Biblio

Privacy issues in recommender systems have attracted the attention of researchers for many years. So far, a number of solutions have been proposed. Unfortunately, most of them are far from practical as they either downgrade the utility or are very inefficient. In this paper, we aim at a more practical solution, by proposing a privacy-preserving hybrid recommender system which consists of an incremental matrix factorization (IMF) component and a user-based collaborative filtering (UCF) component. The IMF component provides the fundamental utility while it allows the service provider to efficiently learn feature vectors in plaintext domain, and the UCF component improves the utility while allows users to carry out their computations in an offline manner. Leveraging somewhat homomorphic encryption (SWHE) schemes, we provide privacy-preserving candidate instantiations for both components. Our experiments demonstrate that the hybrid solution is much more efficient than existing solutions.

Machine learning algorithms based on deep neural networks (NN) have achieved remarkable results and are being extensively used in different domains. On the other hand, with increasing growth of cloud services, several Machine Learning as a Service (MLaaS) are offered where training and deploying machine learning models are performed on cloud providers' infrastructure. However, machine learning algorithms require access to raw data which is often privacy sensitive and can create potential security and privacy risks. To address this issue, we develop new techniques to provide solutions for applying deep neural network algorithms to the encrypted data. In this paper, we show that it is feasible and practical to train neural networks using encrypted data and to make encrypted predictions, and also return the predictions in an encrypted form. We demonstrate applicability of the proposed techniques and evaluate its performance. The empirical results show that it provides accurate privacy-preserving training and classification.

We propose a privacy-preserving framework for learning visual classifiers by leveraging distributed private image data. This framework is designed to aggregate multiple classifiers updated locally using private data and to ensure that no private information about the data is exposed during and after its learning procedure. We utilize a homomorphic cryptosystem that can aggregate the local classifiers while they are encrypted and thus kept secret. To overcome the high computational cost of homomorphic encryption of high-dimensional classifiers, we (1) impose sparsity constraints on local classifier updates and (2) propose a novel efficient encryption scheme named doublypermuted homomorphic encryption (DPHE) which is tailored to sparse high-dimensional data. DPHE (i) decomposes sparse data into its constituent non-zero values and their corresponding support indices, (ii) applies homomorphic encryption only to the non-zero values, and (iii) employs double permutations on the support indices to make them secret. Our experimental evaluation on several public datasets shows that the proposed approach achieves comparable performance against state-of-the-art visual recognition methods while preserving privacy and significantly outperforms other privacy-preserving methods.

Genetic data are important dataset utilised in genetic epidemiology to investigate biologically coded information within the human genome. Enormous research has been delved into in recent years in order to fully sequence and understand the genome. Personalised medicine, patient response to treatments and relationships between specific genes and certain characteristics such as phenotypes and diseases, are positive impacts of studying the genome, just to mention a few. The sensitivity, longevity and non-modifiable nature of genetic data make it even more interesting, consequently, the security and privacy for the storage and processing of genomic data beg for attention. A common activity carried out by geneticists is the association analysis between allele-allele, or even a genetic locus and a disease. We demonstrate the use of cryptographic techniques such as homomorphic encryption schemes and multiparty computations, how such analysis can be carried out in a privacy friendly manner. We compute a 3 × 3 contingency table, and then, genome analyses algorithms such as linkage disequilibrium (LD) measures, all on the encrypted domain. Our computation guarantees privacy of the genome data under our security settings, and provides up to 98.4% improvement, compared to an existing solution.

As the key component of the smart grid, smart meters fill in the gap between electrical utilities and household users. Todays smart meters are capable of collecting household power information in real-time, providing precise power dispatching control services for electrical utilities and informing real-time power price for users, which significantly improve the user experiences. However, the use of data also brings a concern about privacy leakage and the trade-off between data usability and user privacy becomes an vital problem. Existing works propose privacy-utility trade-off frameworks against statistical inference attack. However, these algorithms are basing on distorted data, and will produce cumulative errors when tracing household power usage and lead to false power state estimation, mislead dispatching control, and become an obstacle for practical application. Furthermore, previous works consider power usage as discrete variables in their optimization problems while realistic smart meter data is continuous variable. In this paper, we propose a mechanism to estimate the trade-off between utility and privacy on a continuous time-series distorted dataset, where we extend previous optimization problems to continuous variables version. Experiments results on smart meter dataset reveal that the proposed mechanism is able to prevent inference to sensitive appliances, preserve insensitive appliances, as well as permit electrical utilities to trace household power usage periodically efficiently.

