Theses in Progress

Aufbau und Inbetriebnahme eines automatisierten Testprüfstandes im Gesamtsystemnetz.

The aim of this thesis is to study decoding of tailbiting convolutional codes, a class of codes suited for short blocklengths. We address their efficient decoding as well as applications to communication systems with feedback.

We investigate OFDM under coarse quantization and non-linear power amplification.

In this thesis, the student will study and summarize the most recent code-based digital signature schemes based on the syndrome decoding problem

Identification is a communication scheme that allows rate doubly exponential in the blocklemght, with the tradeoff that identities cannot be decoded (as messages do) but can only be verified.

• https://ieeexplore.ieee.org/document/42172
• https://ieeexplore.ieee.org/document/42173

The double exponential growth presents various challenges in the finite regime: there are heavy computational costs introduced at the encoder and decoder and heavy trade-offs between the error and the codes sizes.

The ultimate goal is to find a fast, reliable implementation while still achieving large code sizes.

Identification codes can be achieved by first removing the errors from the channel with regular transmission channel coding, and then sending a challenge though the corrected channel. For every identity i, The channenge is generated by picking a random input m and computing the corresponding output T_i(m) using a function T_i that depends on the identity. The challenge is then the pair m,T_i(m) and the receiver wanting to verify an identity j will verify whether j=i by testing the challenge. This is done by recomputing the output with T_j and verifying whether T_j(m)= T_i(m). The errors are reduced by ensuring that the various functions collide on a small fraction of the possible inputs.

It turns out that choosing good sets of funtions {T_i} is the same as choosing error-correction codes {c_i} with large distance, where now each codeword c_i defines a function by mapping positions m (sometimes called code locators) to symbols c_im of the codeword.
We can thus construct identification codes by choosing error-correction codes where we are only interested in the performance of the error correction encoders (we are not interested in the error-correction decoder or error-correction codes).

Your task will be implementing the identification codes described in

• https://ieeexplore.ieee.org/abstract/document/782144

aiming at the fastest implementation, and testing their performance in comparison to other current implementations.

For reference, our previous work on identification based on Reed-Solomon and Reed-Muller code can be found at

• https://arxiv.org/abs/2107.07649
• https://arxiv.org/abs/2107.06801
• https://arxiv.org/abs/2007.06372

The coding will be in Python/Sagemath.
The working language will be in English.

Environment: we collaborate with LTI. At LNT and LTI there is currently a lot of funding for research in identification. Therefore you will find a large group of people that might be available for discussion and collaboration.

Federated learning is a machine learning paradigm where decentralized entities (clients) collaboratively learn using their private data. A central server acts as a coordinator of the learning process. Due to the sensitivity of the private data involved,  the data cannot be transferred. A salient problem of federated learning is the presence of malicious clients, which are clients that try to destroy the learning process. Malicious clients can do this by corrupting their data and/or by modifying their local model updates. The goal of this project is to understand how model poisoning attacks and defense strategies perform under different scenarios of federated learning using experiments. 

References: 

[1]- https://www.ndss-symposium.org/wp-content/uploads/ndss2021_6C-3_24498_paper.pdf

[2]- https://arxiv.org/pdf/1903.03936.pdf

[3]- https://arxiv.org/pdf/2304.00160.pdf

• Basic knowledge of machine learning
• Python programming skills, knowledge of PyTorch is an advantage 

DiVincenzo in 2000, provides a comprehensive framework for identifying the necessary requirements to achieve successful quantum computation. In this project, we search for ideal conditions necessary for buidling distributed quantum devices. 

In this thesis, the student will study the mathematics of linear codes and how they can be used to design cryptosystems. 

This thesis aims to provide a novel approach for all-digital clock synchronization aimed at free-space optical communication systems by combining Lee algorithm, a feed-forward timing error detector, with a Kalman filter followed by an interpolator. The approach combines information from the channel conditions under which the timing error measurement was made with the estimate based on physical model and previous measurements. 

DNA Data Storage

Data storage on DNA molecules is a promising approach for archiving massive data.

In classical DNA storage systems, binary information is encoded into sequences consisting of the four DNA bases {A, C, G, T}. The encoded sequences are used to generate DNA molecules called strands using the biochemical process of DNA synthesis. The synthesized strands are stored together in a tube. To retrieve the binary information, the strand must be read via DNA sequencing and decoded back into the binary representation.

The synthesis and the sequencing procedures are error-prone, and with the natural degradation of DNA they introduce errors to the DNA strands. To ensure data reliability, the errors have to be corrected by algorithms and error-correcting codes (ECCs).

A 5min video with an overview of DNA storage: https://youtu.be/r8qWc9X4f6k?si=Yzm5sOW-a6VDnBu3

Composite DNA

Recently, to allow higher potential information capacity, [1,2] introduced the DNA composite synthesis method. In this method, the multiple copies created by the standard DNA synthesis method are utilized to create composite DNA symbols, defined by a mixture of DNA bases and their ratios in a specific position of the strands. By defining different mixtures and ratios, the alphabet can be extended to have more than 4 symbols. More formally, a composite DNA symbol in a specific position can be abstracted as a quartet of probabilities {p_A, p_C, p_G, p_T}, in which p_X, 0 ≤ p_X ≤ 1, is the fraction of the base X in {A, C, G, T} in the mixture and p_A+p_C+ p_G+ p_T =1. Thus, to identify composite symbols it is required to sequence multiple reads and then to estimate p_A, p_C, p_G, p_T in each position.

Problem description

ECCs for DNA data storage differ in many aspects from classical error correction codes. In this model, new error type gain relevance, like deletions and insertions which affect the synchronization of the sequences. Especially for composite DNA data storage, these error types received only little attention.

The most related work to this problem was recently studied by Zhang et al. in [6]. The authors initiated the study of error-correcting codes for DNA composite. They considered an error model for composite symbols, which assumes that errors occur in at most t symbols, and their magnitude is limited by l. They presented several code constructions as well as bounds for this model. In this thesis, we want to analyse a different way to model the composite synthesis method and studies additional error models. We already have some results for substitution and single deletion errors. This thesis should focus on evaluating more error models in the channel model. 

This should only roughly introduce the problem. No need to review all references. If you are interested, please reach out to me, and we can discuss a suitable direction for you.

References

[1] L. Anavy, I. Vaknin, O. Atar, R. Amit, and Z. Yakhini, “Data storage in DNA with fewer synthesis cycles using composite DNA letters,” Nat Biotechnol, vol. 37, no. 10, pp. 1229–1236, Oct. 2019, doi: 10.1038/s41587-019-0240-x.

[2] Y. Choi et al., “High information capacity DNA-based data storage with augmented encoding characters using degenerate bases,” Sci Rep, vol. 9, no. 1, Art. no. 1, Apr. 2019, doi: 10.1038/s41598-019-43105-w.

[3] V. Guruswami and J. Håstad, “Explicit Two-Deletion Codes With Redundancy Matching the Existential Bound,” IEEE Transactions on Information Theory, vol. 67, no. 10, pp. 6384–6394, Oct. 2021, doi: 10.1109/TIT.2021.3069446.

[4] J. Sima, N. Raviv, and J. Bruck, “Two Deletion Correcting Codes From Indicator Vectors,” IEEE Trans. