Biblio

Large-scale parallel applications with complex global data dependencies beyond those of reductions pose significant scalability challenges in an asynchronous runtime system. Internodal challenges include identifying the all-to-all communication of data dependencies among the nodes. Intranodal challenges include gathering together these data dependencies into usable data objects while avoiding data duplication. This paper addresses these challenges within the context of a large-scale, industrial coal boiler simulation using the Uintah asynchronous many-task runtime system on GPU architectures. We show significant reduction in time spent analyzing data dependencies through refinements in our dependency search algorithm. Multiple task graphs are used to eliminate subsequent analysis when task graphs change in predictable and repeatable ways. Using a combined data store and task scheduler redesign reduces data dependency duplication ensuring that problems fit within host and GPU memory. These modifications did not require any changes to application code or sweeping changes to the Uintah runtime system. We report results running on the DOE Titan system on 119K CPU cores and 7.5K GPUs simultaneously. Our solutions can be generalized to other task dependency problems with global dependencies among thousands of nodes which must be processed efficiently at large scale.

As the era of cloud technology arises, more and more people are beginning to migrate their applications and personal data to the cloud. This makes web-based applications an attractive target for cyber-attacks. As a result, web-based applications now need more protections than ever. However, current anomaly-based web attack detection approaches face the difficulties like unsatisfying accuracy and lack of generalization. And the rule-based web attack detection can hardly fight unknown attacks and is relatively easy to bypass. Therefore, we propose a novel deep learning approach to detect anomalous requests. Our approach is to first train two Recurrent Neural Networks (RNNs) with the complicated recurrent unit (LSTM unit or GRU unit) to learn the normal request patterns using only normal requests unsupervisedly and then supervisedly train a neural network classifier which takes the output of RNNs as the input to discriminate between anomalous and normal requests. We tested our model on two datasets and the results showed that our model was competitive with the state-of-the-art. Our approach frees us from feature selection. Also to the best of our knowledge, this is the first time that the RNN is applied on anomaly-based web attack detection systems.

Recent increases in the field of infrastructure has led to the emerging of cloud computing a virtualized computing platform. This technology provides a lot of pros like rapid elasticity, ubiquitous network access and on-demand access etc. Compare to other technologies cloud computing provides many essential services. As the elasticity and scalability increases the chance for vulnerability of the system is also high. There are many known and unknown security risks and challenges present in this environment. In this research an environment is proposed which can handle security issues and deploys various security levels. The system handles the security of various infrastructure like VM and also handles the Dynamic infrastructure request control. One of the key feature of proposed approach is Dual authorization in which all account related data will be authorized by two privileged administrators of the cloud. The auto scalability feature of the cloud is be made secure for on-demand service request handling by providing an on-demand scheduler who will process the on-demand request and assign the required infrastructure. Combining these two approaches provides a secure environment for cloud users as well as handle On-demand Infrastructure request.

Data mining and information extraction from data is a field that has gained relevance in recent years thanks to techniques based on artificial intelligence and use of machine and deep learning. The main aim of the present work is the development of a tool based on a previous behaviour study of security audit tools (oriented to SQL pentesting) with the purpose of creating testing sets capable of performing an accurate detection of a SQL attack. The study is based on the information collected through the generated web server logs in a pentesting laboratory environment. Then, making use of the common extracted patterns from the logs, each attack vector has been classified in risk levels (dangerous attack, normal attack, non-attack, etc.). Finally, a training with the generated data was performed in order to obtain a classifier system that has a variable performance between 97 and 99 percent in positive attack detection. The training data is shared to other servers in order to create a distributed network capable of deciding if a query is an attack or is a real petition and inform to connected clients in order to block the petitions from the attacker's IP.

