Search

Attacks on Federated Learning (FL) can severely reduce the quality of the generated models and limit the usefulness of this emerging learning paradigm that enables on-premise decentralized learning. However, existing untargeted attacks are not practical for many scenarios as they assume that i) the attacker knows every update of benign clients, or ii) the attacker has a large dataset to locally train updates imitating benign parties. In this paper, we propose a data-free untargeted attack (DFA) that synthesizes malicious data to craft adversarial models without eavesdropping on the transmission of benign clients at all or requiring a large quantity of task-specific training data. We design two variants of DFA, namely DFA-R and DFA-G, which differ in how they trade off stealthiness and effectiveness. Specifically, DFA-R iteratively optimizes a malicious data layer to minimize the prediction confidence of all outputs of the global model, whereas DFA-G interactively trains a malicious data generator network by steering the output of the global model toward a particular class. Experimental results on Fashion-MNIST, Cifar-10, and SVHN show that DFA, despite requiring fewer assumptions than existing attacks, achieves similar or even higher attack success rate than state-of-the-art untargeted attacks against various state-of-the-art defense mechanisms. Concretely, they can evade all considered defense mechanisms in at least 50% of the cases for CIFAR-10 and often reduce the accuracy by more than a factor of 2. Consequently, we design REFD, a defense specifically crafted to protect against data-free attacks. REFD leverages a reference dataset to detect updates that are biased or have a low confidence. It greatly improves upon existing defenses by filtering out the malicious updates and achieves high global model accuracy., Green Open Access added to TU Delft Institutional Repository 'You share, we take care!' - Taverne project https://www.openaccess.nl/en/you-share-we-take-care Otherwise as indicated in the copyright section: the publisher is the copyright holder of this work and the author uses the Dutch legislation to make this work public., Data-Intensive Systems

Catalyzed by the popularity of blockchain technology, there has recently been a renewed interest in the design, implementation and evaluation of decentralized systems. Most of these systems are intended to be deployed at scale and in heterogeneous environments with real users and unpredictable workloads. Nevertheless, most research in this field evaluates such systems in controlled environments that poorly reflect the complex conditions of real-world environments. In this work, we argue that deployment is crucial to understanding decentralized mechanisms in a real-world environment and an enabler to building more robust and sustainable systems. We highlight the merits of deployment by comparing this approach with other experimental setups and show how our lab applied a deployment-first methodology. We then outline how we use Tribler, our peer-to-peer file-sharing application, to deploy and monitor decentralized mechanisms at scale. We illustrate the application of our methodology by describing a deployment trial in experimental tokenomics. Finally, we summarize four lessons learned from multiple deployment trials where we applied our methodology., Green Open Access added to TU Delft Institutional Repository 'You share, we take care!' - Taverne project https://www.openaccess.nl/en/you-share-we-take-care Otherwise as indicated in the copyright section: the publisher is the copyright holder of this work and the author uses the Dutch legislation to make this work public., Data-Intensive Systems

Federated learning (FL) enables collaborative learning between parties, called clients, without sharing the original and potentially sensitive data. To ensure fast convergence in the presence of such heterogeneous clients, it is imperative to timely select clients who can effectively contribute to learning. A realistic but overlooked case of heterogeneous clients are Mavericks, who monopolize the possession of certain data types, e.g., children hospitals possess most of the data on pediatric cardiology. In this paper, we address the importance and tackle the challenges of Mavericks by exploring two types of client selection strategies. First, we show theoretically and through simulations that the common contribution-based approach, Shapley Value, underestimates the contribution of Mavericks and is hence not effective as a measure to select clients. Then, we propose FedEMD, an adaptive strategy with competitive overhead based on the Wasserstein distance, supported by a proven convergence bound. As FedEMD adapts the selection probability such that Mavericks are preferably selected when the model benefits from improvement on rare classes, it consistently ensures the fast convergence in the presence of different types of Mavericks. Compared to existing strategies, including Shapley Value-based ones, FedEMD improves the convergence speed of neural network classifiers with FedAvg aggregation by 26.9% and its performance is consistent across various levels of heterogeneity., Data-Intensive Systems

