an extremely popular task in both academia and industry. Recently, offline training has replaced online iterative optimization, enabling nearly real-time stylization. When those stylization networks are applied directly to high-resolution images, however, the style of localized regions often appears less similar to the desired artistic style. This is because the transfer process fails to capture small, intricate textures and maintain correct texture scales of the artworks. Here we propose a multimodal convolutional neural network that takes into consideration faithful representations of both color and luminance channels, and performs stylization hierarchically with multiple losses of increasing scales. Compared to state-of-the-art networks, our network can also perform style transfer in nearly real-time by performing much more sophisticated training offline. By properly handling style and texture cues at multiple scales using several modalities, we can transfer not just large-scale, obvious style cues but also subtle, exquisite ones. That is, our scheme can generate results that are visually pleasing and more similar to multiple desired artistic styles with color and texture cues at multiple scales.

Recent years, the issue of cyber security has become ever more prevalent in the analysis and design of electrical cyber-physical systems (ECPSs). In this paper, we present the TrueTime Network Library for modeling the framework of ECPSs and focuses on the vulnerability analysis of ECPSs under DoS attacks. Model predictive control algorithm is used to control the ECPS under disturbance or attacks. The performance of decentralized and distributed control strategies are compared on the simulation platform. It has been proved that DoS attacks happen at dada collecting sensors or control instructions actuators will influence the system differently.

With the integration of computing, communication, and physical processes, the modern power grid is becoming a large and complex cyber physical power system (CPPS). This trend is intended to modernize and improve the efficiency of the power grid, yet it makes the CPPS vulnerable to potential cascading failures caused by cyber-attacks, e.g., the attacks that are originated by the cyber network of CPPS. To prevent these risks, it is essential to analyze how cyber-attacks can be conducted against the CPPS and how they can affect the power systems. In light of that General Packet Radio Service (GPRS) has been widely used in CPPS, this paper provides a case study by examining possible cyber-attacks against the cyber-physical power systems with GPRS-based SCADA system. We analyze the vulnerabilities of GPRS-based SCADA systems and focus on DoS attacks and message spoofing attacks. Furthermore, we show the consequence of these attacks against power systems by a simulation using the IEEE 9-node system, and the results show the validity of cascading failures propagated through the systems under our proposed attacks.

This paper considers a framework of electrical cyber-physical systems (ECPSs) in which each bus and branch in a power grid is equipped with a controller and a sensor. By means of measuring the damages of cyber attacks in terms of cutting off transmission lines, three solution approaches are proposed to assess and deal with the damages caused by faults or cyber attacks. Splitting incident is treated as a special situation in cascading failure propagation. A new simulation platform is built for simulating the protection procedure of ECPSs under faults. The vulnerability of ECPSs under faults is analyzed by experimental results based on IEEE 39-bus system.

A high definition(HD) wide dynamic video surveillance system is designed and implemented based on Field Programmable Gate Array(FPGA). This system is composed of three subsystems, which are video capture, video wide dynamic processing and video display subsystem. The images in the video are captured directly through the camera that is configured in a pattern have long exposure in odd frames and short exposure in even frames. The video data stream is buffered in DDR2 SDRAM to obtain two adjacent frames. Later, the image data fusion is completed by fusing the long exposure image with the short exposure image (pixel by pixel). The video image display subsystem can display the image through a HDMI interface. The system is designed on the platform of Lattice ECP3-70EA FPGA, and camera is the Panasonic MN34229 sensor. The experimental result shows that this system can expand dynamic range of the HD video with 30 frames per second and a resolution equal to 1920*1080 pixels by real-time wide dynamic range (WDR) video processing, and has a high practical value.

Although connecting a microgrid to modern power systems can alleviate issues arising from a large penetration of distributed generation, it can also cause severe voltage instability problems. This paper presents an online method to analyze voltage security in a microgrid using convolutional neural networks. To transform the traditional voltage stability problem into a classification problem, three steps are considered: 1) creating data sets using offline simulation results; 2) training the model with dimensional reduction and convolutional neural networks; 3) testing the online data set and evaluating performance. A case study in the modified IEEE 14-bus system shows the accuracy of the proposed analysis method increases by 6% compared to back-propagation neural network and has better performance than decision tree and support vector machine. The proposed algorithm has great potential in future applications.

