Biblio

Current testing for Deep Neural Networks (DNNs) focuses on quantity of test cases but ignores diversity. To the best of our knowledge, DeepXplore is the first white-box framework for Deep Learning testing by triggering differential behaviors between multiple DNNs and increasing neuron coverage to improve diversity. Since it is based on multiple DNNs facing problems that (1) the framework is not friendly to a single DNN, (2) if incorrect predictions made by all DNNs simultaneously, DeepXplore cannot generate test cases. This paper presents Test4Deep, a white-box testing framework based on a single DNN. Test4Deep avoids mistakes of multiple DNNs by inducing inconsistencies between predicted labels of original inputs and that of generated test inputs. Meanwhile, Test4Deep improves neuron coverage to capture more diversity by attempting to activate more inactivated neurons. The proposed method was evaluated on three popular datasets with nine DNNs. Compared to DeepXplore, Test4Deep produced average 4.59% (maximum 10.49%) more test cases that all found errors and faults of DNNs. These test cases got 19.57% more diversity increment and 25.88% increment of neuron coverage. Test4Deep can further be used to improve the accuracy of DNNs by average up to 5.72% (maximum 7.0%).

Surrogate models have proved to be a suitable replacement for complex simulation models in various applications. Runtime considerations, complexity reduction and privacy concerns play a role in the decision to use a surrogate model. The choice of an appropriate surrogate model though is often tedious and largely dependent on the individual model properties. A tool can help to facilitate this process. To this end, we present a surrogate modeling process supporting tool that simplifies the process of generation and application of surrogate models in a co-simulation framework. We evaluate the tool in our application context, energy system co-simulation, and apply it to different simulation models from that domain with a focus on decentralized energy units.

Cyber-Physical Systems (CPS) are playing important roles in the critical infrastructure now. A prominent family of CPSs are networked control systems in which the control and feedback signals are carried over computer networks like the Internet. Communication over insecure networks make system vulnerable to cyber attacks. In this article, we design an intrusion detection and compensation framework based on system/plant identification to fight covert attacks. We collect error statistics of the output estimation during the learning phase of system operation and after that, monitor the system behavior to see if it significantly deviates from the expected outputs. A compensating controller is further designed to intervene and replace the classic controller once the attack is detected. The proposed model is tested on a DC motor as the plant and is put against a deception signal amplification attack over the forward link. Simulation results show that the detection algorithm well detects the intrusion and the compensator is also successful in alleviating the attack effects.

Style transfer is a research hotspot in computer vision. Up to now, it is still a challenge although many researches have been conducted on it for high quality style transfer. In this work, we propose an algorithm named ASTCNN which is a real-time Arbitrary Style Transfer Convolution Neural Network. The ASTCNN consists of two independent encoders and a decoder. The encoders respectively extract style and content features from style and content and the decoder generates the style transferred image images. Experimental results show that ASTCNN achieves higher quality output image than the state-of-the-art style transfer algorithms and the floating point computation of ASTCNN is 23.3% less than theirs.

In order to solve the problem that there is no effective means to find the optimal number of hidden nodes of single-hidden-layer feedforward neural network, in this paper, a method will be introduced to solve it effectively by using singular value decomposition. First, the training data need to be normalized strictly by attribute-based data normalization and sample-based data normalization. Then, the normalized data is decomposed based on the singular value decomposition, and the number of hidden nodes is determined according to main eigenvalues. The experimental results of MNIST data set and APS data set show that the feedforward neural network can attain satisfactory performance in the classification task.

Advanced persistent threats (APTs) are a particularly troubling challenge for software systems. The adversarial nature of the security domain, and APTs in particular, poses unresolved challenges to the design of self-* systems, such as how to defend against multiple types of attackers with different goals and capabilities. In this interaction, the observability of each side is an important and under-investigated issue in the self-* domain. We propose a model of APT defense that elevates observability as a first-class concern. We evaluate this model by showing how an informed approach that uses observability improves the defender's utility compared to a uniform random strategy, can enable robust planning through sensitivity analysis, and can inform observability-related architectural design decisions.