Due to its high efficiency, routing based on greedy embeddings of rooted spanning trees is a promising approach for dynamic, large-scale networks with restricted topologies. Friend-to-friend (F2F) overlays, one key application of embedding-based routing, aim to prevent disclosure of their participants to malicious members by restricting exchange of messages to mutually trusted nodes. Since embeddings assign a unique integer vector to each node that encodes its position in a spanning tree of the overlay, attackers can infer network structure from knowledge about assigned vectors. As this information can be used to identify participants, an evaluation of the scale of leakage is needed. In this work, we analyze in detail which information malicious participants can infer from knowledge about assigned vectors. Also, we show that by monitoring packet trajectories, malicious participants cannot unambiguously infer links between nodes of unidentified participants. Using simulation, we find that the vector assignment procedure has a strong impact on the feasibility of inference. In F2F overlay networks, using vectors of randomly chosen numbers for routing decreases the mean number of discovered individuals by one order of magnitude compared to the popular approach of using child enumeration indexes as vector elements., Green Open Access added to TU Delft Institutional Repository 'You share, we take care!' - Taverne project https://www.openaccess.nl/en/you-share-we-take-care Otherwise as indicated in the copyright section: the publisher is the copyright holder of this work and the author uses the Dutch legislation to make this work public., Data-Intensive Systems

Routing based on greedy network embeddings enables efficient and privacypreserving routing in overlays where connectivity is restricted to mutually trusted nodes. In previous works, we proposed security enhancements to the embedding and routing procedures to protect against denial-of-service attacks by malicious overlay participants. In this work, we propose an improved timeout scheme to reduce the stabilization overhead of secure tree maintenance in response to node failures and malicious behavior. Furthermore, we present an attack-resistant packet replication scheme that leverages alternative paths discovered during routing, Distributed Systems

Lightning, the prevailing solution to Bitcoin's scalability issue, uses onion routing to hide senders and recipients of payments. Yet, the path between the sender and the recipient along which payments are routed is selected such that it is short, cost efficient, and fast. The low degree of randomness in the path selection entails that anonymity sets are small. However, quantifying the anonymity provided by Lightning is challenging due to the existence of multiple implementations that differ with regard to the path selection algorithm and exist in parallel within the network. In this paper, we propose a general method allowing a local internal attacker to determine sender and recipient anonymity sets. Based on an in-depth code review of three Lightning implementations, we analyze how an adversary can predict the sender and the recipient of a multi-hop transaction. Our simulations indicate that only one adversarial node on a payment path uniquely identifies at least one of sender and recipient for around 70% of the transactions observed by the adversary. Moreover, multiple colluding attackers can almost always identify sender and receiver uniquely., Distributed Systems

Routing based on greedy network embeddings enables efficient and privacypreserving routing in overlays where connectivity is restricted to mutually trusted nodes. In previous works, we proposed security enhancements to the embedding and routing procedures to protect against denial-of-service attacks by malicious overlay participants. In this work, we propose an improved timeout scheme to reduce the stabilization overhead of secure tree maintenance in response to node failures and malicious behavior. Furthermore, we present an attack-resistant packet replication scheme that leverages alternative paths discovered during routing, Dataintensive Systems

Lightning, the prevailing solution to Bitcoin's scalability issue, uses onion routing to hide senders and recipients of payments. Yet, the path between the sender and the recipient along which payments are routed is selected such that it is short, cost efficient, and fast. The low degree of randomness in the path selection entails that anonymity sets are small. However, quantifying the anonymity provided by Lightning is challenging due to the existence of multiple implementations that differ with regard to the path selection algorithm and exist in parallel within the network. In this paper, we propose a general method allowing a local internal attacker to determine sender and recipient anonymity sets. Based on an in-depth code review of three Lightning implementations, we analyze how an adversary can predict the sender and the recipient of a multi-hop transaction. Our simulations indicate that only one adversarial node on a payment path uniquely identifies at least one of sender and recipient for around 70% of the transactions observed by the adversary. Moreover, multiple colluding attackers can almost always identify sender and receiver uniquely., Data-Intensive Systems

