"uuid","repository link","title","author","contributor","publication year","abstract","subject topic","language","publication type","publisher","isbn","issn","patent","patent status","bibliographic note","access restriction","embargo date","faculty","department","research group","programme","project","coordinates"
"uuid:7ed823a0-5a2c-4439-8fd0-60e02235f24e","http://resolver.tudelft.nl/uuid:7ed823a0-5a2c-4439-8fd0-60e02235f24e","Evaluating differential privacy on language processing federated learning","Van Opstal, Quinten (TU Delft Electrical Engineering, Mathematics and Computer Science)","Huang, J. (mentor); Chen, Lydia Y. (mentor); Zuniga, Marco (graduation committee); Delft University of Technology (degree granting institution)","2024","Federated learning provides a lot of opportunities, especially with the built-in privacy considerations. There is however one attack that might compromise the utility of federated learning: backdoor attacks [14]. There are already some existing defenses, like flame [13] but they are computationally expensive [14]. This paper evaluates a version of differential privacy, where the Gaussian noise added to the aggravated model of the clipped updates is smaller than usually. This is often referred to as weak differential privacy or weakDP. This paper evaluates weakDP with different parameters to find if weakDP can be used as a defense for a language processing federated learning classifier against a backdoor attack.","Federated learning; Differential Privacy; Non-iid","en","bachelor thesis","","","","","","A link to the associated github project. - https://github.com/QuintenVanOpstal/OOD_Federated_Learning.git","","","","","","Computer Science and Engineering","CSE3000 Research Project",""
"uuid:60f7bc45-2550-4e03-a570-ae2a4bb01b14","http://resolver.tudelft.nl/uuid:60f7bc45-2550-4e03-a570-ae2a4bb01b14","Analysis on the Vulnerability of Multi-Server Federated Learning Against Model Poisoning Attacks","Nenovski, Lazar (TU Delft Electrical Engineering, Mathematics and Computer Science)","Huang, J. (mentor); Chen, Lydia Y. (mentor); Zuniga, Marco (graduation committee); Delft University of Technology (degree granting institution)","2024","Abstract— Federated Learning (FL) makes it possible for a network of clients to jointly train a machine learning model, while also keeping the training data private. There are several approaches when designing a FL network and while most existing research is focused on a single-server design, new and promising variations are arising that make use of multiple servers, witch have the benefit of speeding up the training process. Unfortunately single-server FL networks are prone to model poisoning attacks by malicious participants, that aim to reduce the accuracy of the trained model. This work showcases the inherent resilience of the multi-server design against existing state-of-the-art attacks tailored around single-server FL, as well as propose two novel attacks that exploits multi-server topology in order to reduce the required knowledge an adversary needs to obtain to carry out the attack, while still remaining effective. Main findings are as follows: In the event that the malicious party has compromised the entire network, existing single-server attacks are sufficient to completely prevent a model from training. If they are limited to knowledge available only within the local reach of their compromised clients, the effect is minimized to where the attacks might get mitigated without any defences being necessary. However in such cases a correlation can be observed between the location of the compromised clients and the effectiveness of an attack. The novel attacks proposed in this paper exploit this relation in order to remain sufficiently effective while requiring only the same amount of data necessary for the multi-server algorithm to function.","Federated Learning; Multi-server FL; Untargeted Attack","en","bachelor thesis","","","","","","","","","","","","Computer Science and Engineering","CSE3000 Research Project",""
"uuid:f90541b8-232b-47be-8470-d79b273279ae","http://resolver.tudelft.nl/uuid:f90541b8-232b-47be-8470-d79b273279ae","Exploring the Impact of Single-Character Attacks in Federated Learning Language Classification: Introducing the Novel Single-Character Strike","van der Meulen, Jan (TU Delft Electrical Engineering, Mathematics and Computer Science)","Chen, Lydia Y. (mentor); Huang, J. (mentor); Zuniga, Marco (graduation committee); Delft University of Technology (degree granting institution)","2024","Federated learning (FL) is a privacy preserving machine learning approach which allows a machine learning model to be trained in a distributed fashion without ever sharing user data. Due to the large amount of valuable text and voice data stored on end-user devices, this approach works particularly well for natural language processing (NLP) tasks. Due to many applications making use of the algorithm and increasing interest in academics, ensuring security is essential. Current backdoor attacks in NLP tasks are still unable to evade some defence mechanisms. Therefore, we propose a novel attack, the single-character strike to address this research gap. Consequently, the following research question is posed: What are the properties of the single-character strike in a language classification task? By experimental analysis the following properties are discovered: the single-character strike is undetectable against five state-of-the-art defences, has low impact on the global model accuracy, trains slower than similar attacks, relies on characters on the edge of the distribution to function, is robust within the global model, and performs best when close to convergence and with more adversarial clients. Emphasizing its imperceptibility and persistence, the attack maintains a 70\% backdoor accuracy after a thousand iterations without training and remains undetectable against: (Multi-)Krum, RFA, Norm Clipping and Weak Differential Privacy. By providing insight into the effective single-character strike, this paper adds to the growing body of work that questions whether federated learning can be secure against backdoor attacks.","federated learning; natural language processing; backdoor attack; single-character-strike; security","en","bachelor thesis","","","","","","","","","","","","Computer Science and Engineering","CSE3000 Research Project",""
"uuid:49346f35-e8eb-4f9d-870a-6ed44755f6be","http://resolver.tudelft.nl/uuid:49346f35-e8eb-4f9d-870a-6ed44755f6be","Mitigating Regional Accent Bias in ASR Systems","Li, Zirui (TU Delft Electrical Engineering, Mathematics and Computer Science)","Scharenborg, O.E. (mentor); Zuniga, Marco (graduation committee); Delft University of Technology (degree granting institution)","2023","End-to-end Automatic Speech Recognition (ASR) systems improved drastically in recent years and they work extremely well on many large datasets. However, research shows that these models failed to capture the variability in speech production and have biases against the variant caused by the regional accented speech. Moreover, ASR research on regional accents is primarily done in languages used by a large population, like English and Arabic, and the effect of regional accented speech on E2E ASR systems in non-popular languages is still unknown. It is important to know the effect of regional accented speech on E2E ASR systems as it helps researchers to build an inclusive E2E ASR system. In this project, I aim to mitigate the biases against regional accented speech. I select standard speech and regional accented speech from CommonVoice's French and German datasets. I combine the state-of-the-art Conformer Recurrent Neural Network Transducer model with Multi-Domain Adversarial Training (MDAT) to boost the performance of regional accented speech while not hurting the performance of the standard speech. Moreover, since the regional accented speech is typically low-resourced, I study the amount of data required for effective MDAT, as well as the effect of different domain classifiers on the performance of Multi-Domain Adversarial Training. Experimental results show that MDAT can mitigate the biases against regional accented speech in both French and German. The best model in French reduces the bias by around 12% and the best model in German reduces the bias by around 7%. Additionally, MDAT is an effective method for bias mitigation as it can achieve similar performance as the MDAT model trained with the full dataset using only a small amount (e.g. 30 minutes) of untranscribed regional accented speech. Finally, different domain classifier architectures were found to have similar effects on the results of MDAT, thus there is no significant differences among the domain classifier in this project.","bias mitigation; automatic speech recognition; regional accented speech; domain adversarial training","en","master thesis","","","","","","","","","","","","Electrical Engineering | Embedded Systems","",""
"uuid:ad40d91a-fba1-49e8-b71e-acb3cc9f4a9d","http://resolver.tudelft.nl/uuid:ad40d91a-fba1-49e8-b71e-acb3cc9f4a9d","Screen antenna: Integrated data transfer and display using RGB LEDs","Mandlik, Nishad (TU Delft Electrical Engineering, Mathematics and Computer Science)","Wang, Q. (mentor); Zuniga, Marco (graduation committee); Delft University of Technology (degree granting institution)","2023","This thesis presents Screen Antenna - A Visible Light Communication (VLC) system that integrates data transmission and reception, with the conventional pixel display capability of RGB LEDs. The system is constructed with off-the-shelf components and runs on the Arduino Due microcontroller. The hardware and software have been designed with the objectives of scalability and flexibility. This research aims to optimize the system's performance in terms of throughput, error rate, and display quality.
