Federated Learning is highly susceptible to backdoor and targeted attacks as participants can manipulate their data and models locally without any oversight on whether they follow the correct process. There are a number of server-side defenses that mitigate the attacks by modifying or rejecting local updates submitted by clients. However, we find that bursty adversarial patterns with a high variance in the number of malicious clients can circumvent the existing defenses. We propose a client-self defense, LeadFL, that is combined with existing server-side defenses to thwart backdoor and targeted attacks. The core idea of LeadFL is a novel regularization term in local model training such that the Hessian matrix of local gradients is nullified. We provide the convergence analysis of LeadFL and its robustness guarantee in terms of certified radius. Our empirical evaluation shows that LeadFL is able to mitigate bursty adversarial patterns for both iid and non-iid data distributions. It frequently reduces the backdoor accuracy from more than 75% for state-of-the-art defenses to less than 10% while its impact on the main task accuracy is always less than for other client-side defenses.

Blockchain technology has revolutionized the way data is stored, managed, and shared across various industries. Its decentralized nature and immutability make it highly attractive in use cases that require transparency, integrity, and accountability. However, some applications demand confidentiality, necessitating the development of permissioned/consortium blockchains that try to strike a balance between transparency and privacy. Hyperledger Fabric is a permissioned blockchain that has gained popularity due to its modular architecture, performance, and scalability. Despite its strengths, the current set of privacy-enhancing features leaves room for improvement.

Therefore, this master thesis aims to explore the potential of introducing various privacy-enhancing technologies to Hyperledger Fabric, including Dynamic Searchable Symmetric Encryption (DSSE), Multi Authority Attribute-based Encryption (MA-ABE), and Trusted Execution Environments (TEE). Based on the promising results of our study, we decided to implement DSSE and MA-ABE. The combination of blockchain technology with TEE was ruled out after the thorough analysis of two research papers on the subject.

Our main result is the first implementation of a provable secure DSSE scheme in the context of consortium blockchains. Moreover, we developed a blockchain-enabled MA-ABE system that utilizes a novel and generic approach to foreign function invocation. Finally, the thesis discusses unexplored challenges in the field of logistics related to electronic consignment notes used in road transport. To address these issues, we designed a blockchain architecture that incorporates our developed privacy-enhanced technologies.

Payment channel networks (PCNs) are a promising solution to the blockchain scalability problem. They move payments off-chain, i.e., not all payments have to be included in the blockchain. Thus, they do not require that every payment is broadcast to all participants and verified by them. Not requiring global consensus reduces latency, computation, and hence energy consumption.

In PCNs, a payer can route a multi-hop payment to a payee via intermediaries. Yet, Lightning, the only prominent payment channel network, has two major issues when it comes to multi-hop payments. First, the payer decides on the path without being able to take local capacity restrictions into account. Second, due to the atomicity of payments, any failure in the path causes a failure of the complete payment. While there are suggested solutions for each of these problems --- namely splitting payments, local routing, and adding redundancy ---, there is no solution that solves both of them.

In this work, we propose JustForward: The payer adds redundancy to the payment by initially committing to sending a higher amount than the actual payment value $v$. Intermediaries decide on how to forward a received payment, potentially splitting it between multiple paths. If they cannot (or do not want to) forward the complete payment value, they may reduce the amount they forward. If paths for at least $v$ coins are found, the receiver and source jointly select the amounts of sub-payments that make up the full payment. Payment commitments are updated accordingly and fulfilled. In order to guarantee atomicity and correctness of the payment value, we use a modified Hashed Time Lock Contract (HTLC) for paying that requires both the payer and the payee to provide a secret pre-image. JustForward furthermore is the first local routing protocol to include fees: The source adds a maximal amount of fees they are willing to pay to the payment. Intermediaries take fees as they see fit, taking into consideration that subsequent nodes on the path will not forward the payment if there are insufficient fees left.

