Powerful Adversary Research Articles

A Most Powerful Adversary

A Most Powerful Adversary

Coded Caching in the Presence of a Wire and a Cache Tapping Adversary of Type II

This paper introduces the notion of cache-tapping into the information theoretic models of coded caching. The wiretap channel II in the presence of multiple receivers equipped with fixed-size cache memories, and an adversary which selects symbols to tap into from cache placement and/or delivery is introduced. The legitimate terminals know neither whether placement, delivery, or both are tapped, nor the positions in which they are tapped. Only the size of the overall tapped set is known. For two receivers and two files, the strong secrecy capacity- the maximum achievable file rate while keeping the overall library strongly secure- is identified. Lower and upper bounds on the strong secrecy file rate are derived when the library has more than two files. Achievability relies on a code design which combines wiretap coding, security embedding codes, one-time pad keys, and coded caching. A genie-aided upper bound, in which the transmitter is provided with user demands before placement, establishes the converse for the two-files case. For more than two files, the upper bound is constructed by three successive channel transformations. Our results establish provable security guarantees against a powerful adversary which optimizes its tapping over both phases of communication in a cache-aided system.

Optimal Feature Manipulation Attacks Against Linear Regression

In this paper, we investigate how to manipulate the coefficients obtained via linear regression by adding carefully designed poisoning data points to the dataset or modify the original data points. Given the energy budget, we first provide the closed-form solution of the optimal poisoning data point when our target is modifying one designated regression coefficient. We then extend the analysis to the more challenging scenario where the attacker aims to change one particular regression coefficient while making others to be changed as small as possible. For this scenario, we introduce a semidefinite relaxation method to design the best attack scheme. Finally, we study a more powerful adversary who can perform a rank-one modification on the feature matrix. We propose an alternating optimization method to find the optimal rank-one modification matrix. Numerical examples are provided to illustrate the analytical results obtained in this paper.

AWA: Adversarial Website Adaptation

One of the most important obligations of privacy-enhancing technologies is to bring confidentiality and privacy to users’ browsing activities on the Internet. The website fingerprinting attack enables a local passive eavesdropper to predict the target user’s browsing activities even she uses anonymous technologies, such as VPNs, IPsec, and Tor. Recently, the growth of deep learning empowers adversaries to conduct the website fingerprinting attack with higher accuracy. In this paper, we propose a new defense against website fingerprinting attack using adversarial deep learning approaches called Adversarial Website Adaptation (AWA). AWA creates a transformer set in each run so that each website has a unique transformer. Each transformer generates adversarial traces to evade the adversary’s classifier. AWA has two versions, including Universal AWA (UAWA) and Non-Universal AWA (NUAWA). Unlike NUAWA, there is no need to access the entire trace of a website in order to generate an adversarial trace in UAWA. We accommodate secret random elements in the training phase of transformers in order for AWA to generate various sets of transformers in each run. We run AWA several times and create multiple sets of transformers. If an adversary and a target user select different sets of transformers, the accuracy of adversary’s classifier is almost 19.52% and 31.94% with almost 22.28% and 26.28% bandwidth overhead in UAWA and NUAWA, respectively. If a more powerful adversary generates adversarial traces through multiple sets of transformers and trains a classifier on them, the accuracy of adversary’s classifier is almost 49.10% and 25.93% with almost 62.52% and 64.33% bandwidth overhead in UAWA and NUAW, respectively.

Asymptotic Privacy Loss Due to Time Series Matching of Dependent Users

The Internet of Things (IoT) promises to improve user utility by tuning applications to a user’s current behavior, but the user’s behavior can be matched to characteristics learned from prior observations to compromise the user’s identity and hence privacy. Our previous work has established the rate at which anonymization must be performed to prevent such matching in a Bayesian setting when faced with a powerful adversary who has extensive knowledge of each user’s past behavior. However, even sophisticated adversaries do not often have such extensive knowledge; hence, in this letter, we turn our attention to an adversary who must learn user behavior from past data traces of limited length under the assumptions that: (i) there exists dependency between data traces of different users; and (ii) the data points of each user are drawn from a normal distribution. Results on the lengths of training sequences and rates of anonymization for the data sequences that result in a loss of user privacy are presented.