To evaluate the system's performance, extensive experiments were conducted under different lighting conditions, PWM frequency and Rx-Tx distances. The experiments focused on achieving the maximum achievable throughput while maintaining a low Bit Error Rate (BER). Through careful optimization of the system parameters, a maximum throughput of 4.4kbps was achieved. Furthermore, the system consistently maintained a Bit Error Rate of less than 0.005\%, ensuring reliable and error-free data transmission.
In addition to data transmission, the impact of data transmission on the display's visual quality was investigated. The Delta E value, which quantifies the observed colour difference between two pixel values, was used as a metric for display quality. The data rates were calibrated such that the display operated with a Delta-E value of less than 1 in more than 90\% of the colour spectrum, indicating minimal colour distortion during data transmission.
This study contributes to the field of VLC by showcasing a practical implementation that combines scalability, visual quality, and performance. Future research directions could focus on expanding the display size and enhancing the data rate by means of more sophisticated hardware.","Visible Light Communication; VLC; RGB LED; LED; LED display; Data communication; LED matrix","en","master thesis","","","","","","","","2025-07-06","","","","Electrical Engineering | Embedded Systems","",""
"uuid:c2a7e111-4e91-4862-a622-4c19616da108","http://resolver.tudelft.nl/uuid:c2a7e111-4e91-4862-a622-4c19616da108","CIM-architecture for acceleration of DNA pre-alignment filters","Miao, Michael (TU Delft Electrical Engineering, Mathematics and Computer Science)","Wong, J.S.S.M. (mentor); Zuniga, Marco (graduation committee); Shahroodi, T. (mentor); Delft University of Technology (degree granting institution)","2023","Due to recent developments in DNA sequencing technology, there is a growing abundance of available genomic data. To process this information for use in fields such as healthcare and forensics, raw sequencing data have to be processed using computationally intensive algorithms. Currently, one of the major bottlenecks in this processing pipeline is the alignment step, which makes use of dynamic-programming algorithms. To reduce computation times, numerous solutions have been proposed aimed at reducing the execution time of the alignment step. This is done either by accelerating alignment itself using hardware accelerators and heuristics or by reducing the amount of input data through the use of pre-alignment filters. The algorithms associated with the latter solution are less computationally intensive than DP-based alignment, which reduces the end-to-end alignment time.
Currently, pre-alignment filters are effective to the point where the alignment bottleneck is shifted to the filtering step. Therefore, the filters are accelerated on hardware solutions such as GPUs and FPGAs. While these solutions show orders of magnitude improvement in execution times, they are insufficient for removing the filtering bottleneck entirely. The performance of these hardware accelerators is limited by the rate at which data can be supplied. As a solution, we propose a CIM-based accelerator to reduce data-movement overheads between the host device and the accelerator. Additionally, this architecture makes use of emerging non-volatile memories to perform Boolean operations directly within its memory elements. In doing so, it can exploit parallelism in the algorithms to achieve higher throughput.
In this work, we explore commonly found operations in existing pre-alignment filters and devise ways to implement them on the CIM-architecture. The proposed architecture is flexible in supporting multiple pre-alignment filters and a wide range of input data. The functionality of the architecture is verified through simulation and its effectiveness is tested using real data sets.
Using this architecture, we can achieve improvement in end-to-end execution time over the state of the art ranging from 7.2x to 119.6x for the evaluated data sets, while also achieving a reduction of up to 59% and 79.7% in chip-area and power consumption, respectively.
Furthermore, the provided work offers a platform for the development of future pre-alignment filtering algorithms to further improve performance.","CIM; Pre-alignment filter; Architecture; DNA analysis; eNVM; Memory; Accelerators","en","master thesis","","","","","","","","2023-04-24","","","","Electrical Engineering | Embedded Systems","",""
"uuid:10211bcd-63e0-41d1-bb57-4ee244b36c60","http://resolver.tudelft.nl/uuid:10211bcd-63e0-41d1-bb57-4ee244b36c60","Analysis and modelling of light sources for visible light communication","van Mierop, Ron (TU Delft Electrical Engineering, Mathematics and Computer Science)","Isabella, O. (mentor); Manganiello, P. (mentor); Zuniga, Marco (graduation committee); Muttillo, M. (mentor); Delft University of Technology (degree granting institution)","2023","The considerable increase in the number of devices needing connectivity, such as mobile phones and Internet of Things (IoT) devices, has led to an exponential rise in data volumes during the last years, that will surely continue over the next decade. Therefore, it will be increasingly challenging to provide sufficient RF resources. A novel alternative to RF communications is Visible Light Communication (VLC). VLC is a communication technology that uses visible light as an information carrier. The use of VLC for indoor applications has been rapidly growing during the last years – Light Fidelity (LiFi) technology is an example of VLC application – with photodiodes being the most widely used receiving devices. However, looking at both indoor and outdoor communication, photovoltaic (PV) cells represent a relevant alternative for detecting the information. One of the advantages of using a PV cell as receiver is the huge sensitive area for detection of the information that simplifies alignment between transmitter and receiver.