Federated Learning is an emerging distributed collaborative learning paradigm adopted by many of today's applications, e.g., keyboard prediction and object recognition. Its core principle is to learn from large amount of users data while preserving data privacy by design as collaborative users only need to share the machine learning models and keep data locally. The main challenge for such systems is to provide incentives to users to contribute high-quality models trained from their local data. In this paper, we aim to answer how well incentives recognize (in)accurate local models from honest and malicious users, and perceive their impacts on the model accuracy of federated learning systems. We first present a thorough survey on two contrasting perspectives: incentive mechanisms to measure the contribution of local models by honest users, and malicious users to deliberately degrade the overall model. We conduct simulation experiments to empirically demonstrate if existing contribution measurement schemes can disclose low-quality models from malicious users. Our results show there exists a clear tradeoff among measurement schemes in terms of the computational efficiency and effectiveness to distill the impact of malicious participants. We conclude this paper by discussing the research directions to design resilient contribution incentives., Distributed Systems

Greedy embeddings on rooted spanning trees are the most promising solution to provide sufficiently scalable routing in dynamic networks with restricted topologies, for instance friend-to-friend overlays such as the Dark Freenet and payment channel networks such as Lightning. Yet, they are not deployed in practice, as electing a root and configuring addresses remains an unsolved problem in adverse environments. Indeed, faulty or malicious nodes might provide incorrect coordinates, prevent the network from stabilizing by simulating dynamics, or not start the assignment of coordinates in their subtree at all. All of the above attacks may result in an inability to route. To mitigate the above attacks, we design a novel embedding algorithm with an adapted distance metric that only relies on interconnections between benign subtrees for successful delivery. In other words, even if roots of (sub-)trees are malicious or faulty, the remaining nodes still receive coordinates and can communicate with nodes in their tree branch as well as other branches reachable via the neighborhood of their benign ancestors. Extensive simulations demonstrate that we thus facilitate efficient routing even when seemingly decisive parts of the network are under adversarial control., Distributed Systems

Payment channel networks like Bitcoin’s Lightning network are an auspicious approach for realizing high transaction throughput and almost-instant confirmations in blockchain networks. However, the ability to successfully conduct payments in such networks relies on the willingness of participants to lock collateral in the network. In Lightning, the key financial incentive to lock collateral are low fees for routing payments of other participants. While users can choose these fees, real-world data indicates that they mainly stick to default fees. By providing insights on beneficial choices for fees, we aim to incentivize users to lock more collateral and improve the effectiveness of the network. In this paper, we consider a node that given the network topology and the channel details establishes channels and chooses fees to maximize its financial gain. Our contributions are i) formalization of the optimization problem, ii) proving that the problem is NP-hard, and iii) designing and evaluating a greedy algorithm to approximate the optimal solution. In each step, our greedy algorithm establishes a channel that maximizes the increase to ’s total reward, which corresponds to maximizing the number of shortest paths passing through. Our simulation study leveraged real-world data sets to quantify the impact of our gain optimization and indicates that our strategy is at least a factor two better than other strategies., Coronavirus update: Authors whose travel is disrupted can arrange to give video presentations, Distributed Systems, Cyber Security

Current leader election algorithms fail in the presence of Sybil attacks, i.e., one malicious entity inserting many nodes, network dynamics, and restricted knowledge about the graph. However, social overlay networks, i.e., peer-to-peer networks with links corresponding to social relationships, face all of the above challenges. Social overlay networks naturally offer privacy, as they avoid connections with strangers, and furthermore prevent a Sybil attacker from controlling a large number of links in the graph. As recent ideas for scalable communication in such overlays rely heavily on attack resistant leader election, solving leader election for such overlays opens the door for decentralized, privacy-preserving, and secure communication at a large scale. In this work, we propose a novel leader election algorithm based on three-majority voting that utilizes timestamps and cryptographic signatures to detect leader faults in an attack resistant manner. We evaluate our algorithm with simulations on real-world as well as synthetic network topologies. Our results indicate that in networks whose degree sequence follows a power law, our leader election algorithm quickly achieves consensus for more than $80\%$ of all nodes. Furthermore, attackers are unlikely to become leaders as long as the number of connections they establish with honest nodes is low., Distributed Systems