A machine learning based attack detection and mitigation using a secure SaaS framework

Software-as-a-service (SaaS) is a license for access to a specific cloud application through the Internet. However, these services are delayed and sometimes totally disrupted because of the Internet’s unavailability that provides opportunities for a lot of attacks. Research into the security of the cloud focuses mainly on prohibiting malicious users from using the cloud to launch attacks, such as those currently done by botnets, including launching a DDoS attack, sending spam, etc. Because SaaS uses computational power from both servers of cloud computing providers and machines of customers, we argue that SaaS may be elaborately exploited in an unprecedented way as an attack vector for botnets. The previous research work has developed a novel framework for detecting the attacks with the aid of a deep learning approach (Doriguzzi-Corin et al., 2020). Although the attack node in the network has been identified, they were not diminished. An enhanced and powerful adversary model is to be provided as a solution to this problem. Hence, this paper intends to develop a novel framework for attack node mitigation using a secure SaaS Framework. The main contribution of this paper provides an attack detection process that takes place in Deep Belief Network (DBN), in which the weight, as well as activation function, are fine-tuned with Median Fitness oriented Sea Lion Optimization algorithm (MFSLnO). If DBN detects an attack node, the control is transferred to a lightweight bait approach that reliably mitigates the most common attack nodes without disrupting regular connections will be deployed. The performance of the proposed work yielded the best results over the traditional models with a packet loss ratio of 16% and throughput of 89%.

INFUSE: Invisible plausibly-deniable file system for NAND flash

Abstract Protecting sensitive data stored on local storage devices e.g., laptops, tablets etc. is essential for privacy. When adversaries are powerful enough to coerce users to reveal encryption keys/passwords, encryption alone becomes insufficient for data protection. Additional mechanisms are required to hide the very presence of sensitive data. Plausibly deniable storage systems (PDS) are designed to defend against such powerful adversaries. Plausible deniability allows a user to deny the existence of certain stored data even when an adversary has access to the storage medium. However, existing plausible deniability solutions leave users at the mercy of adversaries suspicious of their very use. Indeed, it may be difficult to justify the use of a plausible deniability system while claiming that no sensitive data is being hidden. This work introduces INFUSE, a plausibly-deniable file system that hides not only contents but also the evidence that a particular system is being used to hide data. INFUSE is “invisible” (identical layout with standard file system), provides redundancy, handles overwrites, survives data loss, and is secure in the presence of multi-snapshot adversaries. INFUSE is efficient. Public data operations are orders of magnitude faster than existing multi-snapshot resilient PD systems, and only 15% slower than a standard non-PD baseline, and hidden data operations perform comparably to existing systems.

Robust Stochastic Bandit Algorithms under Probabilistic Unbounded Adversarial Attack

The multi-armed bandit formalism has been extensively studied under various attack models, in which an adversary can modify the reward revealed to the player. Previous studies focused on scenarios where the attack value either is bounded at each round or has a vanishing probability of occurrence. These models do not capture powerful adversaries that can catastrophically perturb the revealed reward. This paper investigates the attack model where an adversary attacks with a certain probability at each round, and its attack value can be arbitrary and unbounded if it attacks. Furthermore, the attack value does not necessarily follow a statistical distribution. We propose a novel sample median-based and exploration-aided UCB algorithm (called med-E-UCB) and a median-based ϵ-greedy algorithm (called med-ϵ-greedy). Both of these algorithms are provably robust to the aforementioned attack model. More specifically we show that both algorithms achieve O(log T) pseudo-regret (i.e., the optimal regret without attacks). We also provide a high probability guarantee of O(log T) regret with respect to random rewards and random occurrence of attacks. These bounds are achieved under arbitrary and unbounded reward perturbation as long as the attack probability does not exceed a certain constant threshold. We provide multiple synthetic simulations of the proposed algorithms to verify these claims and showcase the inability of existing techniques to achieve sublinear regret. We also provide experimental results of the algorithm operating in a cognitive radio setting using multiple software-defined radios.

