Visible to the public Biblio

Found 2435 results

Filters: Keyword is privacy  [Clear All Filters]
2022-01-10
Vast, Rahul, Sawant, Shruti, Thorbole, Aishwarya, Badgujar, Vishal.  2021.  Artificial Intelligence Based Security Orchestration, Automation and Response System. 2021 6th International Conference for Convergence in Technology (I2CT). :1–5.
Cybersecurity is becoming very crucial in the today's world where technology is now not limited to just computers, smartphones, etc. It is slowly entering into things that are used on daily basis like home appliances, automobiles, etc. Thus, opening a new door for people with wrong intent. With the increase in speed of technology dealing with such issues also requires quick response from security people. Thus, dealing with huge variety of devices quickly will require some extent of automation in this field. Generating threat intelligence automatically and also including those which are multilingual will also add plus point to prevent well known major attacks. Here we are proposing an AI based SOAR system in which the data from various sources like firewalls, IDS, etc. is collected with individual event profiling using a deep-learning detection method. For this the very first step is that the collected data from different sources will be converted into a standardized format i.e. to categorize the data collected from different sources. For standardized format Here our system finds out about the true positive alert for which the appropriate/ needful steps will be taken such as the generation of Indicators of Compromise report and the additional evidences with the help of Security Information and Event Management system. The security alerts will be notified to the security teams with the degree of threat.
Ibrahim, Mariam, Nabulsi, Intisar.  2021.  Security Analysis of Smart Home Systems Applying Attack Graph. 2021 Fifth World Conference on Smart Trends in Systems Security and Sustainability (WorldS4). :230–234.
In this work, security analysis of a Smart Home System (SHS) is inspected. The paper focuses on describing common and likely cyber security threats against SHS. This includes both their influence on human privacy and safety. The SHS is properly presented and formed applying Architecture Analysis and Design Language (AADL), exhibiting the system layout, weaknesses, attack practices, besides their requirements and post settings. The obtained model is later inspected along with a security requirement with JKind model tester software for security endangerment. The overall attack graph causing system compromise is graphically given using Graphviz.
Liu, Fuwen, Su, Li, Yang, Bo, Du, Haitao, Qi, Minpeng, He, Shen.  2021.  Security Enhancements to Subscriber Privacy Protection Scheme in 5G Systems. 2021 International Wireless Communications and Mobile Computing (IWCMC). :451–456.
Subscription permanent identifier has been concealed in the 5G systems by using the asymmetric encryption scheme as specified in standard 3GPP TS 33.501 to protect the subscriber privacy. The standardized scheme is however subject to the SUPI guess attack as the public key of the home network is publicly available. Moreover, it lacks the inherent mechanism to prevent SUCI replay attacks. In this paper, we propose three methods to enhance the security of the 3GPP scheme to thwart the SUPI guess attack and replay attack. One of these methods is suggested to be used to strengthen the security of the current subscriber protection scheme.
Setiawan, Fauzan Budi, Magfirawaty.  2021.  Securing Data Communication Through MQTT Protocol with AES-256 Encryption Algorithm CBC Mode on ESP32-Based Smart Homes. 2021 International Conference on Computer System, Information Technology, and Electrical Engineering (COSITE). :166–170.
The Internet of Things (IoT) is a technology that allows connection between devices using the internet to collect and exchange data with each other. Privacy and security have become the most pressing issues in the IoT network, especially in the smart home. Nevertheless, there are still many smart home devices that have not implemented security and privacy policies. This study proposes a remote sensor control system built on ESP32 to implement a smart home through the Message Queuing Telemetry Transport(MQTT) protocol by applying the Advanced Encryption Standard (AES) algorithm with a 256-bit key. It addresses security issues in the smart home by encrypting messages sent from users to sensors. Besides ESP32, the system implementation also uses Raspberry Pi and smartphone with Android applications. The network was analyzed using Wireshark, and it showed that the message sent was encrypted. This implementation could prevent brute force attacks, with the result that it could guarantee the confidentiality of a message. Meanwhile, from several experiments conducted in this study, the difference in the average time of sending encrypted and unencrypted messages was not too significant, i.e., 20 ms.