Vehicular Ad-Hoc Networks (VANET) are the creation of several vehicles communicating with each other in order to create a network capable of communication and data exchange. One of the most promising methods for security and trust amongst vehicular networks is the usage of Public Key Infrastructure (PKI). However, current implementations of PKI as a security solution for determining the validity and authenticity of vehicles in a VANET is not efficient due to the usage of large amounts of delay and computational overhead. In this paper, we investigate the potential of PKI when predictively and preemptively passing along certificates to roadside units (RSU) in an effort to lower delay and computational overhead in a dynamic environment. We look to accomplish this through utilizing fog computing and propose a new protocol to pass certificates along the projected path.

One of the most critical challenges facing cyber defense nowadays is the complexity of recent released cyber-attacks, which are capable of disrupting critical industries and jeopardizing national economy. In this context, moving beyond common security approaches to make it possible to neutralize and react to security attacks at their early stages, becomes a requisite. We develop in this paper a formal model for the proactive assessment of security damages. We define a network of observer agents capable of observing incomplete information about attacks and affected cyber systems, and generating security observations useful for the identification of ongoing attack scenarios and their evolution in the future. A set of analytics are developed for the generation and management of scenario contexts as a set of measures useful for the proactive assessment of damages in the future, and the launching of countermeasures. A case study is provided to exemplify the proposal.

The increased number of cyber attacks makes the availability of services a major security concern. One common type of cyber threat is distributed denial of service (DDoS). A DDoS attack is aimed at disrupting the legitimate users from accessing the services. It is easier for an insider having legitimate access to the system to deceive any security controls resulting in insider attack. This paper proposes an Early Detection and Isolation Policy (EDIP)to mitigate insider-assisted DDoS attacks. EDIP detects insider among all legitimate clients present in the system at proxy level and isolate it from innocent clients by migrating it to attack proxy. Further an effective algorithm for detection and isolation of insider is developed with the aim of maximizing attack isolation while minimizing disruption to benign clients. In addition, concept of load balancing is used to prevent proxies from getting overloaded.

Since MANETs are infrastructure-less, they heavily use secret sharing techniques to distribute and decentralize the role of a trusted third party, where the MANET secret s is shared among the legitimate nodes using (t, n) threshold secret sharing scheme. For long lived MANETs, the shared secret is periodically updated without changing the MANET secret based on proactive secret sharing using Elliptic Curve Cryptography(ECC). Hence, the adversary trying to learn the secret, needs to gain at-least t partial shares in the same time period. If the time period and the threshold value t are selected properly, proactive verifiable secret sharing can maintain the overall security of the information in long lived MANETs. The conventional cryptographic algorithms are heavy weight, require lot of computation power thus consuming lot of resources. In our proposal we used Elliptic Curve Cryptography to verify commitments as it requires smaller keys compared to existing proactive secret sharing techniques and makes it useful for MANETs, Which are formed of resource constraint devices.

Cyber-physical systems (CPS) are interconnections of heterogeneous hardware and software components (e.g., sensors, actuators, physical systems/processes, computational nodes and controllers, and communication subsystems). Increasing network connectivity of CPS computational nodes facilitates maintenance and on-demand reprogrammability and reduces operator workload. However, such increasing connectivity also raises the potential for cyber-attacks that attempt unauthorized modifications of run-time parameters or control logic in the computational nodes to hamper process stability or performance. In this paper, we analyze the effectiveness of real-time monitoring using digital and analog side channels. While analog side channels might not typically provide sufficient granularity to observe each iteration of a periodic loop in the code in the CPS device, the temporal averaging inherent to side channel sensory modalities enables observation of persistent changes to the contents of a computational loop through their resulting effect on the level of activity of the device. Changes to code can be detected by observing readings from side channel sensors over a period of time. Experimental studies are performed on an ARM-based single board computer.

In this paper, we explore the usage of printed tags to authenticate products. Printed tags are a cheap alternative to RFID and other tag based systems and do not require specialized equipment. Due to the simplistic nature of such printed codes, many security issues like tag impersonation, server impersonation, reader impersonation, replay attacks and denial of service present in RFID based solutions need to be handled differently. We propose a cost-efficient scheme based on static tag based hash chains to address these security threats. We analyze the security characteristics of this scheme and compare it to other product authentication schemes that use RFID tags. Finally, we show that our proposed statically printed QR codes can be at least as secure as RFID tags.

Malware technology makes it difficult for malware analyst to detect same malware files with different obfuscation technique. In this paper we are trying to tackle that problem by analyzing the sequence of system call from an executable file. Malware files which actually are the same should have almost identical or at least a similar sequence of system calls. In this paper, we are going to create a model for each malware class consists of malwares from different families based on its sequence of system calls. Method/algorithm that's used in this paper is profile hidden markov model which is a very well-known tool in the biological informatics field for comparing DNA and protein sequences. Malware classes that we are going to build are trojan and worm class. Accuracy for these classes are pretty high, it's above 90% with also a high false positive rate around 37%.