We prove that the proposed protocol achieves balance security, atomicity, bounded loss for the sender, and optionally unlinkability of split sub-payments of the same overall payments. Furthermore, our evaluation on both synthetic and real-world Lightning topologies shows JustForward outperforms existing algorithms in terms of the fraction of successful payments as long as the degree of redundancy is not too high. Concretely, for a moderate amount of redundancy, JustForward increases the success ratio of payments by about 10%.

Double Auctions are mechanisms to trade commodities such as electricity or parts of the wireless spectrum at optimal prices. Bidders and sellers simultaneously submit quantity-price pairs to an auctioneer, denoting the quantity they want to buy or sell at specific prices. The auctioneer aggregates the offers into demand and supply curves to compute the auction result by finding the intersection between supply and demand. In this way, commodities exchange owners in an economically efficient manner, driven by the market. In an ideal scenario, the auctioneer is a trusted third party that does not abuse the information they gain. However, in reality, offers reveal usage patterns of customers, such as electricity usage, or may be used by the auctioneer for their economic gain as insider information. The auctioneer also has opportunities to manipulate results of which there are real-life allegations in electricity trading or advertisement auctions. These concerns call for solutions that conduct the auction in a privacy-preserving and verifiable way while not compromising the auction functionality or economic efficiency.

Proposed solutions for privacy-preserving and verifiable double auctions offer confidentiality but do not allow participants to verify results independently or vice versa without interaction of participants in the full auction procedure. We specifically focus on electricity trading to design a solution covering the above concerns. To the best of our knowledge, we propose the first privacy-preserving and verifiable double auction scheme without interactivity of all participants, tailored to electricity trading on (inter)national exchanges. Using cryptographic schemes, including Homomorphic encryption, Commitment schemes, and Zero-knowledge-proofs, we propose a solution to establish a double auction protocol that preserves privacy and allows for verification.

Searchable Symmetric Encryption (SSE) schemes provide secure search over encrypted databases while allowing admitted information leakages. Generally, the leakages can be categorized into access, search, and volume pattern. In most existing Searchable Encryption (SE) schemes, these leakages are caused by practical designs but are considered an acceptable price to achieve high search efficiency. Many attacks on SSE schemes have shown that such leakages could be easily exploited to retrieve the underlying keywords for search queries. Each attack abuses a different leakage pattern and uses different techniques to achieve high query recovery accuracy. An attacker could be passive or active, where an active attacker can inject files in an SSE scheme, while a passive attacker only observes the queried data. Some passive attacks use the number of files returned by a query to create a match with a candidate keyword. Others use the co-occurrence of multiple keywords in the files to match a query with the same occurrence. We continue this research and design a new Volume and Access Pattern Leakage-abuse Attack (VAL-Attack) that exploits both the access and volume patterns. Our proposed attack only leverages leaked documents and the keywords present in those documents as auxiliary knowledge and can effectively retrieve document and keyword matches from leaked data. Furthermore, the recovery performs with great accuracy and without false positives. We compare VALAttack with two recent well-defined attacks on several real-world datasets to highlight the effectiveness of our attack and present the performance under popular countermeasures.

We can see and we can hear through the internet, but not yet touch. Tactile Internet (TI) is the paradigm that will enable the transmission of tactile feedback through the internet. TI requires Ultra Low Latency (ULL) networking to prevent desynchronization. Reaching the ULL requirements for medium to long distances is not possible without breaking the known physical limits of the speed of light. We propose instead to bypass these requirements by running a physical simulation of the controlled environment to perform predictions before the information can arrive. To this end, efficient low-polygon 3D meshes from the objects in the environment are required for physical simulation. The focus of this work is on reconstructing these meshes from point cloud scanners. Through a combination of Marching Cubes mesh reconstruction and Hoppe's Mesh Simplification, point clouds with up to 20 mm Gaussian noise can be accurately reconstructed into compact low polygon meshes.