Advanced persistent threats (APTs) are a particularly troubling challenge for software systems. The adversarial nature of the security domain, and APTs in particular, poses unresolved challenges to the design of self-* systems, such as how to defend against multiple types of attackers with different goals and capabilities. In this interaction, the observability of each side is an important and under-investigated issue in the self-* domain. We propose a model of APT defense that elevates observability as a first-class concern. We evaluate this model by showing how an informed approach that uses observability improves the defender's utility compared to a uniform random strategy, can enable robust planning through sensitivity analysis, and can inform observability-related architectural design decisions.

Many popular online social networks, such as Twitter, Tum-blr, and Sina Weibo, adopt too simple privacy models to satisfy users’diverse needs for privacy protection. In platforms with no (i.e., completely open) or binary (i.e., “public” and “friends-only”) access con-trol, users cannot control the dissemination boundary of the contentthey share. For instance, on Twitter, tweets in “public” accounts areaccessible to everyone including search engines, while tweets in “pro-tected” accounts are visible toallthe followers. In this work, we presentArcanato  enable  fine-grained access control for social network content sharing. In particular, we target the Twitter platform and intro-duce the “private tweet” function, which allows users to disseminateparticular tweets to designated group(s) of followers. Arcana employsCiphertext-Policy Attribute-based Encryption (CP-ABE) to implement social circle detection and private tweet encryption so that  access-controlled  tweets  are  only  readable  by  designated  recipients.  To  bestealthy, Arcana further embeds the protected content as digital water-marks in image tweets. We have implemented the Arcana prototype asa Chrome browser plug-in, and demonstrated its flexibility and effec-tiveness. Different from existing approaches that require trusted third-parties or additional server/broker/mediator, Arcana is light-weight andcompletely transparent to Twitter – all the communications, includingkey distribution and private tweet dissemination, are exchanged as Twit-ter messages. Therefore, with small API modifications, Arcana could beeasily ported to other online social networking platforms to support fine-grained access control.

The Internet of Things (IoT) is a concept that allows the networking of various objects of everyday life and communications on the Internet without human interaction. The IoT consists of Low-Power and Lossy Networks (LLN) which for routing use a special protocol called Routing over Low-Power and Lossy Networks (RPL). Due to the resource-constrained nature of RPL networks, they may be exposed to a variety of internal attacks. Neighbor attack and DIS attack are the specific internal attacks at this protocol. This paper presents an anomaly-based lightweight Intrusion Detection System (IDS) based on threshold values for detecting attacks on the RPL protocol. The results of the simulation using Cooja show that the proposed model has a very high True Positive Rate (TPR) and in some cases, it can be 100%, while the False Positive Rate (FPR) is very low. The results show that the proposed model is fully effective in detecting attacks and applicable to large-scale networks.

Community structure detection in social networks has become a big challenge. Various methods in the literature have been presented to solve this challenge. Recently, several methods have also been proposed to solve this challenge based on a mapping-reduction model, in which data and algorithms are divided between different process nodes so that the complexity of time and memory of community detection in large social networks is reduced. In this paper, a mapping-reduction model is first proposed to detect the structure of communities. Then the proposed framework is rewritten according to a new mechanism called distributed cache memory; distributed cache memory can store different values associated with different keys and, if necessary, put them at different computational nodes. Finally, the proposed rewritten framework has been implemented using SPARK tools and its implementation results have been reported on several major social networks. The performed experiments show the effectiveness of the proposed framework by varying the values of various parameters.

At present, the AE method has the advantages of live measurement, online monitoring and easy fault location, so it is very suitable for insulation defect detection of power equipments such as GIS, etc. In this paper, development of a data processing software for PD acoustic detection based on a general fast analysis algorithm model is introduced. With considering the signal flow chart of current acoustic detection system widely used in operation and maintenance of power system equipments, the main function of the developed PD AE signals analysis software was designed, including the detailed analysis of individual data file, identification with phase compensation based on 2D PRPD histograms, batch processing analysis of data files, management of discharge fingerprint library and display of typical defect discharge data. And all of the corresponding developed software pages are displayed.