Different light sources can be used in VLC. Typically, either LEDs or LASERs are considered, depending on the characteristics of the link (such as distance, type of receiver, indoor/outdoor application). These light sources differ in terms of spectrum, directionality, optical power density and bandwidth. The performance of the whole VLC link strongly depends on the characteristics of the light source, since it affects the ability of the receiver, such as a PV-device, to detect the information when it overlaps with the ambient light, that can reach very high values, especially in outdoor applications where the ambient light is the sunlight. Therefore, the modelling and analysis of the performance of different light sources in a PV-based VLC link will pave the way towards the realisation of a PV-based communication system of the future; and it is the focus of this thesis project.
The project goals were achieved by first reviewing the characteristics of light sources, to understand their advantages and drawbacks in (PV-based) VLC. This was followed by the development of models of the VLC data-link, with a focus on the light source, which took into account various factors such as the type of light source, its location relative to the receiver, and its dynamic behaviour. This was followed by the realisation of a test setup, to characterise different light sources, and the models were then used to simulate the light distribution from the actual light sources. Finally, the framework was used to simulate a LED-based solar simulator and an outdoor VLC data-link.","Visible light communication; VLC; LED; PV Receiver","en","master thesis","","","","","","","","2025-03-13","","","","Electrical Engineering","",""
"uuid:a7450fad-43ff-446e-ba8e-d7a10fc50029","http://resolver.tudelft.nl/uuid:a7450fad-43ff-446e-ba8e-d7a10fc50029","People Counting Using Low-cost FMCW MIMO Radar: Achieving Tracking for Counting and Classification of Groups of People using FMCW Radar","Ren, Liyuan (TU Delft Electrical Engineering, Mathematics and Computer Science)","Fioranelli, F. (mentor); Yarovoy, Alexander (graduation committee); Zuniga, Marco (graduation committee); Delft University of Technology (degree granting institution)","2022","For the development of automatic People Counting systems, radar is increasingly becoming a popular technology because of the increasingly stringent privacy requirements for people demographic information and the requirement to operate in a challenging environment. Because of the complexity of multi-target movement and the diversity of application scenarios, Radar-based People Counting methods are required to have sufficient robustness. However, based on the review of the current literature, the grouping phenomenon (i.e., multiple individuals moving close together as a single group) was not often considered in the experimental scenarios.
This thesis aims to study one of the most complex motions of individuals, grouping, and address the People Counting problem more in general, including the cases of grouping of multiple individuals. After studying the characteristics of Group People (defined as a group of people sharing neighboring, adjacent locations and moving together), with the help of multiple-input-multiple-output (MIMO) frequency-modulated continuous wave (FMCW) Radar, the combination of the Range-Azimuth map and spectrogram/cadence velocity diagram (CVD) is proposed to solve Group People Counting.
Algorithm-wise, there are two categories of existing Radar-based People Counting methods, namely, tracking for counting methods and feature-based counting methods. It was found that these two categories of methods have complementary strengths. Thus, the proposed method combines the tracking for counting approach and feature-based counting approach into a new processing pipeline to estimate the number of people in each group in the scene while tracking each group. Based on it, the proposed method achieves ""Beyond classification"", which is the output the unlabeled classes not defined at the training stage. Moreover, compared with other state-of-the-art (SOTA) Radar-based People Counting methods, the proposed method outperforms them, and thus it is proved that the grouping problem can be solved in the Radar-based People Counting field.","people counting; Automotive radar; grouping","en","master thesis","","","","","","","","","","","","Electrical Engineering","",""
"uuid:7d7f279c-61b9-4739-b03c-637f065d460d","http://resolver.tudelft.nl/uuid:7d7f279c-61b9-4739-b03c-637f065d460d","FLVoogd: Robust And Privacy Preserving Federated Learning","Tian, Yuhang (TU Delft Electrical Engineering, Mathematics and Computer Science)","Liang, K. (mentor); Wang, R. (graduation committee); Verwer, S.E. (graduation committee); Zuniga, Marco (graduation committee); Delft University of Technology (degree granting institution)","2022","In this work, we propose FLVoogd, an updated federated learning method in which servers and clients collaboratively eliminate Byzantine attacks while preserving privacy. In particular, servers use automatic Density-based Spatial Clustering of Applications with Noise (DBSCAN) combined with S2PC to cluster the benign majority without acquiring sensitive personal information. Meanwhile, clients build dual models and perform test-based distance controlling to adjust their local models toward the global one to achieve personalizing. Our framework is automatic and adaptive that servers/clients don't need to tune the parameters during the training. In addition, our framework leverages Secure Multi-party Computation (SMPC) operations, including multiplications, additions, and comparison, where costly operations, like division and square root, are not required. Evaluations are carried out on some conventional datasets from the image classification field. The result shows that FLVoogd can effectively reject malicious uploads in most scenarios; meanwhile, it avoids data leakage from the server-side.","federated learning; secure-multi-party computation; differential privacy","en","master thesis","","","","","","","","","","","","","",""
"uuid:f6e0bcde-b7cb-4685-add0-40aab03a1e18","http://resolver.tudelft.nl/uuid:f6e0bcde-b7cb-4685-add0-40aab03a1e18","Investigating scaling techniques and the cost-efficiency of distributed to single FPGA compositions for Full Waveform Inversion","Dierick, Luc (TU Delft Electrical Engineering, Mathematics and Computer Science; TU Delft Computer Engineering)","Al-Ars, Z. (mentor); Petri-König, J. (graduation committee); Zuniga, Marco (graduation committee); Gunay, Gokhan (graduation committee); Delft University of Technology (degree granting institution)","2022","In recent years, the big data era has produced an increasing volume and complexity of data that requires processing. To analyze and process these large amounts of data, applications are being scaled on large clusters using distributed data processing frameworks. A more recent trend utilizes hardware accelerators to offload computationally intensive tasks and reduce compute time and energy consumption. As a result, a rapid growth of data center deployment containing heterogeneous compute infrastructures is observed. Alternative to the more commonly used general-purpose GPUs (GPGPUS), the field programmable gate array (FPGA) is becoming an increasingly popular choice of accelerator. Its effectiveness to accelerate highly parallel applications in combination with the flexibility due to its reconfigurable nature make it well suited for a wide range of applications. As a spatial compute resource, the problem size a single FPGA can process is bounded by the available programmable logic and memory. However, applications that do not require the full resources of an FPGA can be vertically scaled by instantiating multiple instances of the hardware design on a single node. A barrier in the adoption of FPGAs is formed by the complexity of hardware design which requires in depth hardware-specific expertise. Additionally, integrating FPGAs in distributed data processing frameworks is a challenge on itself.