2021-12-21
Xu, Xiaojun, Wang, Qi, Li, Huichen, Borisov, Nikita, Gunter, Carl A., Li, Bo.  2021.  Detecting AI Trojans Using Meta Neural Analysis. 2021 IEEE Symposium on Security and Privacy (SP). :103–120.
In machine learning Trojan attacks, an adversary trains a corrupted model that obtains good performance on normal data but behaves maliciously on data samples with certain trigger patterns. Several approaches have been proposed to detect such attacks, but they make undesirable assumptions about the attack strategies or require direct access to the trained models, which restricts their utility in practice.This paper addresses these challenges by introducing a Meta Neural Trojan Detection (MNTD) pipeline that does not make assumptions on the attack strategies and only needs black-box access to models. The strategy is to train a meta-classifier that predicts whether a given target model is Trojaned. To train the meta-model without knowledge of the attack strategy, we introduce a technique called jumbo learning that samples a set of Trojaned models following a general distribution. We then dynamically optimize a query set together with the meta-classifier to distinguish between Trojaned and benign models.We evaluate MNTD with experiments on vision, speech, tabular data and natural language text datasets, and against different Trojan attacks such as data poisoning attack, model manipulation attack, and latent attack. We show that MNTD achieves 97% detection AUC score and significantly outperforms existing detection approaches. In addition, MNTD generalizes well and achieves high detection performance against unforeseen attacks. We also propose a robust MNTD pipeline which achieves around 90% detection AUC even when the attacker aims to evade the detection with full knowledge of the system.
Oliver, Ian.  2021.  Trust, Security and Privacy through Remote Attestation in 5G and 6G Systems. 2021 IEEE 4th 5G World Forum (5GWF). :368–373.
Digitalisation of domains such as medical and railway utilising cloud and networking technologies such as 5G and forthcoming 6G systems presents additional security challenges. The establishment of the identity, integrity and provenance of devices, services and other functional components removed a number of attack vectors and addresses a number of so called zero-trust security requirements. The addition of trusted hardware, such as TPM, and related remote attestation integrated with the networking and cloud infrastructure will be necessary requirement.
Xiaojian, Zhang, Liandong, Chen, Jie, Fan, Xiangqun, Wang, Qi, Wang.  2021.  Power IoT Security Protection Architecture Based on Zero Trust Framework. 2021 IEEE 5th International Conference on Cryptography, Security and Privacy (CSP). :166–170.
The construction of the power Internet of Things has led various terminals to access the corporate network on a large scale. The internal and external business interaction and data exchange are more extensive. The current security protection system is based on border isolation protection. This is difficult to meet the needs of the power Internet of Things connection and open shared services. This paper studies the application scheme of the ``zero trust'' typical business scenario of the power Internet of Things with ``Continuous Identity Authentication and Dynamic Access Control'' as the core, and designs the power internet security protection architecture based on zero trust.
Ayed, Mohamed Ali, Talhi, Chamseddine.  2021.  Federated Learning for Anomaly-Based Intrusion Detection. 2021 International Symposium on Networks, Computers and Communications (ISNCC). :1–8.
We are attending a severe zero-day cyber attacks. Machine learning based anomaly detection is definitely the most efficient defence in depth approach. It consists to analyzing the network traffic in order to distinguish the normal behaviour from the abnormal one. This approach is usually implemented in a central server where all the network traffic is analyzed which can rise privacy issues. In fact, with the increasing adoption of Cloud infrastructures, it is important to reduce as much as possible the outsourcing of such sensitive information to the several network nodes. A better approach is to ask each node to analyze its own data and then to exchange its learning finding (model) with a coordinator. In this paper, we investigate the application of federated learning for network-based intrusion detection. Our experiment was conducted based on the C ICIDS2017 dataset. We present a f ederated learning on a deep learning algorithm C NN based on model averaging. It is a self-learning system for detecting anomalies caused by malicious adversaries without human intervention and can cope with new and unknown attacks without decreasing performance. These experimentation demonstrate that this approach is effective in detecting intrusion.
Elumar, Eray Can, Sood, Mansi, Ya\u gan, Osman.  2021.  On the Connectivity and Giant Component Size of Random K-out Graphs Under Randomly Deleted Nodes. 2021 IEEE International Symposium on Information Theory (ISIT). :2572–2577.