Industrial production plants traditionally include sensors for monitoring or documenting processes, and actuators for enabling corrective actions in cases of misconfigurations, failures, or dangerous events. With the advent of the IoT, embedded controllers link these `things' to local networks that often are of low power wireless kind, and are interconnected via gateways to some cloud from the global Internet. Inter-networked sensors and actuators in the industrial IoT form a critical subsystem while frequently operating under harsh conditions. It is currently under debate how to approach inter-networking of critical industrial components in a safe and secure manner.In this paper, we analyze the potentials of ICN for providing a secure and robust networking solution for constrained controllers in industrial safety systems. We showcase hazardous gas sensing in widespread industrial environments, such as refineries, and compare with IP-based approaches such as CoAP and MQTT. Our findings indicate that the content-centric security model, as well as enhanced DoS resistance are important arguments for deploying Information Centric Networking in a safety-critical industrial IoT. Evaluation of the crypto efforts on the RIOT operating system for content security reveal its feasibility for common deployment scenarios.

Network Function Virtualization (NFV) is a novel paradigm which enables the deployment of network functions on commodity hardware. As such, it also stands for a deployment en-abler for any novel networking function or networking paradigm such as Named Data Networking (NDN), the most promising solution relying on the Information-Centric Networking (ICN) paradigm. However, dedicated solutions for the security and performance orchestration of such an emerging paradigm are still lacking thus preventing its adoption by network operators. In this paper, we propose a first step toward a content-oriented orchestration whose purpose is to deploy, manage and secure an NDN virtual network. We present the way we leverage the TOSCA standard, using a crafted NDN oriented extension to enable the specification of both deployment and operational behavior requirements of NDN services. We also highlight NDN-related security and performance policies to produce counter-measures against anomalies that can either come from attacks or performance incidents.

This paper proposes a triple-layer, high capacity, message security scheme. The first two layers are of a cryptographic nature, whereas the third layer is of a steganographic nature. In the first layer, AES-128 encryption is performed on the secret message. In the second layer, a chaotic logistic map encryption is applied on the output of the first secure layer to increase the security of the scheme. In the third layer of security, a 2D image steganography technique is performed, where the least significant bit (LSB) -embedding is done according to a zigzag pattern in each of the three color planes of the cover image (i.e. RGB). The distinguishing feature of the proposed scheme is that the secret data is hidden in a zigzag manner that cannot be predicted by a third party. Moreover, our scheme achieves higher values of peak signal to noise ratio (PPSNR), mean square error (MSE), the structural similarity index metric (SSIM), normal cross correlation (NCC) and image fidelity (IF) compared to its counterparts form the literature. In addition, a histogram analysis as well as the high achieved capacity are magnificent indicators for a reliable and high capacity steganographic scheme.

A parallel brute force attack on RC4 algorithm based on FPGA (Field Programmable Gate Array) with an efficient style has been presented. The main idea of this design is to use number of forecast keying methods to reduce the overall clock pulses required depended to key searching operation by utilizes on-chip BRAMs (block RAMs) of FPGA for maximizing the total number of key searching unit with taking into account the highest clock rate. Depending on scheme, 32 key searching units and main controller will be used in one Xilinx XC3S1600E-4 FPGA device, all these units working in parallel and each unit will be searching in a specific range of keys, by comparing the current result with the well-known cipher text if its match the found flag signal will change from 0 to 1 and the main controller will receive this signal and stop the searching operation. This scheme operating at 128-MHz clock frequency and gives us key searching speed of 7.7 × 106 keys/sec. Testing all possible keys (40-bits length), requires only around 39.5h.

The anomaly detection of time series is a hotspot of time series data mining. The own characteristics of different anomaly detectors determine the abnormal data that they are good at. There is no detector can be optimizing in all types of anomalies. Moreover, it still has difficulties in industrial production due to problems such as a single detector can't be optimized at different time windows of the same time series. This paper proposes an adaptive model based on time series characteristics and selecting appropriate detector and run-time parameters for anomaly detection, which is called ATSDLN(Adaptive Time Series Detector Learning Network). We take the time series as the input of the model, and learn the time series representation through FCN. In order to realize the adaptive selection of detectors and run-time parameters according to the input time series, the outputs of FCN are the inputs of two sub-networks: the detector selection network and the run-time parameters selection network. In addition, the way that the variable layer width design of the parameter selection sub-network and the introduction of transfer learning make the model be with more expandability. Through experiments, it is found that ATSDLN can select appropriate anomaly detector and run-time parameters, and have strong expandability, which can quickly transfer. We investigate the performance of ATSDLN in public data sets, our methods outperform other methods in most cases with higher effect and better adaptation. We also show experimental results on public data sets to demonstrate how model structure and transfer learning affect the effectiveness.