These challenges are being addressed in two directions. High level synthesis (HLS) tools and compilers are being developed to decrease the complexity of hardware design by allowing users to develop FPGA designs in high level languages. Additionally, there is an increased availability of ready-to-use FPGA designs for common applications in hardware libraries such as Vitis libraries.
To aid the adoption of FPGAs and improve their accessibility, this work presents OctoRay: a python framework with a focus on ease-of-use that allows users to flexibly and transparently scale applications both vertically and horizontally on FPGA clusters. Scaling a binarized convolutional neural network (CNN) with OctoRay resulted in performance improvements linear to the number of nodes, or copied instances applied. The framework was also used to analyze the cost-efficiency of a cluster of low-end PYNQ-Z1 FPGAs compared to a data center class Alveo U280 FPGA. A partly in hardware accelerated implementation of Full Waveform Inversion (FWI), a seismic imaging algorithm, was developed and used to conduct the investigation. It was concluded that 32 PYNQ-Z1s are required to match the performance of a single Alveo U280 FPGA. An important bottleneck in the performance of the PYNQ-Z1s was the low-performance host processor on which a significant portion of FWI was executed. The small number of resources available on a PYNQ-Z1 limited the attainable accuracy of FWI to a bare minimum. The FWI hardware design with the same specifications made for the high-end FPGA only utilized a fraction of its resources, far from harnessing its full potential. It was concluded that, unlike FWI, applications that do not require the abundance of resources a high-end FPGA offers, but do benefit from rapid development cycles and low energy consumption are suited for a distributed low-end FPGA composition.","Scalability; FPGA; Full Waveform Inversion","en","master thesis","","","","","","","","","","","","Computer Engineering","",""
"uuid:615a9965-3685-439e-8599-9c913b9902da","http://resolver.tudelft.nl/uuid:615a9965-3685-439e-8599-9c913b9902da","A Survey on Accelerating Sparse CNN Inference on GPUs","Chen, Qilin (TU Delft Electrical Engineering, Mathematics and Computer Science)","Mohamed, Hasan (mentor); Liu, Shih-Chii (mentor); Tömen, N. (mentor); Zuniga, Marco (graduation committee); Delft University of Technology (degree granting institution)","2022","Convolutional neural networks (CNNs) are often pruned to achieve faster training and inference speed while also requiring less memory. Nevertheless, during computation, most modern GPUs cannot take advantage of the sparsity automatically, especially on networks with unstructured sparsity. Therefore, many libraries that exploit sparsity, have been proposed for accelerating CNN inference on GPUs. However, there is little research on systematically comparing them. In this paper, some state-of-the-art libraries for accelerating sparse CNN inference on GPUs are reviewed and benchmarked. Most of the libraries speedup the convolution and/or pooling operations by skipping zero calculations, therefore, they are able to perform sparse matrix calculations faster. However, many of them have hardware and software restrictions and are hard to integrate into a new model to perform end-to-end inference.","Convolutional Neural Networks (CNNs); Sparsity; Accelerators; Inference","en","bachelor thesis","","","","","","","","","","","","Computer Science and Engineering","CSE3000 Research Project",""
"uuid:6b5ff97f-ebfb-49ad-9f37-4ff4c3a6ad22","http://resolver.tudelft.nl/uuid:6b5ff97f-ebfb-49ad-9f37-4ff4c3a6ad22","Increasing security of an e-auction smart contracts with Intel SGX trusted hardware","Deshamudre, Rohan (TU Delft Electrical Engineering, Mathematics and Computer Science)","Liang, K. (mentor); Zuniga, Marco (graduation committee); Delft University of Technology (degree granting institution)","2022","Smart contracts allow for the collaboration and transaction processes between multiple parties/organisations to be automated and conducted in a neutral environment. In many situations these agreements are confidential and running a smart contract that contains private/sensitive information on a public blockchain network which is transparent and shares data with all nodes is not a sensible method of execution. With the use of hyperledger fabric which is a private and permissioned network merged with Intel SGX trusted execution environments, a feasible solution can be proposed to tackle this problem.
This paper first analyses some of the security issues faced by smart contracts on hyperledger fabric, compares the current solutions for blockchain based e-auction systems and discusses security measures such as encryption, attestation and combination with different trusted hardware modules.