There are several security requirements identification methods proposed by researchers in up-front requirements engineering (RE). However, in open source software (OSS) projects, developers use lightweight representation and refine requirements frequently by writing comments. They also tend to discuss security aspect in comments by providing code snippets, attachments, and external resource links. Since most security requirements identification methods in up-front RE are based on textual information retrieval techniques, these methods are not suitable for OSS projects or just-in-time RE. In our study, we propose a new model based on logistic regression to identify security requirements in OSS projects. We used five metrics to build security requirements identification models and tested the performance of these metrics by applying those models to three OSS projects. Our results show that four out of five metrics achieved high performance in intra-project testing.

In this paper, a practical quantum public-key encryption model is proposed by studying the recent quantum public-key encryption. This proposed model makes explicit stipulations on the generation, distribution, authentication, and usage of the secret keys, thus forms a black-box operation. Meanwhile, this proposed model encapsulates the process of encryption and decryption for the users, and forms a blackbox client-side. In our models, each module is independent and can be replaced arbitrarily without affecting the proposed model. Therefore, this model has a good guiding significance for the design and development of the quantum public key encryption schemes.

Third-party IME (Input Method Editor) apps are often the preference means of interaction for Android users' input. In this paper, we first discuss the insecurity of IME apps, including the Potentially Harmful Apps (PHA) and malicious IME apps, which may leak users' sensitive keystrokes. The current defense system, such as I-BOX, is vulnerable to the prefix-substitution attack and the colluding attack due to the post-IME nature. We provide a deeper understanding that all the designs with the post-IME nature are subject to the prefix-substitution and colluding attacks. To remedy the above post-IME system's flaws, we propose a new idea, pre-IME, which guarantees that "Is this touch event a sensitive keystroke?" analysis will always access user touch events prior to the execution of any IME app code. We designed an innovative TrustZone-based framework named IM-Visor which has the pre-IME nature. Specifically, IM-Visor creates the isolation environment named STIE as soon as a user intends to type on a soft keyboard, then the STIE intercepts, translates and analyzes the user's touch input. If the input is sensitive, the translation of keystrokes will be delivered to user apps through a trusted path. Otherwise, IM-Visor replays non-sensitive keystroke touch events for IME apps or replays non-keystroke touch events for other apps. A prototype of IM-Visor has been implemented and tested with several most popular IMEs. The experimental results show that IM-Visor has small runtime overheads.

In this paper, we review big data characteristics and security challenges in the cloud and visit different cloud domains and security regulations. We propose using integrated auditing for secure data storage and transaction logs, real-time compliance and security monitoring, regulatory compliance, data environment, identity and access management, infrastructure auditing, availability, privacy, legality, cyber threats, and granular auditing to achieve big data security. We apply a stochastic process model to conduct security analyses in availability and mean time to security failure. Potential future works are also discussed.

The collaborative recommendation mechanism is beneficial for the subject in an open network to find efficiently enough referrers who directly interacted with the object and obtain their trust data. The uncertainty analysis to the collected trust data selects the reliable trust data of trustworthy referrers, and then calculates the statistical trust value on certain reliability for any object. After that the subject can judge its trustworthiness and further make a decision about interaction based on the given threshold. The feasibility of this method is verified by three experiments which are designed to validate the model's ability to fight against malicious service, the exaggeration and slander attack. The interactive success rate is significantly improved by using the new model, and the malicious entities are distinguished more effectively than the comparative model.

HDFS has been widely used for storing massive scale data which is vulnerable to site disaster. The file system backup is an important strategy for data retention. In this paper, we present an efficient, easy- to-use Backup and Disaster Recovery System for HDFS. The system includes a client based on HDFS with additional feature of remote backup, and a remote server with a HDFS cluster to keep the backup data. It supports full backup and regularly incremental backup to the server with very low cost and high throughout. In our experiment, the average speed of backup and recovery is up to 95 MB/s, approaching the theoretical maximum speed of gigabit Ethernet.