Differential privacy has emerged as a promising mechanism for privacy-safe data mining. One popular differential privacy mechanism allows researchers to pose queries over a dataset, and adds random noise to all output points to protect privacy. While differential privacy produces useful data in many scenarios, added noise may jeopardize utility for queries posed over small populations or over long-tailed datasets. Gehrke et al. proposed crowd-blending privacy, with random noise added only to those output points where fewer than k individuals (a configurable parameter) contribute to the point in the same manner. This approach has a lower privacy guarantee, but preserves more research utility than differential privacy. We propose an even more liberal privacy goal—commoner privacy—which fuzzes (omits, aggregates or adds noise to) only those output points where an individual's contribution to this point is an outlier. By hiding outliers, our mechanism hides the presence or absence of an individual in a dataset. We propose one mechanism that achieves commoner privacy—interactive k-anonymity. We also discuss query composition and show how we can guarantee privacy via either a pre-sampling step or via query introspection. We implement interactive k-anonymity and query introspection in a system called Patrol for network trace processing. Our evaluation shows that commoner privacy prevents common attacks while preserving orders of magnitude higher research utility than differential privacy, and at least 9-49 times the utility of crowd-blending privacy.

Modern browsers implement different security policies such as the Content Security Policy (CSP), a mechanism designed to mitigate popular web vulnerabilities, and the Same Origin Policy (SOP), a mechanism that governs interactions between resources of web pages. In this work, we describe how CSP may be violated due to the SOP when a page contains an embedded iframe from the same origin. We analyse 1 million pages from 10,000 top Alexa sites and report that at least 31.1% of current CSP-enabled pages are potentially vulnerable to CSP violations. Further considering real-world situations where those pages are involved in same-origin nested browsing contexts, we found that in at least 23.5% of the cases, CSP violations are possible. During our study, we also identified a divergence among browsers implementations in the enforcement of CSP in srcdoc sandboxed iframes, which actually reveals a problem in Gecko-based browsers CSP implementation. To ameliorate the problematic conflicts of the security mechanisms, we discuss measures to avoid CSP violations.

The huge amount of sensory data collected from mobile devices has offered great potentials to promote more significant services based on user data extracted from sensor readings. However, releasing user data could also seriously threaten user privacy. It is possible to directly collect sensitive information from released user data without user permissions. Furthermore, third party users can also infer sensitive information contained in released data in a latent manner by utilizing data mining techniques. In this paper, we formally define these two types of threats as inherent data privacy and latent data privacy and construct a data-sanitization strategy that can optimize the tradeoff between data utility and customized two types of privacy. The key novel idea lies that the developed strategy can combat against powerful third party users with broad knowledge about users and launching optimal inference attacks. We show that our strategy does not reduce the benefit brought by user data much, while sensitive information can still be protected. To the best of our knowledge, this is the first work that preserves both inherent data privacy and latent data privacy.

To facilitate collaboration over sensitive data, we present DataSynthesizer, a tool that takes a sensitive dataset as input and generates a structurally and statistically similar synthetic dataset with strong privacy guarantees. The data owners need not release their data, while potential collaborators can begin developing models and methods with some confidence that their results will work similarly on the real dataset. The distinguishing feature of DataSynthesizer is its usability — the data owner does not have to specify any parameters to start generating and sharing data safely and effectively. DataSynthesizer consists of three high-level modules — DataDescriber, DataGenerator and ModelInspector. The first, DataDescriber, investigates the data types, correlations and distributions of the attributes in the private dataset, and produces a data summary, adding noise to the distributions to preserve privacy. DataGenerator samples from the summary computed by DataDescriber and outputs synthetic data. ModelInspector shows an intuitive description of the data summary that was computed by DataDescriber, allowing the data owner to evaluate the accuracy of the summarization process and adjust any parameters, if desired. We describe DataSynthesizer and illustrate its use in an urban science context, where sharing sensitive, legally encumbered data between agencies and with outside collaborators is reported as the primary obstacle to data-driven governance. The code implementing all parts of this work is publicly available at https://github.com/DataResponsibly/DataSynthesizer.

This study examines the relationship between task communication and relationship communication, and collaboration by exploring the mediating effect of interpersonal trust in a virtual team environment. A theoretical model was developed to examine this relationship where cognitive trust and affective trust are defined as mediation variables between communication and collaboration. The main results of this study show that firstly, there is a significant correlation with a large effect size between communication, trust, and collaboration. Secondly, interpersonal trust plays an important role as a mediator in the relationship between communication and collaboration, especially in relationship communication within virtual teams.