How to protect users' private data during network data mining has become a hot issue in the fields of big data and network information security. Most current researches on differential privacy k-means clustering algorithms focus on optimizing the selection of initial centroids. However, the traditional privacy budget allocation has the problem that the random noise becomes too large as the number of iterations increases, which will reduce the performance of data clustering. To solve the problem, we improved the way of privacy budget allocation in differentially private clustering algorithm DPk-means, and proposed APDPk-means, a new differential privacy clustering algorithm based on arithmetic progression privacy budget allocation. APDPk-means decomposes the total privacy budget into a decreasing arithmetic progression, allocating the privacy budgets from large to small in the iterative process, so as to ensure the rapid convergence in early iteration. The experiment results show that compared with the other differentially private k-means algorithms, APDPk-means has better performance in availability and quality of the clustering result under the same level of privacy protection.

Internet of Things is the newfound information architecture based on the Internet that develops interactions between objects and services in a secure and reliable environment. As the availability of many smart devices rises, secure and scalable mass storage systems for aggregate data is required in IoTs applications. In this paper, we propose a new method for storing aggregate data in IoTs by use of ( t, n) -threshold secret sharing scheme in the cloud storage. In this method, original data is divided into t blocks that each block is considered as a share. This method is scalable and traceable, i.e., new data can be inserted or part of original data can be deleted, without changing shares, also cloud service providers' fault in sending invalid shares are detectable.

The IoT industry has developed rapidly in recent years, which has attracted the attention of security researchers. However, the researchers are hampered by the wide variety of IoT device operating systems and their hardware architectures. Especially for the lightweight IoT devices, many manufacturers do not provide the device firmware images, embedded firmware source code or even the develop documents. As a result, it hinders traditional static analysis and dynamic analysis techniques. In this paper, we propose a novel dynamic analysis framework, called FIoT, which aims at finding memory corruption vulnerabilities in lightweight IoT device firmware images. The key idea is dynamically run the binary code snippets through symbolic execution with carrying out a fuzzing test. Specifically, we generate code snippets through traversing the control-flow graph (CFG) in a backward manner. We improved the CFG recovery approach and backward slice approach for better performance. To reduce the influence of the binary firmware, FIoT leverages loading address determination analysis and library function identification approach. We have implemented a prototype of FIoT and conducted experiments. Our results show that FIoT can complete the Fuzzing test within 40 seconds in average. Considering 170 seconds for static analysis, FIoT can load and analyze a lightweight IoT firmware within 210 seconds in total. Furthermore, we illustrate the effectiveness of FIoT by applying it over 115 firmware images from 17 manufacturers. We have found 35 images exist memory corruptions, which are all zero-day vulnerabilities.

To increase computation throughput, general purpose Graphics Processing Units (GPUs) have been leveraged to accelerate computationally intensive workloads. GPUs have been used as cryptographic engines, improving encryption/decryption throughput and leveraging the GPU's Single Instruction Multiple Thread (SIMT) model. RSA is a widely used public-key cipher and has been ported onto GPUs for signing and decrypting large files. Although performance has been significantly improved, the security of RSA on GPUs is vulnerable to side-channel timing attacks and is an exposure overlooked in previous studies. GPUs tend to be naturally resilient to side-channel attacks, given that they execute a large number of concurrent threads, performing many RSA operations on different data in parallel. Given the degree of parallel execution on a GPU, there will be a significant amount of noise introduced into the timing channel given the thousands of concurrent threads executing concurrently. In this work, we build a timing model to capture the parallel characteristics of an RSA public-key cipher implemented on a GPU. We consider optimizations that include using Montgomery multiplication and sliding-window exponentiation to implement cryptographic operations. Our timing model considers the challenges of parallel execution, complications that do not occur in single-threaded computing platforms. Based on our timing model, we launch successful timing attacks on RSA running on a GPU, extracting the private key of RSA. We also present an effective error detection and correction mechanism. Our results demonstrate that GPU acceleration of RSA is vulnerable to side-channel timing attacks. We propose several countermeasures to defend against this class of attacks.