The second part proposes a solution to combine Intel SGX trusted hardware with a e-auctions fabric smart contract, discusses the architecture and step by step implementation of a prototype, explains the security enhancements of this method by comparing just hyperledger vs the new method, and lastly outlines the limitations and future extensions. This solution mitigates many vulnerabilities by isolating execution of chaincode with sensitive data in a TEE and minimizing the trusted computing base to reduce latency and overhead.","Smart Contracts; Intel SGX; Hyperledger Fabric","en","bachelor thesis","","","","","","","","","","","","Computer Science and Engineering","CSE3000 Research Project",""
"uuid:5def242c-0440-4601-bb0e-34e62343d81b","http://resolver.tudelft.nl/uuid:5def242c-0440-4601-bb0e-34e62343d81b","Creating a TPM based smart contract for the Medical Supply Chain in Hyperledger Fabric","Nanhekhan, Kevin (TU Delft Electrical Engineering, Mathematics and Computer Science)","Liang, K. (mentor); Zuniga, Marco (graduation committee); Delft University of Technology (degree granting institution)","2022","Smart contracts play an important role within the blockchain by ensuring that valid transactions are being recorded. However, there are critical concerns regarding the security and privacy of data within these blockchain applications. This research provides information on how the integration of the Trusted Platform Module can achieve more security in a blockchain application built on Hyperledger Fabric. Despite the implemented prototype and Trusted Platform Module functions working separately, combining the two into one working prototype was not successful. Tests analysis has been performed showing that the prototype performs worse due to higher latencies and has no security vulnerabilities. However, these are not conclusive statements, as not only the prototype did not work fully but the analysis tools were lacking. Recommendations have been made to use the described process of this research for future research and also the development of more analysis tools.","Smart Contract; Hyperledger Fabric; Trusted Platform Module; TPM; Blockchain","en","bachelor thesis","","","","","","","","","","","","Computer Science and Engineering","CSE3000 Research Project",""
"uuid:4386d065-dce4-4215-95eb-e0ad5bcfea57","http://resolver.tudelft.nl/uuid:4386d065-dce4-4215-95eb-e0ad5bcfea57","Sign Hyperledger Fabric Smart Contracts using the TPM","Starke, Zeddrich (TU Delft Electrical Engineering, Mathematics and Computer Science)","Liang, K. (mentor); Zuniga, Marco (graduation committee); Delft University of Technology (degree granting institution)","2022","Block-chain technology is gaining momentum in both industry and academics. With
this momentum there are a lot of potential gains, but also potential risk involved. This papers proposes a solution for security risks, like a man-in-the-middle-attack, of the permissioned block-chain distributed ledger software Hyperledger Fabric. A prototype is introduces that uses the Trusted Hardware Module for attestation. By combining the key hierarchy present in the TPM, to sign assets before storing them on the ledger.","","en","bachelor thesis","","","","","","","","","","","","Computer Science and Engineering","CSE3000 Research Project",""
"uuid:6a3debc5-fb0d-4c6c-8406-4e4fc67f2ea3","http://resolver.tudelft.nl/uuid:6a3debc5-fb0d-4c6c-8406-4e4fc67f2ea3","Secure smart contract attestation using Intel SGX","Chatterjee, Agniv (TU Delft Electrical Engineering, Mathematics and Computer Science)","Liang, K. (mentor); Chen, H. (mentor); Zuniga, Marco (graduation committee); Delft University of Technology (degree granting institution)","2022","Blockchain networks are increasingly recognized as a disruptive technology across sectors such as online services, finance, supply chain, administration etc. They are underpinned by smart contracts which provide programmatic instruction for the blockchain to operate. A major obstacle in the widespread adoption of blockchain technology is the security of the underlying smart contracts and potentially exploitative flaws in their technical makeup that pose a risk to data privacy. Modern trusted execution environments, such as Intel SGX, leverage hardware through process of attestation and have been proposed to preserve privacy in smart contracts; however, practical research & development in this field has seen slower progress. This paper explores the process of attestation by which Intel SGX enhances smart contract security, examines development & execution of a prototype smart contract that utilizes SGX for secure e-voting and evaluates benefits & limitations of the process. Finally, we also propose improvements to our approach and present further scope of research on the topic.","","en","bachelor thesis","","","","","","","","","","","","Computer Science and Engineering","CSE3000 Research Project",""
"uuid:bdc8259e-43e7-4e81-b573-c2f8c1442892","http://resolver.tudelft.nl/uuid:bdc8259e-43e7-4e81-b573-c2f8c1442892","RoCE based 100GbE RDMA network stack on FPGA hardware","SONG, TIANLI (TU Delft Electrical Engineering, Mathematics and Computer Science)","Al-Ars, Z. (mentor); Ahmad, T. (graduation committee); Zuniga, Marco (graduation committee); Delft University of Technology (degree granting institution)","2021","Big data analytics is one of the foundations for booming technologies such as machine learning, genetics/genomics, and computer vision. These big data applications require a large amount of data transfers for distributed and parallel processing. Networking is thus a crucial facilitator and could make big impact on big data processing.
In a computing system with a common network stack such as the TCP/IP protocol suite, many expensive memory operations are necessary to process networking traffic. This means a large percentage of CPU resources are occupied by networking rather than data processing. The memory copying overhead introduced by networking not only reduces the throughput but also increases the latency. In this case, networking is becoming a major bottleneck for big data applications. This problem can be solved by applying Remote Direct Memory Access (RDMA) technology to the network stack. RDMA enables a zero-copy mechanism and has CPU bypass ability. With RDMA implemented, both the throughput and latency can be improved.
In this work, we developed an open source 100 Gbps RDMA network stack on Field Programmable Gate Array (FPGA) hardware. The developed stack follows the RDMA over Converged Ethernet (RoCE) architecture and targets the Alveo FPGA platform. The stack includes a User kernel that can be customized for user applications. This means that computing applications can also be offloaded to this RoCE stack. Finally, we evaluate the stack and compare it with existing TCP/IP and RDMA stacks like the EasyNet and StRoM. The results show that the developed RDMA stack achieves a throughput of 100 Gbps and an RDMA READ operation latency around 4 us and an RDMA WRITE latency around 3.5 us for 64B data. It shows a great throughput advantage over the TCP/IP stack for message sizes smaller than 1 MB. The latency is also slightly lower than the TCP/IP stack.Big data analytics is one of the foundations for booming technologies such as machine learning, genetics/genomics, and computer vision. These big data applications require a large amount of data transfers for distributed and parallel processing. Networking is thus a crucial facilitator and could make big impact on big data processing.
In a computing system with a common network stack such as the TCP/IP protocol suite, many expensive memory operations are necessary to process networking traffic. This means a large percentage of CPU resources are occupied by networking rather than data processing. The memory copying overhead introduced by networking not only reduces the throughput but also increases the latency. In this case, networking is becoming a major bottleneck for big data applications. This problem can be solved by applying Remote Direct Memory Access (RDMA) technology to the network stack. RDMA enables a zero-copy mechanism and has CPU bypass ability. With RDMA implemented, both the throughput and latency can be improved.