Random K-out graphs, denoted \$$\backslash$mathbbH(n;K)\$, are generated by each of the \$n\$ nodes drawing \$K\$ out-edges towards \$K\$ distinct nodes selected uniformly at random, and then ignoring the orientation of the arcs. Recently, random K-out graphs have been used in applications as diverse as random (pairwise) key predistribution in ad-hoc networks, anonymous message routing in crypto-currency networks, and differentially-private federated averaging. In many applications, connectivity of the random K-out graph when some of its nodes are dishonest, have failed, or have been captured is of practical interest. We provide a comprehensive set of results on the connectivity and giant component size of \$$\backslash$mathbbH(n;K\_n,$\backslash$gamma\_n)\$, i.e., random K-out graph when \textsubscriptn of its nodes, selected uniformly at random, are deleted. First, we derive conditions for \textsubscriptn and \$n\$ that ensure, with high probability (whp), the connectivity of the remaining graph when the number of deleted nodes is \$$\backslash$gamma\_n=Ømega(n)\$ and \$$\backslash$gamma\_n=o(n)\$, respectively. Next, we derive conditions for \$$\backslash$mathbbH(n;K\_n, $\backslash$gamma\_n)\$ to have a giant component, i.e., a connected subgraph with \$Ømega(n)\$ nodes, whp. This is also done for different scalings of \textsubscriptn and upper bounds are provided for the number of nodes outside the giant component. Simulation results are presented to validate the usefulness of the results in the finite node regime.
Diamond, Benjamin E..  2021.  Many-out-of-Many Proofs and Applications to Anonymous Zether. 2021 IEEE Symposium on Security and Privacy (SP). :1800–1817.
Anonymous Zether, proposed by Bünz, Agrawal, Zamani, and Boneh (FC'20), is a private payment design whose wallets demand little bandwidth and need not remain online; this unique property makes it a compelling choice for resource-constrained devices. In this work, we describe an efficient construction of Anonymous Zether. Our protocol features proofs which grow only logarithmically in the size of the "anonymity sets" used, improving upon the linear growth attained by prior efforts. It also features competitive transaction sizes in practice (on the order of 3 kilobytes).Our central tool is a new family of extensions to Groth and Kohlweiss's one-out-of-many proofs (Eurocrypt 2015), which efficiently prove statements about many messages among a list of commitments. These extensions prove knowledge of a secret subset of a public list, and assert that the commitments in the subset satisfy certain properties (expressed as linear equations). Remarkably, our communication remains logarithmic; our computation increases only by a logarithmic multiplicative factor. This technique is likely to be of independent interest.We present an open-source, Ethereum-based implementation of our Anonymous Zether construction.
Hamouid, Khaled, Omar, Mawloud, Adi, Kamel.  2021.  A Privacy-Preserving Authentication Model Based on Anonymous Certificates in IoT. 2021 Wireless Days (WD). :1–6.
This paper proposes an anonymity based mechanism for providing privacy in IoT environment. Proposed scheme allows IoT entities to anonymously interacting and authenticating with each other, or even proving that they have trustworthy relationship without disclosing their identities. Authentication is based on an anonymous certificates mechanism where interacting IoT entities could unlinkably prove possession of a valid certificate without revealing any incorporated identity-related information, thereby preserving their privacy and thwarting tracking and profiling attacks. Through a security analysis, we demonstrate the reliability of our solution.
2021-12-20
Abi Sen, Adnan Ahmed, M Alawfi, Ibrahim Moeed, Aloufi, Hazim Faisal, Bahbouh, Nour Mahmoud, Alsaawy, Yazed.  2021.  Comparison among Cooperation, Anonymity and Cloak Area Approaches for Preserving Privacy of IoT. 2021 8th International Conference on Computing for Sustainable Global Development (INDIACom). :413–416.