Trust in human-automation interaction is increasingly imperative as AI and robots become ubiquitous at home, school, and work. Interdependence theory allows for the identification of one-on-one interactions that require trust by analyzing the structure of the potential outcomes. This paper synthesizes multiple, formerly disparate research approaches by extending Interdependence theory to create a unified framework for outcome-based trust in human-automation interaction. This framework quantitatively contextualizes validated empirical results from social psychology on relationship formation, stability, and betrayal. It also contributes insights into trust-related concepts, such as power and commitment, which help further our understanding of trustworthy system design. This new integrated interactional approach reveals how trust and trustworthiness machines from merely reliable tools to trusted teammates working hand-in-actuator toward an automated future.

Entropy sources are designed to provide unpredictable random numbers for cryptographic systems. As an assessment of the sources, Shannon entropy is usually adopted to quantitatively measure the unpredictability of the outputs. In several related works about the entropy evaluation of ring oscillator-based (RO-based) entropy sources, authors evaluated the unpredictability with the average conditional Shannon entropy (ACE) of the source, moreover provided a lower bound of the ACE (LBoACE). However, in this paper, we have demonstrated that when the adversaries have access to the history outputs of the entropy source, for example, by some intrusive attacks, the LBoACE may overestimate the actual unpredictability of the next output for the adversaries. In this situation, we suggest to adopt the specific conditional Shannon entropy (SCE) which exactly measures the unpredictability of the future output with the knowledge of previous output sequences and so is more consistent with the reality than the ACE. In particular, to be conservative, we propose to take the lower bound of the SCE (LBoSCE) as an estimation of the worst-case entropy of the sources. We put forward a detailed method to estimate this worst-case entropy of RO-based entropy sources, which we have also verified by experiment on an FPGA device. We recommend to adopt this method to provide a conservative assessment of the unpredictability when the entropy source works in a vulnerable environment and the adversaries might obtain the previous outputs.

Expected and unexpected risks in cloud computing, which included data security, data segregation, and the lack of control and knowledge, have led to some dilemmas in several fields. Among all of these dilemmas, the privacy problem is even more paramount, which has largely constrained the prevalence and development of cloud computing. There are several privacy protection algorithms proposed nowadays, which generally include two categories, Anonymity algorithm, and differential privacy mechanism. Since many types of research have already focused on the efficiency of the algorithms, few of them emphasized the different orientation and demerits between the two algorithms. Motivated by this emerging research challenge, we have conducted a comprehensive survey on the two popular privacy protection algorithms, namely K-Anonymity Algorithm and Differential Privacy Algorithm. Based on their principles, implementations, and algorithm orientations, we have done the evaluations of these two algorithms. Several expectations and comparisons are also conducted based on the current cloud computing privacy environment and its future requirements.

The usage of robot is rapidly growth in our society. The communication link and applications connect the robots to their clients or users. This communication link and applications are normally connected through some kind of network connections. This network system is amenable of being attached and vulnerable to the security threats. It is a critical part for ensuring security and privacy for robotic platforms. The paper, also discusses about several cyber-physical security threats that are only for robotic platforms. The peer to peer applications use in the robotic platforms for threats target integrity, availability and confidential security purposes. A Remote Administration Tool (RAT) was introduced for specific security attacks. An impact oriented process was performed for analyzing the assessment outcomes of the attacks. Tests and experiments of attacks were performed in simulation environment which was based on Gazbo Turtlebot simulator and physically on the robot. A software tool was used for simulating, debugging and experimenting on ROS platform. Integrity attacks performed for modifying commands and manipulated the robot behavior. Availability attacks were affected for Denial-of-Service (DoS) and the robot was not listened to Turtlebot commands. Integrity and availability attacks resulted sensitive information on the robot.