In this work, we developed an open source 100 Gbps RDMA network stack on Field Programmable Gate Array (FPGA) hardware. The developed stack follows the RDMA over Converged Ethernet (RoCE) architecture and targets the Alveo FPGA platform. The stack includes a User kernel that can be customized for user applications. This means that computing applications can also be offloaded to this RoCE stack. Finally, we evaluate the stack and compare it with existing TCP/IP and RDMA stacks like the EasyNet and StRoM. The results show that the developed RDMA stack achieves a throughput of 100 Gbps and an RDMA READ operation latency around 4 us and an RDMA WRITE latency around 3.5 us for 64B data. It shows a great throughput advantage over the TCP/IP stack for message sizes smaller than 1 MB. The latency is also slightly lower than the TCP/IP stack.","FPGA; RDMA; Networking; 100Gbps; RoCE","en","master thesis","","","","","","","","","","","","Electrical Engineering","",""
"uuid:852f21d1-1229-4c94-ad08-ae8a0f937d2b","http://resolver.tudelft.nl/uuid:852f21d1-1229-4c94-ad08-ae8a0f937d2b","Comparing Model-Free Deep Reinforcement Learning Algorithms on Stock Market","Meral, Murat Kaan Meral (TU Delft Electrical Engineering, Mathematics and Computer Science; TU Delft Software Technology)","Neustroev, G. (mentor); de Weerdt, M.M. (graduation committee); Zuniga, Marco (graduation committee); Delft University of Technology (degree granting institution)","2021","Automated asset trading is a crucial method used by financial entities such as investment firms or hedge funds. It allows them to allocate their capital in order to maximize their rate of returns. In scientific literature, there are multiple models suggested to solve this problem. However, these models either lack the complexity to understand the market or scalability for the market in general. On the other hand, deep reinforcement learning is a great framework that can solve these problem. In this study we aim to understand the performance of model-free deep reinforcement learning algorithms in terms of training speed, financial performance and generalizability by training and comparing them on a smaller representative market. Proximal Policy Approximation (PPO) and Twin Delayed Deep Deterministic Policy Gradient (TD3) were used as a representatives of policy approximation and QLearning algorithms respectively. Our study have found that while proximal policy algorithms offer higher speed due to smaller training data they use at each timestep, Q-Learning algorithms offer a better general performance in terms of stability and generalizability. With respect to financial performance on training stocks, this study did not find a statistically important difference in performances.","Deep Reinforcement Learning; Stock Market; Policy Optimization; Q-Learning","en","bachelor thesis","","","","","","","","","","","","Computer Science and Engineering","CSE3000 Research Project",""
"uuid:a98cc32e-9d10-41a3-8fc3-c840f1a8654d","http://resolver.tudelft.nl/uuid:a98cc32e-9d10-41a3-8fc3-c840f1a8654d","Matching in Multi-Agent Pathfinding using M*","Dönszelmann, Jonathan (TU Delft Electrical Engineering, Mathematics and Computer Science; TU Delft Intelligent Systems)","Mulderij, J. (mentor); de Weerdt, M.M. (mentor); Zuniga, Marco (graduation committee); Delft University of Technology (degree granting institution)","2021","Multi-agent pathfinding (MAPF) is the process of finding collision-free paths for multiple agents. MAPF can be extended by grouping agents into teams. In a team, agents need to be assigned (or matched) to one of the team's goals such that the sum of individual cost} is minimised. This extension is called MAPF with matching (MAPFM). M* is a complete and optimal algorithm to solve MAPF problems. In this paper, two strategies are proposed which allow M* to solve MAPFM problems. These strategies are called inmatching and prematching. It is shown that prematching is generally preferable to inmatching, the benefits of different optimisations for M* are compared, and it is shown that the performance of M* performs very comparably to other A*-derived algorithms.","MAPF; MAPFM; Pathfinding; Algorithm; M*; matching","en","bachelor thesis","","","","","","","","","","","","Computer Science and Engineering","CSE3000 Research Project",""
"uuid:6065df0a-51e4-46a3-9727-b0ab68cec8fd","http://resolver.tudelft.nl/uuid:6065df0a-51e4-46a3-9727-b0ab68cec8fd","Multi-Agent Pathfinding with Matching using Increasing Cost Tree Search","van der Woude, Thom (TU Delft Electrical Engineering, Mathematics and Computer Science; TU Delft Software Technology)","Mulderij, J. (mentor); de Weerdt, M.M. (mentor); Zuniga, Marco (graduation committee); Delft University of Technology (degree granting institution)","2021","Both the assignment problem and the multi-agent pathfinding problem are common problems in the fields of robotics and transportation. The joint problem of multi-agent pathfinding extended with the assignment of goals to agents, matching, is something that has not been studied much; few methods exist today that solve it. In this work, two types of algorithms based on the Increasing Cost Tree Search (ICTS) algorithm for multi-agent pathfinding are presented that can optimally solve this joint problem: exhaustive algorithms that reduce the problem to solving many multi-agent pathfinding problems using regular ICTS, and algorithms that search a generalized increasing cost tree. These are compared to each other experimentally on a set of grid maps, and it is shown that exhaustive methods typically outperform the generalized ICTS. Lastly, an exhaustive ICTS algorithm is compared to alternative algorithms based on other multi-agent pathfinding approaches to put its performance into the broader context of algorithms for multi-agent pathfinding with matching.","Multi-Agent Pathfinding; Matching; Increasing Cost Tree Search; Exhaustive Search; Assignment","en","bachelor thesis","","","","","","","","","","","","Computer Science and Engineering","CSE3000 Research Project",""
"uuid:e73165e6-d60b-43cc-8abd-c6046f0ab638","http://resolver.tudelft.nl/uuid:e73165e6-d60b-43cc-8abd-c6046f0ab638","Adapting CBM to optimize the Sum of Costs","Baauw, Robbin (TU Delft Electrical Engineering, Mathematics and Computer Science; TU Delft Intelligent Systems)","Mulderij, J. (mentor); de Weerdt, M.M. (mentor); Zuniga, Marco (graduation committee); Delft University of Technology (degree granting institution)","2021","In the Multi-Agent Pathfinding with Matching (MAPFM) problem, agents from a team are matched with and routed towards one of their team's goals without colliding with other agents. The sum of path costs of all agents is minimized. In prior works, Conflict Based Min-Cost-Flow (CBM) has been proposed. This algorithm solves a similar problem that instead minimizes the maximum path length. In this paper, an extension upon CBM is presented, called CBMxSOC. It consists of several changes to CBM that allow it to minimize the sum of path costs. CBMxSOC is experimentally compared to other MAPFM solvers and is shown to be able to scale to many agents when there are few conflicts between different teams.