Federated Learning is an emerging distributed collaborative learning paradigm adopted by many of today's applications, e.g., keyboard prediction and object recognition. Its core principle is to learn from large amount of users data while preserving data privacy by design as collaborative users only need to share the machine learning models and keep data locally. The main challenge for such systems is to provide incentives to users to contribute high-quality models trained from their local data. In this paper, we aim to answer how well incentives recognize (in)accurate local models from honest and malicious users, and perceive their impacts on the model accuracy of federated learning systems. We first present a thorough survey on two contrasting perspectives: incentive mechanisms to measure the contribution of local models by honest users, and malicious users to deliberately degrade the overall model. We conduct simulation experiments to empirically demonstrate if existing contribution measurement schemes can disclose low-quality models from malicious users. Our results show there exists a clear tradeoff among measurement schemes in terms of the computational efficiency and effectiveness to distill the impact of malicious participants. We conclude this paper by discussing the research directions to design resilient contribution incentives., Data-Intensive Systems

Greedy embeddings on rooted spanning trees are the most promising solution to provide sufficiently scalable routing in dynamic networks with restricted topologies, for instance friend-to-friend overlays such as the Dark Freenet and payment channel networks such as Lightning. Yet, they are not deployed in practice, as electing a root and configuring addresses remains an unsolved problem in adverse environments. Indeed, faulty or malicious nodes might provide incorrect coordinates, prevent the network from stabilizing by simulating dynamics, or not start the assignment of coordinates in their subtree at all. All of the above attacks may result in an inability to route. To mitigate the above attacks, we design a novel embedding algorithm with an adapted distance metric that only relies on interconnections between benign subtrees for successful delivery. In other words, even if roots of (sub-)trees are malicious or faulty, the remaining nodes still receive coordinates and can communicate with nodes in their tree branch as well as other branches reachable via the neighborhood of their benign ancestors. Extensive simulations demonstrate that we thus facilitate efficient routing even when seemingly decisive parts of the network are under adversarial control., Data-Intensive Systems

Payment channel networks like Bitcoin’s Lightning network are an auspicious approach for realizing high transaction throughput and almost-instant confirmations in blockchain networks. However, the ability to successfully conduct payments in such networks relies on the willingness of participants to lock collateral in the network. In Lightning, the key financial incentive to lock collateral are low fees for routing payments of other participants. While users can choose these fees, real-world data indicates that they mainly stick to default fees. By providing insights on beneficial choices for fees, we aim to incentivize users to lock more collateral and improve the effectiveness of the network. In this paper, we consider a node that given the network topology and the channel details establishes channels and chooses fees to maximize its financial gain. Our contributions are i) formalization of the optimization problem, ii) proving that the problem is NP-hard, and iii) designing and evaluating a greedy algorithm to approximate the optimal solution. In each step, our greedy algorithm establishes a channel that maximizes the increase to ’s total reward, which corresponds to maximizing the number of shortest paths passing through. Our simulation study leveraged real-world data sets to quantify the impact of our gain optimization and indicates that our strategy is at least a factor two better than other strategies., Coronavirus update: Authors whose travel is disrupted can arrange to give video presentations, Data-Intensive Systems, Cyber Security

Current leader election algorithms fail in the presence of Sybil attacks, i.e., one malicious entity inserting many nodes, network dynamics, and restricted knowledge about the graph. However, social overlay networks, i.e., peer-to-peer networks with links corresponding to social relationships, face all of the above challenges. Social overlay networks naturally offer privacy, as they avoid connections with strangers, and furthermore prevent a Sybil attacker from controlling a large number of links in the graph. As recent ideas for scalable communication in such overlays rely heavily on attack resistant leader election, solving leader election for such overlays opens the door for decentralized, privacy-preserving, and secure communication at a large scale. In this work, we propose a novel leader election algorithm based on three-majority voting that utilizes timestamps and cryptographic signatures to detect leader faults in an attack resistant manner. We evaluate our algorithm with simulations on real-world as well as synthetic network topologies. Our results indicate that in networks whose degree sequence follows a power law, our leader election algorithm quickly achieves consensus for more than $80\%$ of all nodes. Furthermore, attackers are unlikely to become leaders as long as the number of connections they establish with honest nodes is low., Data-Intensive Systems

Nodes in route-restricted overlays have an im-mutable set of neighbors, explicitly specified by their users. Pop-ular examples include payment networks such as the Lightningnetwork as well as social overlays such as the Dark Freenet.Routing algorithms are central to such overlays as they enablecommunication between nodes that are not directly connected.Recent results show that algorithms based on spanning treesare the most promising provably efficient choice. However, allsuggested solutions fail to address how distributed spanning treealgorithms can deal with active denial of service attacks bymalicious nodes.In this work, we design a novel self-stabilizing spanning treeconstruction algorithm that utilizes cryptographic signatures andprove that it reduces the set of nodes affected by active attacks.Our simulations substantiate this theoretical result withconcretevalues based on real-world data sets. In particular, our resultsindicate that our algorithm reduces the number of affectednodes by up to 74% compared to state-of-the-art attack-resistantspanning tree constructions, Distributed Systems