T0RTT: Non-Interactive Immediate Forward-Secret Single-Pass Circuit Construction

Abstract Maintaining privacy on the Internet with the presence of powerful adversaries such as nation-state attackers is a challenging topic, and the Tor project is currently the most important tool to protect against this threat. The circuit construction protocol (CCP) negotiates cryptographic keys for Tor circuits, which overlay TCP/IP by routing Tor cells over n onion routers. The current circuit construction protocol provides strong security guarantees such as forward secrecy by exchanging 𝒪(n 2) messages. For several years it has been an open question if the same strong security guarantees could be achieved with less message overhead, which is desirable because of the inherent latency in overlay networks. Several publications described CCPs which require only 𝒪(n) message exchanges, but significantly reduce the security of the resulting Tor circuit. It was even conjectured that it is impossible to achieve both message complexity 𝒪(n) and forward secrecy immediately after circuit construction (so-called immediate forward secrecy). Inspired by the latest advancements in zero round-trip time key exchange (0-RTT), we present a new CCP protocol Tor 0-RTT (T0RTT). Using modern cryptographic primitives such as puncturable encryption allow to achieve immediate forward secrecy using only 𝒪(n) messages. We implemented these new primitives to give a first indication of possible problems and how to overcome them in order to build practical CCPs with 𝒪(n) messages and immediate forward secrecy in the future.

A Dung Beetle's Victory: The Moral of theLife of Aesop(Vita G)

AbstractTheLife of Aesopis an entertaining yet profound account of Aesop's life dating from the first to second centuriesad. Although it is widely agreed that theLife of Aesopmay be read as a ‘metafable’, there has been, in my view, a widespread and perversely negative interpretation of the supposed moral of this life story: that ‘pride comes before a fall’. This supposed moral is not borne out by the ending, in which Aesop's prophecies of doom prove to be correct, the Delphians are thrice punished for executing Aesop, and Aesop himself achieves everlasting fame as a storyteller. In this paper, I will argue that a more fitting moral for theLife of Aesopis that ‘even the weakest may find a means to avenge a wrong’. This is the moral that accompanies the quintessentially Aesopic fable of the dung beetle, the hare, and the eagle in which a tiny dung beetle triumphs over a powerful adversary. This fable is pointedly narrated by Aesop to the Delphians just before he is put to death. By reading theLife of Aesopas an exposition of this fable, I will demonstrate that Aesop, just like the dung beetle, is not the loser but the ultimate victor.

Evaluation of hydrated extract of phaseolus Vulgaris L. (bean plant) on hypoglycemia and hypolipidemic in streptozotocin-induced diabetic albino Wistar rats

The current research was intended to comprehend hypoglycemic and anti-lipidaemic exercises of hydrated common bean (phaseolus Vulgaris L.) seed extracts on streptozotocin-induced diabetic albino rats. At a set portion fluctuate of 100, 200,300 mg/kg body weight of common bean extracts was orally directed as one portion for every day to polygenic disorder rats for a measure of thirty days. The impact of P.vulgaris L. on hypoglycemic, glycosylated hemoprotein (HbA1c) and blood serum lipid profile (Total cholesterin), Triglyceride (TG), very low-density lipoprotein (VLDL), low-density lipoprotein (LDL), High-density lipoprotein (HDL)) in plasma were estimated in the regular and diabetic induced rat. The outcomes demonstrated that quick glucose,serum TC, TG, LDL, VLDL, levels were significantly (p<0.05) attenuate, while blood serum HDL, the level was extensively (p<0.05) upgraded inside the diabetic rats. The inconclusive amount of pace of 300 mg/kg is more reasonable than that of a hundred mg/kg. Our examination so shows that Phaseolus vulgaris L has a powerful adversary to diabetic and anti-lipidaemic impacts on streptozotocin-induced diabetic rats, and results were comparable to reference drug glibenclamide.

Stronger trust and privacy in social networks via local cooperation1

AbstractIn our article, we present several protocols that allow to efficiently construct large groups of users based only on local relations of trust. What is more, our approach proves to need only very small computational and communication overhead. Moreover, we give guarantees that a trusted core of the network is defended, even facing a powerful adversary capable of controlling a vast majority of users. This is non-trivial property in real-life networks, as those are usually modelled using preferential attachment graphs, which are extremely prone to attacks on the hub nodes. We show that using our protocols we can achieve similar robustness as Erdős–Renyí graphs, which, on the contrary, are very resistant against attacks focused on chosen nodes. Our protocols have been tested on graphs representing real-world social networks using high performance computing due to the size of the networks. In addition for some protocols, we provided a formal analysis to prove some phenomena in random graphs following power-law distribution, which we use as a network model. Finally, we explicitly demonstrate how our approach can be used to amplify security offered by some privacy-preserving protocols. We believe however that our results can be also seen as a contribution to fundamental observation about the nature of social networks. These results may help to design protocols, whenever it is necessary to gather a big group of users in highly dynamic or even adversarial settings.