To enhance the encryption and anti-translation capability of the information, we constructed a five-dimensional chaotic system. Combined with the Lü system, a time-switched system with multiple chaotic attractors is realized in the form of a digital circuit. Some characteristics of the five-dimensional system are analyzed, such as Poincare mapping, the Lyapunov exponent spectrum, and bifurcation diagram. The analysis shows that the system exhibits chaotic characteristics for a wide range of parameter values. We constructed a time-switched expression between multiple chaotic attractors using the communication between a microcontroller unit (MCU) and field programmable gate array (FPGA). The system can quickly switch between different chaotic attractors within the chaotic system and between chaotic systems at any time, leading to signal sources with more variability, diversity, and complexity for chaotic encryption.

We always use CAPTCHA(Completely Automated Public Turing test to Tell Computers and Humans Apart) to prevent automated bot for data entry. Although there are various kinds of CAPTCHAs, text-based scheme is still applied most widely, because it is one of the most convenient and user-friendly way for daily user [1]. The fact is that segmentations of different types of CAPTCHAs are not always the same, which means one of CAPTCHA's bottleneck is the segmentation. Once we could accurately split the character, the problem could be solved much easier. Unfortunately, the best way to divide them is still case by case, which is to say there is no universal way to achieve it. In this paper, we present a novel algorithm to achieve state-of-the-art performance, what was more, we also constructed a new convolutional neural network as an add-on recognition part to stabilize our state-of-the-art performance of the whole CAPTCHA system. The CAPTCHA datasets we are using is from the State Administration for Industry& Commerce of the People's Republic of China. In this datasets, there are totally 33 entrances of CAPTCHAs. In this experiments, we assume that each of the entrance is known. Results are provided showing how our algorithms work well towards these CAPTCHAs.

The traditional text classification methods usually follow this process: first, a sentence can be considered as a bag of words (BOW), then transformed into sentence feature vector which can be classified by some methods, such as maximum entropy (ME), Naive Bayes (NB), support vector machines (SVM), and so on. However, when these methods are applied to text classification, we usually can not obtain an ideal result. The most important reason is that the semantic relations between words is very important for text categorization, however, the traditional method can not capture it. Sentiment classification, as a special case of text classification, is binary classification (positive or negative). Inspired by the sentiment analysis, we use a novel deep learning-based recurrent neural networks (RNNs)model for automatic security audit of short messages from prisons, which can classify short messages(secure and non-insecure). In this paper, the feature of short messages is extracted by word2vec which captures word order information, and each sentence is mapped to a feature vector. In particular, words with similar meaning are mapped to a similar position in the vector space, and then classified by RNNs. RNNs are now widely used and the network structure of RNNs determines that it can easily process the sequence data. We preprocess short messages, extract typical features from existing security and non-security short messages via word2vec, and classify short messages through RNNs which accept a fixed-sized vector as input and produce a fixed-sized vector as output. The experimental results show that the RNNs model achieves an average 92.7% accuracy which is higher than SVM.

Control theory and SDN (Software Defined Networking) are key components for NFV (Network Function Virtualization) deployment. However little has been done to use a control-theoretic approach for SDN and NFV management. In this demo, we describe a use case for NFV management using control theory and SDN. We use the management architecture of RINA (a clean-slate Recursive InterNetwork Architecture) to manage Virtual Network Function (VNF) instances over the GENI testbed. We deploy Snort, an Intrusion Detection System (IDS) as the VNF. Our network topology has source and destination hosts, multiple IDSes, an Open vSwitch (OVS) and an OpenFlow controller. A distributed management application running on RINA measures the state of the VNF instances and communicates this information to a Proportional Integral (PI) controller, which then provides load balancing information to the OpenFlow controller. The latter controller in turn updates traffic flow forwarding rules on the OVS switch, thus balancing load across the VNF instances. This demo demonstrates the benefits of using such a control-theoretic load balancing approach and the RINA management architecture in virtualized environments for NFV management. It also illustrates that the GENI testbed can easily support a wide range of SDN and NFV related experiments.