Most popular Web applications rely on persistent databases based on languages like SQL for declarative specification of data models and the operations that read and modify them. As applications scale up in user base, they often face challenges responding quickly enough to the high volume of requests. A common aid is caching of database results in the application's memory space, taking advantage of program-specific knowledge of which caching schemes are sound and useful, embodied in handwritten modifications that make the program less maintainable. These modifications also require nontrivial reasoning about the read-write dependencies across operations. In this paper, we present a compiler optimization that automatically adds sound SQL caching to Web applications coded in the Ur/Web domain-specific functional language, with no modifications required to source code. We use a custom cache implementation that supports concurrent operations without compromising the transactional semantics of the database abstraction. Through experiments with microbenchmarks and production Ur/Web applications, we show that our optimization in many cases enables an easy doubling or more of an application's throughput, requiring nothing more than passing an extra command-line flag to the compiler.

Due to the increase in design complexity and cost of VLSI chips, a number of design houses outsource manufacturing and import designs in a way to reduce the cost. This results in a decrease of the authenticity and security of the manufactured product. Since product development involves outside sources, circuit designers can not guarantee that their hardware has not been altered. It is often possible that attackers include additional hardware in order to gain privileges over the original circuit or cause damage to the product. These added circuits are called ``Hardware Trojans''. In this paper, we investigate introducing necessary modules needed for detection of hardware Trojans. We also introduce necessary programmable logic fabric that can be used in the implementation of the hardware assertion checkers. Our target is to utilize the provided programable fabric in a System on Chip (SoC) and optimize the hardware assertion to cover the detection of most hardware trojans in each core of the target SoC.

This paper describes the work done to design a SoC platform for real-time on-line pattern search in TCP packets for Deep Packet Inspection (DPI) applications. The platform is based on a Xilinx Zynq programmable SoC and includes an accelerator that implements a pattern search engine that extends the original Boyer-Moore algorithm with timing and logical rules, that produces a very complex set of rules. Also, the platform implements different modes of operation, including SIMD and MISD parallelism, which can be configured on-line. The platform is scalable depending of the analysis requirement up to 8 Gbps. High-Level synthesis and platform based design methodologies have been used to reduce the time to market of the completed system.

This article presents PrOLoc, a localization system that combines partially homomorphic encryption with a new way of structuring the localization problem to enable emcient and accurate computation of a target's location while preserving the privacy of the observers.

Channel state information (CSI) has been recently shown to be useful in performing security attacks in public WiFi environments. By analyzing how CSI is affected by the finger motions, CSI-based attacks can effectively reconstruct text-based passwords and locking patterns. This paper presents WiGuard, a novel system to protect sensitive on-screen gestures in a public place. Our approach carefully exploits the WiFi channel interference to introduce noise into the attacker's CSI measurement to reduce the success rate of the attack. Our approach automatically detects when a CSI-based attack happens. We evaluate our approach by applying it to protect text-based passwords and pattern locks on mobile devices. Experimental results show that our approach is able to reduce the success rate of CSI attacks from 92% to 42% for text-based passwords and from 82% to 22% for pattern lock.

This work concerns distributed consensus algorithms and application to a network intrusion detection system (NIDS) [21]. We consider the problem of defending the system against multiple data falsification attacks (Byzantine attacks), a vulnerability of distributed peer-to-peer consensus algorithms that has not been widely addressed in its practicality. We consider both naive (independent) and colluding attackers. We test three defense strategy implementations, two classified as outlier detection methods and one reputation-based method. We have narrowed our attention to outlier and reputation-based methods because they are relatively light computationally speaking. We have left out control theoretic methods which are likely the most effective methods, however their computational cost increase rapidly with the number of attackers. We compare the efficiency of these three implementations for their computational cost, detection performance, convergence behavior and possible impacts on the intrusion detection accuracy of the NIDS. Tests are performed based on simulations of distributed denial of service attacks using the KSL-KDD data set.

High-accuracy localization is a prerequisite for many wireless applications. To obtain accurate location information, it is often required to share users' positional knowledge and this brings the risk of leaking location information to adversaries during the localization process. This paper develops a theory and algorithms for protecting location secrecy. In particular, we first introduce a location secrecy metric (LSM) for a general measurement model of an eavesdropper. Compared to previous work, the measurement model accounts for parameters such as channel conditions and time offsets in addition to the positions of users. We determine the expression of the LSM for typical scenarios and show how the LSM depends on the capability of an eavesdropper and the quality of the eavesdropper's measurement. Based on the insights gained from the analysis, we consider a case study in wireless localization network and develop an algorithm that diminish the eavesdropper's capabilities by exploiting the reciprocity of channels. Numerical results show that the proposed algorithm can effectively increase the LSM and protect location secrecy.