Embeddedreal-time systems that have cost or energy constraints are usually limited inprocessing power and memory. This limitation typically leads to applyingsimpler execution models such as non-preemptive scheduling. A problem with anon-preemptive real-time system is that finding a schedule without causingdeadline misses is NP-hard. Finding a schedule is therefore done with online schedulingpolicies, i.e. policies that make decisions upon arrival or after execution ofa task to schedule the next one. Onlinenon-preemptive scheduling policies are typically priority based, namely, theypick the highest priority job to schedule based on criteria as period or absolutedeadline. These are work-conserving policies which cannot keep the processoridle while there are still pending jobs in the ready queue. Non-work-conservingpolicies allow an idle cpu while there are still jobs in the ready queue. Whilethis increases schedulability, it also has an increased overhead. Current stateof the art policies like Precautious Rate Monotonic (PRM) and Critical WindowsEarliest Deadline First (CW-EDF) have an idle-time insertion policy which caninsert an idle time in the schedule (between the execution of the jobs) whilestill having pending jobs. PRM verifies whether the highest priority job in theready queue is able to finish without causing a deadline miss for the highestpriority task in the system, which is the task with the lowest period. PRMimproves schedulability with increased overhead of O(1).CW-EDF comes with additional overhead O(n log n) in which it verifies ifscheduling the highest priority pending job will result in a deadline miss for thenext job of all other tasks in the system. PRM has low overhead, while CW-EDFhas better schedulability despite having higher overhead. Alimitation of these non-work-conserving policies is the missing support for event-triggeredtask, where jobs are released at unknown time instants. Namely, the existingnon-work-conserving policies are designed for strict periodic tasks. In thisthesis we will introduce a policy which has schedulability as high as CW-EDF,but prior to running the system it detects the critical tasks which have aninfluence on the idle-time insertion policy. We ignore the non-critical tasks duringthe idle-time insertion policy, reducing the runtime overhead. We also introducea policy which will support time-triggered tasks, by using an arrival curvewhich stores the possible behavior of the event-triggered task. With this arrivalcurve we will reduce the number of deadline misses compared to the existingnon-work-conserving policies. ","real-time systems; non-work-conserving; event-triggered; arrival curve","en","master thesis","","","","","","","","","","","","Electrical Engineering | Embedded Systems","",""
"uuid:bcfb9bb6-a21e-4f0b-b98d-5410e399ff34","http://resolver.tudelft.nl/uuid:bcfb9bb6-a21e-4f0b-b98d-5410e399ff34","Modeling Inference Time of Deep Neural Networks on Memory-constrained Systems","Brouwer, Hans (TU Delft Electrical Engineering, Mathematics and Computer Science)","Chen, Lydia Y. (mentor); Ghiassi, S. (graduation committee); Cox, B.A. (graduation committee); Zuniga, Marco (graduation committee); Delft University of Technology (degree granting institution)","2020","Deep neural networks have revolutionized multiple fields within computer science. It is important to have a comprehensive understanding of the memory requirements and performance of deep networks on low-resource systems. While there have been efforts to this end, the effects of severe memory limits and heavy swapping are understudied. We have profiled multiple deep networks under varying memory restrictions and on different hardware. Using this data, we develop two modeling approaches to predict the execution time of a network based on a description of its layers and the available memory. The first modeling approach is based on engineering predictive features through a theoretical analysis of the computations required to execute a layer. The second approach uses a LASSO regression to select predictive features from an expanded set of predictors. Both approaches achieve a mean absolute percentage error of 5% on log-transformed data, but suffer degraded performance on transformation of predictions back to regular space.","Performance Analysis; Deep Neural Networks; Inference; Memory-constrained; Machine Learning; Prediction; Modeling","en","bachelor thesis","","","","","","","","","","","","Computer Science and Engineering","CSE3000 Research Project",""
"uuid:5eb53059-d4bf-4a39-92ae-1bfbfa8c4f64","http://resolver.tudelft.nl/uuid:5eb53059-d4bf-4a39-92ae-1bfbfa8c4f64","Offline Compression of Convolutional Neural Networks on Edge Devices","Tulling, Simon (TU Delft Electrical Engineering, Mathematics and Computer Science; TU Delft Distributed Systems)","Chen, Lydia Y. (mentor); Ghiassi, S. (graduation committee); Cox, B.A. (graduation committee); Zuniga, Marco (graduation committee); Delft University of Technology (degree granting institution)","2020","Edge Devices and Artificial Intelligence are important and ever increasing fields in technology. Yet their combination is lacking because the neural networks used in AI are being made increasingly large and complex while edge devices lack the resources to keep up with these developments. Neural network model compression will allow these edge devices to run these models due to overcoming memory constraints. This paper proposes to use both singular value decomposition and canonical polyadic decomposition as a way to decrease the size of convolutional neural networks at the cost of some accuracy. This compression pipeline can be run on an edge device and is configurable to change the trade-off between file size and accuracy. This creates a possibility to run convolutional neural networks natively on edge devices.","","en","bachelor thesis","","","","","","","","","","","","Computer Science and Engineering","CSE3000 Research Project",""
"uuid:758d345d-ecdf-478e-a534-a23300dbe877","http://resolver.tudelft.nl/uuid:758d345d-ecdf-478e-a534-a23300dbe877","Detection and Tracking of a Fast-Moving Object in Squash using a Low-Cost Approach","Sachdeva, Saumil (TU Delft Electrical Engineering, Mathematics and Computer Science; TU Delft Computer Engineering)","van Genderen, A.J. (mentor); Wong, J.S.S.M. (mentor); Zuniga, Marco (graduation committee); Delft University of Technology (degree granting institution)","2019","Advancement in technology has given rise to the need for technology to be utilized in extracting meaningful game-play information from sports. To do so in a ball-game like squash, the prime objective is to perform ball-tracking in an adequate and efficient manner. In squash, ball-tracking is complex due to the small size of the ball, the high-speed movement and constant occlusion due to the continuous movement of the players. The current state-of-the-art ball tracking methods use high-speed cameras along with high-computation power resources to solve these problems in similar sports such as tennis. The aim of this thesis is to solve the challenges in ball-tracking for squash using a low-cost approach with low-computation power resources and a single camera view. A ball-detection system with a high-accuracy and a ball-tracking system which can optimally tackle the problem of occlusion is developed using computer-vision techniques and by utilizing the cues from the game itself. The implementation is carried out on a Raspberry-Pi which is characterized as a low-computation platform with an Arm Cortex-A53 processor. The results show that the tracking-problem can be solved using a low-cost approach for the challenging scenarios that are present in squash. The 2D trajectory of the ball generated as a result can be used for various applications such as line-calling, shot analysis and game analysis.","Squash; Ball-Tracking; Computer Vision; Low-Cost","en","master thesis","","","","","","","","","","","","Electrical Engineering | Embedded Systems","",""