As a result of the importance of privacy at present, especially with the modern applications and technologies that have spread in the last decade, many techniques and methods have appeared to preserve privacy and protect users' data from tracking, profiling, or identification. The most popular of these technologies are those which rely on peer-to-peer or third-party cooperation. But, by reviewing a significant portion of existing research articles related to privacy, we find considerable confusion amongst several concepts and ways of protection, such as the concept of cloak area, Anonymizer, cooperation, and Third Party Peers (TTP). In this research, we revisit and review these approaches, which contain an overlap between them to distinguish each one clearly with the help of graphs and to remove their ambiguity. In this way, we shall be able provide a ready-reckoner to those interested in this field to easily differentiate between them and thus work to develop them and provide new methods. In other words, this research seeks to enhance the privacy and security in smart applications and technologies in the IoT and smart city environments.
Buccafurri, Francesco, De Angelis, Vincenzo, Idone, Maria Francesca, Labrini, Cecilia.  2021.  Extending Routes in Tor to Achieve Recipient Anonymity against the Global Adversary. 2021 International Conference on Cyberworlds (CW). :238–245.
Tor is a famous routing overlay network based on the Onion multi-layered encryption to support communication anonymity in a threat model in which some network nodes are malicious. However, Tor does not provide any protection against the global passive adversary. In this threat model, an idea to obtain recipient anonymity, which is enough to have relationship anonymity, is to hide the recipient among a sufficiently large anonymity set. However, this would lead to high latency both in the set-up phase (which has a quadratic cost in the number of involved nodes) and in the successive communication. In this paper, we propose a way to arrange a Tor circuit with a tree-like topology, in which the anonymity set consists of all its nodes, whereas set-up and communication latency depends on the number of the sole branch nodes (which is a small fraction of all the nodes). Basically, the cost goes down from quadratic to linear. Anonymity is obtained by applying a broadcast-based technique for the forward message, and cover traffic (generated by the terminal-chain nodes) plus mixing over branch nodes, for the response.
Mygdalis, Vasileios, Tefas, Anastasios, Pitas, Ioannis.  2021.  Introducing K-Anonymity Principles to Adversarial Attacks for Privacy Protection in Image Classification Problems. 2021 IEEE 31st International Workshop on Machine Learning for Signal Processing (MLSP). :1–6.
The network output activation values for a given input can be employed to produce a sorted ranking. Adversarial attacks typically generate the least amount of perturbation required to change the classifier label. In that sense, generated adversarial attack perturbation only affects the output in the 1st sorted ranking position. We argue that meaningful information about the adversarial examples i.e., their original labels, is still encoded in the network output ranking and could potentially be extracted, using rule-based reasoning. To this end, we introduce a novel adversarial attack methodology inspired by the K-anonymity principles, that generates adversarial examples that are not only misclassified, but their output sorted ranking spreads uniformly along K different positions. Any additional perturbation arising from the strength of the proposed objectives, is regularized by a visual similarity-based term. Experimental results denote that the proposed approach achieves the optimization goals inspired by K-anonymity with reduced perturbation as well.
Deng, Han, Wang, Zhechon, Zhang, Yazhen.  2021.  Overview of Privacy Protection Data Release Anonymity Technology. 2021 7th IEEE Intl Conference on Big Data Security on Cloud (BigDataSecurity), IEEE Intl Conference on High Performance and Smart Computing, (HPSC) and IEEE Intl Conference on Intelligent Data and Security (IDS). :151–156.
The collection of digital information by governments, companies and individuals creates tremendous opportunities for knowledge and information-based decision-making. Driven by mutual benefit and laws and regulations, there is a need for data exchange and publication between all parties. However, data in its original form usually contains sensitive information about individuals and publishing such data would violate personal privacy. Privacy Protection Data Distribution (PPDP) provides methods and tools to release useful information while protecting data privacy. In recent years, PPDP has received extensive attention from the research community, and many solutions have been proposed for different data release scenarios. How to ensure the availability of data under the premise of protecting user privacy is the core problem to be solved in this field. This paper studies the existing achievements of privacy protection data release anonymity technology, focusing on the existing anonymity technology in three aspects of high-dimensional, high-deficiency, and complex relational data, and analyzes and summarizes them.
Kohlweiss, Markulf, Madathil, Varun, Nayak, Kartik, Scafuro, Alessandra.  2021.  On the Anonymity Guarantees of Anonymous Proof-of-Stake Protocols. 2021 IEEE Symposium on Security and Privacy (SP). :1818–1833.