Data outsourcing in cloud is emerging as a successful paradigm that benefits organizations and enterprises with high-performance, low-cost, scalable data storage and sharing services. However, this paradigm also brings forth new challenges for data confidentiality because the outsourced are not under the physic control of the data owners. The existing schemes to achieve the security and usability goal usually apply encryption to the data before outsourcing them to the storage service providers (SSP), and disclose the decryption keys only to authorized user. They cannot ensure the security of data while operating data in cloud where the third-party services are usually semi-trustworthy, and need lots of time to deal with the data. We construct a privacy data management system appending hierarchical access control called HAC-DMS, which can not only assure security but also save plenty of time when updating data in cloud.

Software Defined Networking (SDN) is a paradigm shift that changes the working principles of IP networks by separating the control logic from routers and switches, and logically centralizing it within a controller. In this architecture the control plane (controller) communicates with the data plane (switches) through a control channel using a standards-compliant protocol, that is, OpenFlow. While having a centralized controller creates an opportunity to monitor and program the entire network, as a side effect, it causes the control plane to become a single point of failure. Denial of service (DoS) attacks or even heavy control traffic conditions can easily become real threats to the proper functioning of the controller, which indirectly detriments the entire network. In this paper, we propose a solution to reduce the control traffic generated primarily during table-miss events. We utilize the buffer\_id feature of the OpenFlow protocol, which has been designed to identify individually buffered packets within a switch, reusing it to identify flows buffered as a series of packets during table-miss, which happens when there is no related rule in the switch flow tables that matches the received packet. Thus, we allow the OpenFlow switch to send only the first packet of a flow to the controller for a table-miss while buffering the rest of the packets in the switch memory until the controller responds or time out occurs. The test results show that OpenFlow traffic is significantly reduced when the proposed method is used.

Today's computing devices keep considerable amounts of sensitive data unencrypted in RAM. When stolen, lost or simply unattended, attackers are capable of accessing the data in RAM with ease. Valuable and possibly classified data falling into the wrongs hands can lead to severe consequences, for instance when disclosed or reused to log in to accounts or to make transactions. We present a lightweight and hardware-independent mechanism to protect confidential data on suspended Linux devices against physical attackers. Our mechanism rapidly encrypts the contents of RAM during suspension and thereby prevents attackers from retrieving confidential data from the device. Existing systems can easily be extended with our mechanism while fully preserving the usability for end users.

In the Content-Centric Networking (CCN) architecture, content confidentiality is treated as an application-layer concern. Data is only encrypted if the producer and consumer agree on a suitable access control policy and enforcement mechanism. In contrast, transport encryption in TCP/IP applications is increasingly opportunistic for better privacy. This type of encryption is woefully lacking in CCN. To that end, we present TRAPS, a protocol to enable transparent packet security and opportunistic encryption for all CCN data. TRAPS builds on the assumption that knowledge of a name gives one access to the corresponding content; otherwise, by design, the content remains encrypted and secure. TRAPS builds on recent advances in memory hard functions and message-locked encryption to protect data in transit. We show that the security of TRAPS is dependent on the distribution of content names and argue that it can be significantly improved if secure sessions are used to transmit small pieces of information from producers to consumers. Our performance assessment indicates TRAPS is capable of providing opportunistic encryption to CCN without significant throughput loss for reasonable packet throughput measurements.

In smart grid, large quantities of data is collected from various applications, such as smart metering substation state monitoring, electric energy data acquisition, and smart home. Big data acquired in smart grid applications is usually sensitive. For instance, in order to dispatch accurately and support the dynamic price, lots of smart meters are installed at user's house to collect the real-time data, but all these collected data are related to user privacy. In this paper, we propose a data aggregation scheme based on secret sharing with fault tolerance in smart grid, which ensures that control center gets the integrated data without revealing user's privacy. Meanwhile, we also consider fault tolerance during the data aggregation. At last, we analyze the security of our scheme and carry out experiments to validate the results.

In vehicular networks, each message is signed by the generating node to ensure accountability for the contents of that message. For privacy reasons, each vehicle uses a collection of certificates, which for accountability reasons are linked at a central authority. One such design is the Security Credential Management System (SCMS) [1], which is the leading credential management system in the US. The SCMS is composed of multiple components, each of which has a different task for key management, which are logically separated. The SCMS is designed to ensure privacy against a single insider compromise, or against outside adversaries. In this paper, we demonstrate that the current SCMS design fails to achieve its design goal, showing that a compromised authority can gain substantial information about certificate linkages. We propose a solution that accommodates threshold-based detection, but uses relabeling and noise to limit the information that can be learned from a single insider adversary. We also analyze our solution using techniques from differential privacy and validate it using traffic-simulator based experiments. Our results show that our proposed solution prevents privacy information leakage against the compromised authority in collusion with outsider attackers.