Following the recent adoption of deep neural networks (DNN) accross a wide range of applications, adversarial attacks against these models have proven to be an indisputable threat. Adversarial samples are crafted with a deliberate intention of undermining a system. In the case of DNNs, the lack of better understanding of their working has prevented the development of efficient defenses. In this paper, we propose a new defense method based on practical observations which is easy to integrate into models and performs better than state-of-the-art defenses. Our proposed solution is meant to reinforce the structure of a DNN, making its prediction more stable and less likely to be fooled by adversarial samples. We conduct an extensive experimental study proving the efficiency of our method against multiple attacks, comparing it to numerous defenses, both in white-box and black-box setups. Additionally, the implementation of our method brings almost no overhead to the training procedure, while maintaining the prediction performance of the original model on clean samples.

The globalization of integrated circuit (IC) supply chain and the emergence of threats, such as intellectual property (IP) piracy, reverse engineering, and hardware Trojans, have forced semiconductor companies to revisit the trust in the supply chain. Logic locking is emerging as a popular and effective countermeasure against these threats. Over the years, multiple logic techniques have been developed. Moreover, a number of attacks have been proposed that expose the security vulnerabilities of these techniques. This paper highlights the key developments in the logic locking research and presents a comprehensive literature review of logic locking.

IPv6 Segment Routing is a major IPv6 extension that provides a modern version of source routing that is currently being developed within the Internet Engineering Task Force (IETF). We propose the first open-source implementation of IPv6 Segment Routing in the Linux kernel. We first describe it in details and explain how it can be used on both endhosts and routers. We then evaluate and compare its performance with plain IPv6 packet forwarding in a lab environment. Our measurements indicate that the performance penalty of inserting IPv6 Segment Routing Headers or encapsulating packets is limited to less than 15%. On the other hand, the optional HMAC security feature of IPv6 Segment Routing is costly in a pure software implementation. Since our implementation has been included in the official Linux 4.10 kernel, we expect that it will be extended by other researchers for new use cases.

The interconnectivity of 6LoWPAN networks with the Internet raises serious security concerns, as constrained 6LoWPAN devices are accessible anywhere from the untrusted global Internet. Also, 6LoWPAN devices are mostly deployed in unattended environments, hence easy to capture and clone. Despite that state of the art crypto solutions provide information security, IPv6 enabled smart objects are vulnerable to attacks from outside and inside 6LoWPAN networks that are aimed to disrupt networks. This paper attempts to identify intrusions aimed to disrupt the Routing Protocol for Low-Power and Lossy Networks (RPL).In order to improve the security within 6LoWPAN networks, we extend SVELTE, an intrusion detection system for the Internet of Things, with an intrusion detection module that uses the ETX (Expected Transmissions) metric. In RPL, ETX is a link reliability metric and monitoring the ETX value can prevent an intruder from actively engaging 6LoWPAN nodes in malicious activities. We also propose geographic hints to identify malicious nodes that conduct attacks against ETX-based networks. We implement these extensions in the Contiki OS and evaluate them using the Cooja simulator.

The rapid deployment of IoT systems on the public Internet is not without concerns for the security and privacy of consumers. Security in IoT systems is often poorly engineered and engineering for privacy does notseemtobea concern for vendors at all. Thecombination of poor security hygiene and access to valuable knowledge renders IoT systems a much-sought target for attacks. IoT systems are not only Internet-accessible but also play the role of servers according to the established client-server communication model and are thus configured with static and/or easily predictable IPv6 addresses, rendering them an easy target for attacks. We present 6HOP, a novel addressing scheme for IoT devices. Our proposal is lightweight in operation, requires minimal administration overhead, and defends against reconnaissance attacks, address based correlation as well as denial-of-service attacks. 6HOP therefore exploits the ample address space available in IPv6 networks and provides effective protection this way.

Field Programmable Gate Arrays (FPGAs) are being increasingly deployed in diverse applications including the emerging Internet of Things (IoT), biomedical, and automotive systems. However, security of the FPGA configuration file (i.e. bitstream), especially during in-field reconfiguration, as well as effective safeguards against unauthorized tampering and piracy during operation, are notably lacking. The current practice of bitstreram encryption is only available in high-end FPGAs, incurs unacceptably high overhead for area/energy-constrained devices, and is susceptible to side channel attacks. In this paper, we present a fundamentally different and novel approach to FPGA security that can protect against all major attacks on FPGA, namely, unauthorized in-field reprogramming, piracy of FPGA intellectual property (IP) blocks, and targeted malicious modification of the bitstream. Our approach employs the security through diversity principle to FPGA, which is often used in the software domain. We make each device architecturally different from the others using both physical (static) and logical (time-varying) configuration keys, ensuring that attackers cannot use a priori knowledge about one device to mount an attack on another. It therefore mitigates the economic motivation for attackers to reverse engineering the bitstream and IP. The approach is compatible with modern remote upgrade techniques, and requires only small modifications to existing FPGA tool flows, making it an attractive addition to the FPGA security suite. Our experimental results show that the proposed approach achieves provably high security against tampering and piracy with worst-case 14% latency overhead and 13% area overhead.