In proof-of-stake (PoS) blockchains, stakeholders that extend the chain are selected according to the amount of stake they own. In S&P 2019 the "Ouroboros Crypsinous" system of Kerber et al. (and concurrently Ganesh et al. in EUROCRYPT 2019) presented a mechanism that hides the identity of the stakeholder when adding blocks, hence preserving anonymity of stakeholders both during payment and mining in the Ouroboros blockchain. They focus on anonymizing the messages of the blockchain protocol, but suggest that potential identity leaks from the network-layer can be removed as well by employing anonymous broadcast channels.In this work we show that this intuition is flawed. Even ideal anonymous broadcast channels do not suffice to protect the identity of the stakeholder who proposes a block.We make the following contributions. First, we show a formal network-attack against Ouroboros Crypsinous, where the adversary can leverage network delays to distinguish who is the stakeholder that added a block on the blockchain. Second, we abstract the above attack and show that whenever the adversary has control over the network delay – within the synchrony bound – loss of anonymity is inherent for any protocol that provides liveness guarantees. We do so, by first proving that it is impossible to devise a (deterministic) state-machine replication protocol that achieves basic liveness guarantees and better than (1-2f) anonymity at the same time (where f is the fraction of corrupted parties). We then connect this result to the PoS setting by presenting the tagging and reverse tagging attack that allows an adversary, across several executions of the PoS protocol, to learn the stake of a target node, by simply delaying messages for the target. We demonstrate that our assumption on the delaying power of the adversary is realistic by describing how our attack could be mounted over the Zcash blockchain network (even when Tor is used). We conclude by suggesting approaches that can mitigate such attacks.
Weng, Jui-Hung, Chi, Po-Wen.  2021.  Multi-Level Privacy Preserving K-Anonymity. 2021 16th Asia Joint Conference on Information Security (AsiaJCIS). :61–67.
k-anonymity is a well-known definition of privacy, which guarantees that any person in the released dataset cannot be distinguished from at least k-1 other individuals. In the protection model, the records are anonymized through generalization or suppression with a fixed value of k. Accordingly, each record has the same level of anonymity in the published dataset. However, different people or items usually have inconsistent privacy requirements. Some records need extra protection while others require a relatively low level of privacy constraint. In this paper, we propose Multi-Level Privacy Preserving K-Anonymity, an advanced protection model based on k-anonymity, which divides records into different groups and requires each group to satisfy its respective privacy requirement. Moreover, we present a practical algorithm using clustering techniques to ensure the property. The evaluation on a real-world dataset confirms that the proposed method has the advantages of offering more flexibility in setting privacy parameters and providing higher data utility than traditional k-anonymity.
Buccafurri, Francesco, Angelis, Vincenzo De, Francesca Idone, Maria, Labrini, Cecilia.  2021.  WIP: An Onion-Based Routing Protocol Strengthening Anonymity. 2021 IEEE 22nd International Symposium on a World of Wireless, Mobile and Multimedia Networks (WoWMoM). :231–235.
Anonymous Communication Networks (ACNs) are networks in which, beyond data confidentiality, also traffic flow confidentiality is provided. The most popular routing approach for ACNs also used in practice is Onion. Onion is based on multiple encryption wrapping combined with the proxy mechanism (relay nodes). However, it offers neither sender anonymity nor recipient anonymity in a global passive adversary model, simply because the adversary can observe (at the first relay node) the traffic coming from the sender, and (at the last relay node) the traffic delivered to the recipient. This may also cause a loss of relationship anonymity if timing attacks are performed. This paper presents Onion-Ring, a routing protocol that improves anonymity of Onion in the global adversary model, by achieving sender anonymity and recipient anonymity, and thus relationship anonymity.
Yang, Yuhan, Zhou, Yong, Wang, Ting, Shi, Yuanming.  2021.  Reconfigurable Intelligent Surface Assisted Federated Learning with Privacy Guarantee. 2021 IEEE International Conference on Communications Workshops (ICC Workshops). :1–6.