Nodes in route-restricted overlays have an im-mutable set of neighbors, explicitly specified by their users. Pop-ular examples include payment networks such as the Lightningnetwork as well as social overlays such as the Dark Freenet.Routing algorithms are central to such overlays as they enablecommunication between nodes that are not directly connected.Recent results show that algorithms based on spanning treesare the most promising provably efficient choice. However, allsuggested solutions fail to address how distributed spanning treealgorithms can deal with active denial of service attacks bymalicious nodes.In this work, we design a novel self-stabilizing spanning treeconstruction algorithm that utilizes cryptographic signatures andprove that it reduces the set of nodes affected by active attacks.Our simulations substantiate this theoretical result withconcretevalues based on real-world data sets. In particular, our resultsindicate that our algorithm reduces the number of affectednodes by up to 74% compared to state-of-the-art attack-resistantspanning tree constructions, Data-Intensive Systems

The genomes of four Lactobacillus reuteri strains isolated from human breast milk and the gastrointestinal tract have been recently sequenced as part of the Human Microbiome Project. Preliminary genome comparisons suggested that these strains belong to two different clades, previously shown to differ with respect to antimicrobial production, biofilm formation, and immunomodulation. To explain possible mechanisms of survival in the host and probiosis, we completed a detailed genomic comparison of two breast milk–derived isolates representative of each group: an established probiotic strain (L. reuteri ATCC 55730) and a strain with promising probiotic features (L. reuteri ATCC PTA 6475). Transcriptomes of L. reuteri strains in different growth phases were monitored using strain-specific microarrays, and compared using a panmetabolic model representing all known metabolic reactions present in these strains. Both strains contained candidate genes involved in the survival and persistence in the gut such as mucus-binding proteins and enzymes scavenging reactive oxygen species. A large operon predicted to encode the synthesis of an exopolysaccharide was identified in strain 55730. Both strains were predicted to produce health-promoting factors, including antimicrobial agents and vitamins (folate, vitamin B12). Additionally, a complete pathway for thiamine biosynthesis was predicted in strain 55730 for the first time in this species. Candidate genes responsible for immunomodulatory properties of each strain were identified by transcriptomic comparisons. The production of bioactive metabolites by human-derived probiotics may be predicted using metabolic modeling and transcriptomics. Such strategies may facilitate selection and optimization of probiotics for health promotion, disease prevention and amelioration., Biotechnology, Applied Sciences

The genomes of four Lactobacillus reuteri strains isolated from human breast milk and the gastrointestinal tract have been recently sequenced as part of the Human Microbiome Project. Preliminary genome comparisons suggested that these strains belong to two different clades, previously shown to differ with respect to antimicrobial production, biofilm formation, and immunomodulation. To explain possible mechanisms of survival in the host and probiosis, we completed a detailed genomic comparison of two breast milk–derived isolates representative of each group: an established probiotic strain (L. reuteri ATCC 55730) and a strain with promising probiotic features (L. reuteri ATCC PTA 6475). Transcriptomes of L. reuteri strains in different growth phases were monitored using strain-specific microarrays, and compared using a panmetabolic model representing all known metabolic reactions present in these strains. Both strains contained candidate genes involved in the survival and persistence in the gut such as mucus-binding proteins and enzymes scavenging reactive oxygen species. A large operon predicted to encode the synthesis of an exopolysaccharide was identified in strain 55730. Both strains were predicted to produce health-promoting factors, including antimicrobial agents and vitamins (folate, vitamin B12). Additionally, a complete pathway for thiamine biosynthesis was predicted in strain 55730 for the first time in this species. Candidate genes responsible for immunomodulatory properties of each strain were identified by transcriptomic comparisons. The production of bioactive metabolites by human-derived probiotics may be predicted using metabolic modeling and transcriptomics. Such strategies may facilitate selection and optimization of probiotics for health promotion, disease prevention and amelioration., Biotechnology, Applied Sciences