TROS: Protecting Humanoids ROS from Privileged Attackers

The spread adoption of humanoid social robots in different application fields is growing the interest of hackers who could violate the privacy of people, or—even worse—threaten humans’ life from physical and emotional/social point of views. Different vectors of attack exist, which are more easily exploitable if physical access to the target robot is available. This is very likely for humanoids that typically reside in untrusted environments where physically access to the robot is allowed and expected, thus permitting anyone to exploit the Linux kernel vulnerability (e.g., through the insertion of a USB pen drive) with the objective of tampering sensitive data. The Robot Operating System (ROS) is at the core of humanoids. Thus, it is crucial for their security. The most-recent solutions introduced in ROS2, SROS, and H-ROS are not sufficient for facing powerful adversaries. In this paper, we first identify the uncovered ROS weaknesses, which are particularly worrying in the case of humanoids. Then, we present our patched ROS solution called Trusted-ROS (TROS) leveraging hardware-assisted trusted computing to shield data managed by ROS, which otherwise would reside in robot’s memory unencrypted. The design of TROS is reported, together with a prototype implementation using a simulated version of the NAO humanoid secured through an Intel SGX hardware. Finally, we evaluated the proposed solution from both security and performance perspectives in order to demonstrate the practicability of our approach.

On the Adversarial Robustness of Subspace Learning

In this paper, we study the adversarial robustness of subspace learning problems. Different from the assumptions made in existing work on robust subspace learning where data samples are contaminated by gross sparse outliers or small dense noises, we consider a more powerful adversary who can first observe the data matrix and then intentionally modify the whole data matrix. We first characterize the optimal rank-one attack strategy that maximizes the subspace distance between the subspace learned from the original data matrix and that learned from the modified data matrix. We then generalize the study to the scenario without the rank constraint and characterize the corresponding optimal attack strategy. Our analysis shows that the optimal strategies depend on the singular values of the original data matrix and the adversary's energy budget. Finally, we provide numerical experiments and practical applications to demonstrate the efficiency of the attack strategies.

A Probabilistic Source Location Privacy Protection Scheme in Wireless Sensor Networks

With the recent developments of wireless sensor networks (WSNs), computing and communication have experienced huge advancement. Meanwhile, security has not received the same attention to go along with such developments. In this paper, we focus on the source location privacy problem in WSNs, a hot research topic in security, and propose a probabilistic source location privacy (PSLP) protection scheme for WSNs. A more powerful adversary, which can use hidden Markov model to estimate the state of the source, is considered in this study. To cope with this type of adversary, phantom nodes and fake sources, which are responsible to mimic the behavior of the source, are utilized to diversify the routing path. Then, the weight of each node is calculated as a criterion to select the next-hop candidate. In addition, two transmission modes are designed to transmit real packets. The simulation results demonstrate that the proposed PSLP scheme improves the safety time without compromising the energy consumption.

The Indignities of Civil Litigation

Dispute resolution has become increasingly shrouded in secrecy, with the proliferation of protective orders in discovery, confidential settlement agreements, and private arbitration. While many civil procedure scholars have criticized this trend for undermining the systemic benefits of public adjudication, the desirability of secrecy in civil litigation proves to be a much more complicated question. On the one hand, some of those same scholars have recently sought to justify civil litigation in terms that, ironically, highlight the benefits of secrecy. Although this new justification remains somewhat inchoate, it is best understood as a claim that the procedures of civil litigation allow individual plaintiffs to realize one aspect of their dignity — which this Article labels “dignity-as-status” — by empowering them to call those who have allegedly wronged them to account and to thereby reassert their standing as equals. The problem is that civil litigation can also undermine another aspect of plaintiffs’ dignity — which this Article labels “dignity-as-image” — by requiring them to divulge sensitive personal information and thus to cede control over their public self-presentation. Secrecy can help to preserve this second aspect of plaintiffs’ dignity. On the other hand, secrecy can also deprive plaintiffs of a potentially powerful expressive weapon in their quest to hold wrongdoers accountable. In conditions of socioeconomic inequality, weaker plaintiffs can sometimes turn the humiliating aspects of civil litigation to their advantage, intentionally revealing sensitive personal information that emphasizes their lower social status in order to shame their more powerful adversaries. It turns out that civil litigation can indeed promote plaintiffs’ dignity-as-status, but by affording them a venue in which to deliberately compromise their dignity-as-image — to humiliate, as much as ennoble, themselves. Given the complex nature of dignity and the complex trade-off between secrecy’s dignitarian benefits and costs, plaintiffs should be given more control over how much of their personal information is disseminated beyond the immediate parties to a lawsuit — a prescription with implications not only for secrecy in civil litigation, but also potentially for several other prominent procedural issues.