In this paper, we consider a wireless federated learning (FL) system concerning differential privacy (DP) guarantee, where multiple edge devices collaboratively train a shared model under the coordination of a central base station (BS) through over-the-air computation (AirComp). However, due to the heterogeneity of wireless links, it is difficult to achieve the optimal trade-off between model privacy and accuracy during the FL model aggregation. To address this issue, we propose to utilize the reconfigurable intelligent surface (RIS) technology to mitigate the communication bottleneck in FL by reconfiguring the wireless propagation environment. Specifically, we aim to minimize the model optimality gap while strictly meeting the DP and transmit power constraints. This is achieved by jointly optimizing the device transmit power, artificial noise, and phase shifts at RIS, followed by developing a two-step alternating minimization framework. Simulation results will demonstrate that the proposed RIS-assisted FL model achieves a better trade-off between accuracy and privacy than the benchmarks.
Shen, Cheng, Liu, Tian, Huang, Jun, Tan, Rui.  2021.  When LoRa Meets EMR: Electromagnetic Covert Channels Can Be Super Resilient. 2021 IEEE Symposium on Security and Privacy (SP). :1304–1317.
Due to the low power of electromagnetic radiation (EMR), EM convert channel has been widely considered as a short-range attack that can be easily mitigated by shielding. This paper overturns this common belief by demonstrating how covert EM signals leaked from typical laptops, desktops and servers are decoded from hundreds of meters away, or penetrate aggressive shield previously considered as sufficient to ensure emission security. We achieve this by designing EMLoRa – a super resilient EM covert channel that exploits memory as a LoRa-like radio. EMLoRa represents the first attempt of designing an EM covert channel using state-of-the-art spread spectrum technology. It tackles a set of unique challenges, such as handling complex spectral characteristics of EMR, tolerating signal distortions caused by CPU contention, and preventing adversarial detectors from demodulating covert signals. Experiment results show that EMLoRa boosts communication range by 20x and improves attenuation resilience by up to 53 dB when compared with prior EM covert channels at the same bit rate. By achieving this, EMLoRa allows an attacker to circumvent security perimeter, breach Faraday cage, and localize air-gapped devices in a wide area using just a small number of inexpensive sensors. To countermeasure EMLoRa, we further explore the feasibility of uncovering EMLoRa's signal using energy- and CNN-based detectors. Experiments show that both detectors suffer limited range, allowing EMLoRa to gain a significant range advantage. Our results call for further research on the countermeasure against spread spectrum-based EM covert channels.
Sahay, Rajeev, Brinton, Christopher G., Love, David J..  2021.  Frequency-based Automated Modulation Classification in the Presence of Adversaries. ICC 2021 - IEEE International Conference on Communications. :1–6.
Automatic modulation classification (AMC) aims to improve the efficiency of crowded radio spectrums by automatically predicting the modulation constellation of wireless RF signals. Recent work has demonstrated the ability of deep learning to achieve robust AMC performance using raw in-phase and quadrature (IQ) time samples. Yet, deep learning models are highly susceptible to adversarial interference, which cause intelligent prediction models to misclassify received samples with high confidence. Furthermore, adversarial interference is often transferable, allowing an adversary to attack multiple deep learning models with a single perturbation crafted for a particular classification network. In this work, we present a novel receiver architecture consisting of deep learning models capable of withstanding transferable adversarial interference. Specifically, we show that adversarial attacks crafted to fool models trained on time-domain features are not easily transferable to models trained using frequency-domain features. In this capacity, we demonstrate classification performance improvements greater than 30% on recurrent neural networks (RNNs) and greater than 50% on convolutional neural networks (CNNs). We further demonstrate our frequency feature-based classification models to achieve accuracies greater than 99% in the absence of attacks.
Masuda, Hiroki, Kita, Kentaro, Koizumi, Yuki, Takemasa, Junji, Hasegawa, Toru.  2021.  Model Fragmentation, Shuffle and Aggregation to Mitigate Model Inversion in Federated Learning. 2021 IEEE International Symposium on Local and Metropolitan Area Networks (LANMAN). :1–6.