Internet Protocol version 6 (IPv6) over Low-power Wireless Personal Area Networks (6LoWPAN) is extensively used in wireless sensor networks due to its capability to transmit IPv6 packets with low bandwidth and limited resources. 6LoWPAN has several operations in each layer. Most existing security challenges are focused on the network layer, which is represented by the Routing Protocol for Low-power and Lossy Networks (RPL). 6LoWPAN, with its routing protocol (RPL), usually uses nodes that have constrained resources (memory, power, and processor). In addition, RPL messages are exchanged among network nodes without any message authentication mechanism, thereby exposing the RPL to various attacks that may lead to network disruptions. A sinkhole attack utilizes the vulnerabilities in an RPL and attracts considerable traffic by advertising falsified data that change the routing preference for other nodes. This paper proposes the neighbor-passive monitoring technique (NPMT) for detecting sinkhole attacks in RPL-based networks. The proposed technique is evaluated using the COOJA simulator in terms of power consumption and detection accuracy. Moreover, NPMT is compared with popular detection mechanisms.

The increase of sensitive data in the current Internet of Things (IoT) raises demands of computation, communication and storage capabilities. Indeed, thanks to RFID tags and wireless sensor networks, anything can be part of IoT. As a result, a large amount of data is generated, which is hard for many IoT devices to handle, as many IoT devices are resource-constrained and cannot use the existing standard security protocols. Cloud computing might seem like a convenient solution, since it offers on-demand access to a shared pool of resources such as processors, storage, applications and services. However this comes as a cost, as unnecessary communications not only burden the core network, but also the data center in the cloud. Therefore, considering suitable approaches such as fog computing and security middleware solutions is crucial. In this paper, we propose a novel middleware architecture to solve the above issues, and discuss the generic concept of using fog computing along with cloud in order to achieve a higher security level. Our security middleware acts as a smart gateway as it is meant to pre-process data at the edge of the network. Depending on the received information, data might either be processed and stored locally on fog or sent to the cloud for further processing. Moreover, in our scheme, IoT constrained devices communicate through the proposed middleware, which provide access to more computing power and enhanced capability to perform secure communications. We discuss these concepts in detail, and explain how our proposal is effective to cope with some of the most relevant IoT security challenges.

Mobile cloud computing paradigm enables cloud servers to extend the limited hardware resources of mobile devices improving availability and reliability of the services provided. Consequently, private, financial, business and critical data pass through wireless access media exposed to malicious attacks. Mobile cloud infrastructure requires new security mechanisms, at the same time as offloading operations need to maintain the advantages of saving processing and energy of the device. Thus, this paper implements a middleware-based computational offloading with cryptographic algorithms and evaluates two mechanisms (symmetric and asymmetric), to provide the integrity and authenticity of data that a smartphone offloads to mobile cloud servers. Also, the paper discusses the factors that impact on power consumption and performance on smartphones that's run resource-intensive applications.

Apps written in JavaScript are an easy target for reverse engineering attacks, e.g. to steal the intellectual property or to create a clone of an app. Unprotected JavaScript apps even contain high level information such as developer comments, if those were not explicitly stripped. This fact becomes more and more important with the increasing popularity of JavaScript as language of choice for both web development and hybrid mobile apps. In this paper, we present a novel JavaScript obfuscator based on the Google Closure Compiler, which transforms readable JavaScript source code into a representation much harder to analyze for adversaries. We evaluate this obfuscator regarding its performance impact and its semantics-preserving property.