Unsupervised robust nonparametric learning of hidden community properties

We consider learning of fundamental properties of communities in large noisy networks, in the prototypical situation where the nodes or users are split into two classes according to a binary property, e.g., according to their opinions or preferences on a topic. For learning these properties, we propose a nonparametric, unsupervised, and scalable graph scan procedure that is, in addition, robust against a class of powerful adversaries. In our setup, one of the communities can fall under the influence of a knowledgeable adversarial leader, who knows the full network structure, has unlimited computational resources and can completely foresee our planned actions on the network. We prove strong consistency of our results in this setup with minimal assumptions. In particular, the learning procedure estimates the baseline activity of normal users asymptotically correctly with probability 1; the only assumption being the existence of a single implicit community of asymptotically negligible logarithmic size. We provide experiments on real and synthetic data to illustrate the performance of our method, including examples with adversaries.

Hardening quantum machine learning against adversaries

Security of machine learning has begun to become a serious issue for present day applications. An important question remaining is whether emerging quantum technologies will help or hinder the security of machine learning. Here we discuss a number of ways that quantum information can be used to help make quantum classifiers more secure or private. In particular, we demonstrate a form of robust principal component analysis that, under some circumstances, can provide an exponential speedup relative to robust methods used at present. To demonstrate this approach we introduce a linear combinations of unitaries Hamiltonian simulation method that we show functions when given an imprecise Hamiltonian oracle, which may be of independent interest. We also introduce a new quantum approach for bagging and boosting that can use quantum superposition over the classifiers or splits of the training set to aggragate over many more models than would be possible classically. Finally, we provide a private form of k-means clustering that can be used to prevent an all powerful adversary from learning more than a small fraction of a bit from any user. These examples show the role that quantum technologies can play in the security of ML and vice versa. This illustrates that quantum computing can provide useful advantages to machine learning apart from speedups.

Collusion-Resistant Processing of SQL Range Predicates

Prior solutions for securely handling SQL range predicates in outsourced Cloud-resident databases have primarily focused on passive attacks in the Honest-but-Curious adversarial model, where the server is only permitted to observe the encrypted query processing. We consider here a significantly more powerful adversary, wherein the server can launch an active attack by clandestinely issuing specific range queries via collusion with a few compromised clients. The security requirement in this environment is that data values from a plaintext domain of size N should not be leaked to within an interval of size H. Unfortunately, all prior encryption schemes for range predicate evaluation are easily breached with only O(log _2psi ) range queries, where psi = N{/}H. To address this lacuna, we present SPLIT, a new encryption scheme where the adversary requires exponentially more—{mathbf{O}}(psi )—range queries to breach the interval constraint and can therefore be easily detected by standard auditing mechanisms. The novel aspect of SPLIT is that each value appearing in a range-sensitive column is first segmented into two parts. These segmented parts are then independently encrypted using a layered composition of a secure block cipher with the order-preserving encryption and prefix-preserving encryption schemes, and the resulting ciphertexts are stored in separate tables. At query processing time, range predicates are rewritten into an equivalent set of table-specific sub-range predicates, and the disjoint union of their results forms the query answer. A detailed evaluation of SPLIT on benchmark database queries indicates that its execution times are well within a factor of two of the corresponding plaintext times, testifying its efficiency in resisting active adversaries.

Authenticated semiquantum dialogue with secure delegated quantum computation over a collective noise channel

Semiquantum communication permits a communication party with only limited quantum ability (i.e., “classical” ability) to communicate securely with a powerful quantum counterpart and will obtain a significant advantage in practice when the completely quantum world has not been built up. At present, various semiquantum schemes for key distribution, secret sharing and secure communication have been proposed. In a quantum dialogue (QD) scenario, two communicants mutually transmit their respective secret messages and may have equal power (such as two classical parties). Based on delegated quantum computation model, this work extends the original semiquantum model to the authenticated semiquantum dialogue (ASQD) protocols, where two “classical” participants can mutually transmit secret messages without any information leakage and quantum operations are securely delegated to a quantum server. To make the proposed ASQD protocols more practical, we assume that the quantum channel is a collective noise channel and the quantum server is untrusted. The security analysis shows that the proposed protocols are robust even when the delegated quantum server is a powerful adversary.