Federated learning is a privacy-preserving learning system where participants locally update a shared model with their own training data. Despite the advantage that training data are not sent to a server, there is still a risk that a state-of-the-art model inversion attack, which may be conducted by the server, infers training data from the models updated by the participants, referred to as individual models. A solution to prevent such attacks is differential privacy, where each participant adds noise to the individual model before sending it to the server. Differential privacy, however, sacrifices the quality of the shared model in compensation for the fact that participants' training data are not leaked. This paper proposes a federated learning system that is resistant to model inversion attacks without sacrificing the quality of the shared model. The core idea is that each participant divides the individual model into model fragments, shuffles, and aggregates them to prevent adversaries from inferring training data. The other benefit of the proposed system is that the resulting shared model is identical to the shared model generated with the naive federated learning.
Luo, Xinjian, Wu, Yuncheng, Xiao, Xiaokui, Ooi, Beng Chin.  2021.  Feature Inference Attack on Model Predictions in Vertical Federated Learning. 2021 IEEE 37th International Conference on Data Engineering (ICDE). :181–192.
Federated learning (FL) is an emerging paradigm for facilitating multiple organizations' data collaboration without revealing their private data to each other. Recently, vertical FL, where the participating organizations hold the same set of samples but with disjoint features and only one organization owns the labels, has received increased attention. This paper presents several feature inference attack methods to investigate the potential privacy leakages in the model prediction stage of vertical FL. The attack methods consider the most stringent setting that the adversary controls only the trained vertical FL model and the model predictions, relying on no background information of the attack target's data distribution. We first propose two specific attacks on the logistic regression (LR) and decision tree (DT) models, according to individual prediction output. We further design a general attack method based on multiple prediction outputs accumulated by the adversary to handle complex models, such as neural networks (NN) and random forest (RF) models. Experimental evaluations demonstrate the effectiveness of the proposed attacks and highlight the need for designing private mechanisms to protect the prediction outputs in vertical FL.
Nasr, Milad, Songi, Shuang, Thakurta, Abhradeep, Papemoti, Nicolas, Carlin, Nicholas.  2021.  Adversary Instantiation: Lower Bounds for Differentially Private Machine Learning. 2021 IEEE Symposium on Security and Privacy (SP). :866–882.
Differentially private (DP) machine learning allows us to train models on private data while limiting data leakage. DP formalizes this data leakage through a cryptographic game, where an adversary must predict if a model was trained on a dataset D, or a dataset D′ that differs in just one example. If observing the training algorithm does not meaningfully increase the adversary's odds of successfully guessing which dataset the model was trained on, then the algorithm is said to be differentially private. Hence, the purpose of privacy analysis is to upper bound the probability that any adversary could successfully guess which dataset the model was trained on.In our paper, we instantiate this hypothetical adversary in order to establish lower bounds on the probability that this distinguishing game can be won. We use this adversary to evaluate the importance of the adversary capabilities allowed in the privacy analysis of DP training algorithms.For DP-SGD, the most common method for training neural networks with differential privacy, our lower bounds are tight and match the theoretical upper bound. This implies that in order to prove better upper bounds, it will be necessary to make use of additional assumptions. Fortunately, we find that our attacks are significantly weaker when additional (realistic) restrictions are put in place on the adversary's capabilities. Thus, in the practical setting common to many real-world deployments, there is a gap between our lower bounds and the upper bounds provided by the analysis: differential privacy is conservative and adversaries may not be able to leak as much information as suggested by the theoretical bound.
Buccafurri, Francesco, De Angelis, Vincenzo, Idone, Maria Francesca, Labrini, Cecilia.  2021.  A Distributed Location Trusted Service Achieving k-Anonymity against the Global Adversary. 2021 22nd IEEE International Conference on Mobile Data Management (MDM). :133–138.
When location-based services (LBS) are delivered, location data should be protected against honest-but-curious LBS providers, them being quasi-identifiers. One of the existing approaches to achieving this goal is location k-anonymity, which leverages the presence of a trusted party, called location trusted service (LTS), playing the role of anonymizer. A drawback of this approach is that the location trusted service is a single point of failure and traces all the users. Moreover, the protection is completely nullified if a global passive adversary is allowed, able to monitor the flow of messages, as the source of the query can be identified despite location k-anonymity. In this paper, we propose a distributed and hierarchical LTS model, overcoming both the above drawbacks. Moreover, position notification is used as cover traffic to hide queries and multicast is minimally adopted to hide responses, to keep k-anonymity also against the global adversary, thus enabling the possibility that LBS are delivered within social networks.