As technology evolves, transmission speeds become faster. Tactile Internet requires ultra-low-latency (ULL) communications to further immerse humans in a remote environment by transmitting movement and force feedback, allowing them to interact with that environment in real-time. However, no transmission speed can be fast enough to support the "1 ms challenge" over long distances due to light speed limitations. A system with over 1ms of delay will feel unnatural to the user and cause "cyber-sickness". A novelty solution solves this by simulating the remote environment locally using point clouds. The system is then able to compute the force feedback immediately. This paper focuses on the desynchronization between simulation and reality that can build up due to disturbances. A framework for testing and observing desync caused by controlled disturbances is built for this purpose. The framework can also be used to test possible solutions to this issue. 1-dimensional simulations show that divergence happens slowly for friction and mass mismatches, providing a time frame during which it can be corrected. 2-dimensional simulations presented non-deterministic results, limiting the observations.

The tactile internet can be described as the next step of the internet. It can empower people, adding physical interactions to what normally is just an audiovisual experience.
It however does need ultra-low latency.
A solution to having delays can be found in environment simulation.
This paper describes a way to find the correct shape of objects within the observed environment.
There already are AI-based approaches, but the problem is that they are not transparent in how they work.
Hence the research question in this paper: Is it possible to create a 3D point cloud completion algorithm from 2.5D data, without resorting to the use of AI?
A three-step approach was chosen: dividing the scene, reconstructing its objects, and turning everything into meshes. The focus was on the reconstruction. The basic principle used is that of symmetry.
The main conclusions are that this approach can work, but only for relatively simple objects.
And whereas the results are easier to understand, the logic-based approach is slower and less accurate than existing approaches.

A problem faced by Tactile Internet (TI) is the distance limitation. A possible solution is to create a simulation of the remote environment, such that the delay experienced by a human operator is reduced. However, to obtain a model of the remote environment, before the user interacts with it, we need to be able to estimate the physical properties of objects in the environment from visual information.
This research paper investigates estimating an object's mass from visual information in the form of a point cloud. The object's volume is estimated by surfacing the mesh and dividing the resulting mesh into tetrahedrons, the volume of each tetrahedron can quickly be computed and summed to give an estimate of the total volume. The overall performance is better than the naive approach of using an OBB, however, it is less accurate than a volume slicing approach.

Some of the most prominent types of attacks against modern cryptographic implementations are side-channel attacks. These attacks leverage some unintended, often physical, leakage of the implementation to retrieve secret information. In recent times, a large part of the focus of side-channel research has been on deep learning methods. These methods operate in a profiled setting where a model is learned based on a copy of the device that is being attacked. This model is subsequently used to create significantly more potent attacks against the target. Attacks using deep learning methods can often defeat even implementations protected with countermeasures, but as implementations become more protected, novel methods are required to successfully generate attacks.
Recently, residual neural networks have been used for side-channel attacks, and these networks show promising attacking performance. However, these novel networks are relatively limited, and a more thorough investigation into the construction of residual networks in the side-channel context is required.
Our contribution is a more thorough investigation into the construction of these residual architectures. We explore several important factors to the construction of these models and generate insights into various methods for this construction. The resulting architectures we find show attacking performance that is competitive with the state-of-the-art methods across various data sets and feature selection scenarios.

In a world where more data gets uploaded to the cloud, it is essential that the data gets stored securely. For users to keep search functionality, searchable symmetric encryption has been developed. SSE works by a user sending a token representing a keyword (or a range), after which the server returns the documents that match the keyword (or values in the queried range). These observed documents are called the access pattern. A number of attacks that work on range SSE schemes have been developed. Yet, most of these attacks work on density or query assumptions that hinder their performance if these
assumptions are false. Furthermore, a number of these attacks use auxiliary information. Yet none of them uses known search tokens.

We, thus, propose a novel approach that uses a PQ-Tree to get the order of the observed documents. It uses the known pairs to assign possibilities for these documents and returns a list of options for each observed token and estimated values for each document. The basic attack guarantees that the correct query is always available by assigning document values based on the known tokens. A refined attack was made that assigns more known tokens by adding the best matches based on confidence score. This refinement allows for more exact token and query matches. Both of these attacks were extended using partial or similar documents from which document volume or rank information could be gathered
to further increase the attack performance in cases where such data is available. We evaluate our attacks against current state-of-the-art and outperform those regarding document and query recovery on all tested datasets. Lastly, we use two countermeasures and see that all attacks are impacted but that our attack still outperforms most.