With the coming of the age of big data, efforts to institutionalize de-identification of personal information to protect privacy but also at the same time, to allow the use of personal information, have been actively carried out and already, many countries are in the stage of implementing and establishing de-identification policies quite actively. But even with such efforts to protect and use personal information at the same time, the danger posed by re-identification based on de-identified information is real enough to warrant serious consideration for a management mechanism of such risks as well as a mechanism for distributing the responsibilities and liabilities that follow these risks in the event of accidents and incidents involving the invasion of privacy. So far, most countries implementing the de-identification policies are focusing on defining what de-identification is and the exemption requirements to allow free use of de-identified personal information; in fact, it seems that there is a lack of discussion and consideration on how to distribute the responsibility of the risks and liabilities involved in the process of de-identification of personal information. This study proposes to take a look at the various de-identification policies worldwide and contemplate on these policies in the perspective of risk-liability theory. Also, the constituencies of the de-identification policies will be identified in order to analyze the roles and responsibilities of each of these constituencies thereby providing the theoretical basis on which to initiate the discussions on the distribution of burden and responsibilities arising from the de-identification policies.

With Internet of Things (IoT) middleware solutions moving towards cloud computing, the problems of trust in cloud platforms and data privacy need to be solved. The emergence of Trusted Execution Environments (TEEs) opens new perspectives to increase security in cloud applications. We propose a privacy-preserving IoT middleware, using Intel Software Guard Extensions (SGX) to create a secure system on untrusted platforms. An encrypted index is used as a database and communication with the application is protected using asymmetric encryption. This set of measures allows our system to process events in an orchestration engine without revealing data to the hosting cloud platform.

The growing adoption of distributed data processing frameworks in a wide diversity of application domains challenges end-to-end integration of properties like security, in particular when considering deployments in the context of large-scale clusters or multi-tenant Cloud infrastructures. This paper therefore introduces SecureStreams, a reactive middleware framework to deploy and process secure streams at scale. Its design combines the high-level reactive dataflow programming paradigm with Intel®'s low-level software guard extensions (SGX) in order to guarantee privacy and integrity of the processed data. The experimental results of SecureStreams are promising: while offering a fluent scripting language based on Lua, our middleware delivers high processing throughput, thus enabling developers to implement secure processing pipelines in just few lines of code.

Multileaved comparison methods generalize interleaved comparison methods to provide a scalable approach for comparing ranking systems based on regular user interactions. Such methods enable the increasingly rapid research and development of search engines. However, existing multileaved comparison methods that provide reliable outcomes do so by degrading the user experience during evaluation. Conversely, current multileaved comparison methods that maintain the user experience cannot guarantee correctness. Our contribution is two-fold. First, we propose a theoretical framework for systematically comparing multileaved comparison methods using the notions of considerateness, which concerns maintaining the user experience, and fidelity, which concerns reliable correct outcomes. Second, we introduce a novel multileaved comparison method, Pairwise Preference Multileaving (PPM), that performs comparisons based on document-pair preferences, and prove that it is considerate and has fidelity. We show empirically that, compared to previous multileaved comparison methods, PPM is more sensitive to user preferences and scalable with the number of rankers being compared.

In this paper, we focus on fraud detection on a signed graph with only a small set of labeled training data. We propose a novel framework that combines deep neural networks and spectral graph analysis. In particular, we use the node projection (called as spectral coordinate) in the low dimensional spectral space of the graph's adjacency matrix as the input of deep neural networks. Spectral coordinates in the spectral space capture the most useful topology information of the network. Due to the small dimension of spectral coordinates (compared with the dimension of the adjacency matrix derived from a graph), training deep neural networks becomes feasible. We develop and evaluate two neural networks, deep autoencoder and convolutional neural network, in our fraud detection framework. Experimental results on a real signed graph show that our spectrum based deep neural networks are effective in fraud detection.

The Internet of Things (IoT) creates value by connecting digital processes to the physical world using embedded sensors, actuators and wireless networks. The IoT is increasingly intertwined with critical industrial processes, yet contemporary IoT devices offer limited security features, creating a large new attack surface and inhibiting the adoption of IoT technologies. Hardware security modules address this problem, however, their use increases the cost of embedded IoT devices. Furthermore, millions of IoT devices are already deployed without hardware security support. This paper addresses this problem by introducing a Security MicroVisor (SμV) middleware, which provides memory isolation and custom security operations using software virtualisation and assembly-level code verification. We showcase SμV by implementing a key security feature: remote attestation. Evaluation shows extremely low overhead in terms of memory, performance and battery lifetime for a representative IoT device.