"uuid:29c4b96e-ffdb-4136-81d7-ca9822b0fa0b","http://resolver.tudelft.nl/uuid:29c4b96e-ffdb-4136-81d7-ca9822b0fa0b","Novel Interaction Method for UHF RFID Tags","Ketel, Thijmen (TU Delft Electrical Engineering, Mathematics and Computer Science)","Pawełczak, Przemysław (mentor); Zuniga, Marco (graduation committee); Langendoen, K.G. (graduation committee); Bozzon, A. (graduation committee); Delft University of Technology (degree granting institution)","2019","RFID technology is slowly replacing traditional bar codes as a way to identify and track objects and individuals. However, consumer-oriented market penetration has been limited as dedicated RFID readers carry a high start-up cost. Furthermore, interactions with individual tags require special-purpose RFID readers. We present a novel RFID tag interaction system based on commercially of the shelf hardware at vastly lower cost compared to conventional systems.","RFID; Embedded systems; Tag Interaction","en","master thesis","","","","","","","","2024-07-04","","","","Electrical Engineering | Embedded Systems","",""
"uuid:e87b1b45-d291-432f-8e8b-f65b859255c9","http://resolver.tudelft.nl/uuid:e87b1b45-d291-432f-8e8b-f65b859255c9","3D Gradient Printing of Energetic Multi-Materials","Rijnders, Bart (TU Delft Electrical Engineering, Mathematics and Computer Science)","Langendoen, K.G. (mentor); Straathof, Michiel (mentor); Zuniga, Marco (graduation committee); Rellermeyer, Jan S. (graduation committee); Delft University of Technology (degree granting institution)","2019","The performance of gun and rocket propellants, which consist of energetic materials, is largely determined by their geometry and composition. Con- ventional production methods limit the performance by putting constraints on both. With additive manufacturing, or 3D-printing, there are signi- cantly fewer geometry constraints and together with the ability to combine multiple materials into a continuous gradient new performance optimization opportunities are created. In the current 3D-printing world it is possible to print single-material or discrete gradient multi-material objects by trans- lating a CAD model to printer instructions. This translation is done with slicer software that slices a 3D-model and outputs printer instructions in a G-Code le. This thesis looks at how an object with a continuous gradient can be printed. A modied version of the popular Cura slicing software is presented that can apply an approximation of a continuous gradient to an input model. The printer paths are simulated with the slicer software and ultimately printed using TNO's multi-material 3D-printer. While the print results show that energetic materials behave in such a dierent way than normal plastics that 3D-printing them it is not an easy task, printing a 3D-model with a multi-material continuous gradient is certainly viable.","3D-printing; Additive Manufacturing; Energetic Materials; Multi-Material Additive Manufacturing","en","master thesis","","","","","","","","2021-07-01","","","","Electrical Engineering | Embedded Systems","",""
"uuid:1d83f7e6-a0ba-45b9-b696-8d858d5dabca","http://resolver.tudelft.nl/uuid:1d83f7e6-a0ba-45b9-b696-8d858d5dabca","Command Recognition on Intermittently-Powered Devices","Schilder, Patrick (TU Delft Electrical Engineering, Mathematics and Computer Science; TU Delft Embedded and Networked Systems)","Langendoen, K.G. (mentor); Wong, J.S.S.M. (graduation committee); Zuniga, Marco (graduation committee); Majid, A.Y. (graduation committee); Delft University of Technology (degree granting institution)","2019","The Internet of Things (IoT) is expected to include billions of tiny devices that collect, process, and communicate sensory data. As of now, batteries power these devices. Batteries, however, are large, expensive, and short-lived - even the rechargeable ones wear out in a few years. Therefore, they are not a sustainable powering solution. Tiny battery-less devices promise a maintenance-free and environment-friendly alternative. They operate by harvesting energy from the environment. Ambient power, however, is marginal and unpredictable. This causes tiny energy-harvesting devices to operate intermittently, violating the requirements of many real-world applications.
This work presents an event-based command-recognition algorithm tailored towards battery-less sensors, taking into account the challenges of intermittent execution and the ultra-low-power hardware. Our algorithm achieves a 97% recognition accuracy with a ten-word vocabulary.","Speech recognition; Energy Harvesting; Embedded Systems","en","master thesis","","","","","","","","2021-05-09","","","","Electrical Engineering | Embedded Systems","",""
"uuid:ac423080-36f4-4005-b266-82b01ed1df81","http://resolver.tudelft.nl/uuid:ac423080-36f4-4005-b266-82b01ed1df81","Design and implementation of a LoRa based Gateway","Dixit, Shubhankar (TU Delft Electrical Engineering, Mathematics and Computer Science; TU Delft Quantum & Computer Engineering)","van Genderen, A.J. (mentor); Cotofana, S.D. (graduation committee); Strietman, Gertjan (graduation committee); Zuniga, Marco (graduation committee); Snippe, Harm Wouter (graduation committee); Delft University of Technology (degree granting institution)","2017","The project focuses on the hardware-software co design of a LoRaWAN based industrial IoT gateway used for proprietary applications. Long Range Wide Area Network, abbreviated as LoRaWAN is a network and data layer running over the LoRa PHY layer which operates at 868 MHz[29]. The surge in LoRa has led big market players like SemTech to licence devices operating over a free network. Also gateway manufacturers have seized on the opportunity of this growing market and 2 industrial gateways, [15], [27] have captured most of the market. In order to break this monopoly, FactoryLab B.V, an Industrial IoT company from Zwijndrecht, The Netherlands has developed a low cost Linux based gateway which can be used for proprietary applications. The project aims at developing and comparing the gateway with industry standards. Although the hardware for the gateway couldn’t be tested in time for the finalization of this report, various tests are performed using an improvised hardware setup which emulates the FactoryLab hardware and the results approximated and compared to the industrial gateways. The Range to Cost ratio for the test setup was calculated to be 4.8 meters/euro and when pitched against the other gateways, showed a maximum increase of 26.6%.","LoRa; LoRaWAN; Gateway; Linux; development; range; optmization","en","master thesis","","","","","","","","","","","","Electrical Engineering | Embedded Systems","",""