The increasing demand for sharing Internet of Things (IoT) increases demand for a secure way of sharing data. Smart contracts could provide this because of its distributed nature. Proxy re-encryption is an encryption (PRE) method that can be used to share information. In this paper, a secure data sharing scheme is presented that uses multi-hop PRE with HyperLedger Fabric (HF). The scheme requires only two parties to participate with the data owner serving as the proxy and lets users remain access after re-encrypting the cipher. The implementation is used as means for demonstration and analysis. It still requires further work for actual deployment but demonstrates that the scheme holds in terms of efficiency and scalability.

Privacy and security are important topics in data aggregation, as companies are aggregating large amounts of sensitive data from their users. The aim of this paper is to improve the security and privacy of data aggregation using smart contracts, homomorphic encryption and post-quantum encryption methods on Hyperledger Fabric. After a literature study, the protocol discussed in \cite{Regueiro2021} was taken as the starting point for an adapted protocol. The adapted protocol is implemented and tested on performance and the security is discussed. The protocol is improved in terms of security by resolving or lowering the chance of security flaws that were present in the protocol in \cite{Regueiro2021}, and introducing post-quantum encryption. However, the adapted protocol in its current implementation is not scalable.

A rise in the use of blockchain systems implies an increase in smart contract usage. Blockchain systems can easily be found these days, most notably in cryptocurrencies and financial management. Furthermore, the Internet of Things (IoT) is being developed with blockchain, since blockchain ensures trust in IoT data and adds great flexibility. However, as with any other system, an increase in usage means the security of the blockchain system is becoming a more important aspect and an interesting area of research. Additionally, several pieces of research suggest that the smart contract is the most vulnerable component of the blockchain. In this paper, a simple prototype for a blockchain-based image encryption scheme for the Internet of Things will be implemented. This prototype will be implemented using Hyperledger Fabric Smart Contract and Self-Encryption scheme. Results will pro- pose that the encryption method implemented in the smart contract is secure enough against any other malicious users.

Blockchains like Bitcoin are known to be victim of scalability issues. The lack in high throughput and low latency form a great bottleneck to its network. A promis- ing solution are layer 2 protocols, more precisely payment channel networks (PCN). Payment success rates are a common metric in these networks. These rates can be in- creased by tweaking the routing of payments in the network. Local routing is a form of routing that allows payments in such networks to be split over multiple paths to reach its receiver. This significantly increases the rate of payment successes, however there is no trivial way to integrate fees in such protocol. This paper focuses on the integration of fees in local routing protocols by proposing a viable solution. Local Routing Fee Protocol (LRFP) is a protocol designed to extend an existing local routing protocol and is proven to be secure. It is a light addition but works as intended. Proofs on se- curity guarantees and a formal description on the protocol form the main contribution of this paper.

The largest payment channel network, Bitcoin Lightning, shows a potential alternative to cur- rent financial systems, overcoming the scalability limitations of blockchain. Source onion routing is used to route payments, but novel routing protocols claim improved effectiveness by employ- ing strategies not compatible with the current state of affairs. In this paper, we compare the current implementations deployed in the Lightning Network, with a novel routing algorithm with splitting and local routing. We find the latter option to significantly boost the success ratio of payments, while inducing some monetary overhead. Then, we propose future directions of research on Lightning protocols to boost their performance.

Authentication of domains has been a crucial part of the growth of web browsing, especially for e-commerce and secure browsing. However, the digital space has expanded from web domains to include devices such as smart cars, smart houses, and other IoT devices. The future for these devices is to communicate autonomously with one another, also known as Machine-to-Machine(M2M) communication. For M2M communication, IoT devices need to be able to authenticate each other. However, IoT devices cannot take over traditional authentication mechanisms. This problem is due to the expected growth in the number of IoT devices and the varying resource capabilities of these devices. In this work, we introduce Bubblechain, a generic, decentralised, and scalable IoT authentication system. In contrast to existing works, Bubblechain can update and remove identities of IoT devices that allows them to authenticate each other in a dynamic setting. Bubblechain also creates a higher trust level for devices that belong to the same Bubble. A Bubble refers to a group of devices that belong together, such as a home or office setting. The system also provides the possibility to authenticate devices from different Bubbles. The authentication process in Bubblechain is autonomous and decentralised.

Network covert timing channels are techniques to covertly transmit information over computer networks, by utilizing the time between subsequent network packets. Previous work on the detection of the various techniques has introduced numerous new methods, with high reported success. From these previous works we have noticed that there is little confirmation on these results in subsequent works, as well as there being a lack of an overview for the efficiency of each method. Next to this, we have found that many works use data in their experiments that may not be representative of real network scenarios. In this thesis we attempt to remedy this lack of information, by performing a broad performance evaluation on the currently existing singular detection metrics. This performance evaluation was done on a total of 18 different detection methods, applied to the 8 most prevalent covert timing channels. For the underlying network data, we gathered SSH and HTTPS traffic from the TU Delft, and applied varying amounts of simulated network jitter to them. From the resulting evaluations we find that there are cases where the detection methods do perform similarly to what has been shown in previous work, but we also find those that have a large difference in performance. Further, we discuss possible strengths and weaknesses of each of the detection methods, based on their performance, and in some cases how this performance might be improved. Using the (simulated) network scenarios we show the effects that jitter and different traffic types can have on each of the detection methods, and also find those that are resilient to network effects. Finally, we combine the full experimental performance evaluations into a comprehensive overview, for each combination of detection method and covert channel technique. We find that the current detection methods are likely not sufficient to be reliably applied in a realistic network setting, and more work needs to be done in this field to reach that point. The overview and discussions we have provided can then serve as a basis for future research, to give an indication of where performance needs to be improved.

Solving propositional satisfiability (SAT) and constraint programming (CP) instances has been a fundamental part of a wide range of modern applications. For this reason a lot of research went into improving the efficiency of modern SAT and CP solvers. Recently much of this research has gone into exploring the possibilities of integrating machine learning approaches with these solvers. However, with hybrid solvers, which combine both SAT and CP, dominating recent benchmarks it is surprising that no research has been done yet to apply those machine learning approach to improve hybrid solvers.This research proposes using a machine learning technique called unsatisfiable core learning to improve the performance of the Lazy Clause Generation solver Chuffed. The approach developed fort his study uses a Graph Convolutional Network model, which is trained on a dataset containing unsatisfiable instances. This machine learning model is then used for predicting unsatisfiable cores on CP instances and the predictions are used to initialise the activity score of the Variable State Independent Sum heuristic which is incorporated in Chuffed. The resulting approach managed to consistently solve a set of Multi-mode Resource-Constrained Project Scheduling instances 2.5% faster on average.These results indicate that, while this technique was originally developed for SAT, it can also be used to improve hybrid SAT/CP solvers.

The adoption of smart wearable devices has been on the rise over the past few years. These wearables are able to track the user's vital signs, making them valuable for use in the healthcare industry. Sharing this information with the user's healthcare provider has the potential to improve medical care by reducing medical misinformation. Currently, patients in the Netherlands are able to inspect their medical file at each healthcare provider they are attending, but they are unable to inspect all their medical files at one place. Moreover, they are not able to contribute to their own medical files. In this thesis, we propose the use of Trusted Execution Environments (TEEs) as an extension to the Polymorphic Encryption and Pseudonymisation (PEP) framework. PEP facilitates the exchange of medical data between multiple parties in a privacy-preserving manner. However, PEP suffers from collusion and scalability issues. The Distributed Polymorphic Encryption and Pseudonymisation (Dist-PEP) protocol is an improvement to PEP, and mitigates these issues. To make further improvements, the Distributed Polymorphic Access Management (Dist-PACMAN) protocol has been introduced to handle the access management more securely. By introducing TEEs, we make further improvements to the efficiency en scalability of the protocol. The result is a privacy-preserving framework that can be used to share the information gathered by the wearables with the user's healthcare provider securely and more efficiently. Additionally, the user can keep and inspect their own medical file at a cloud provider.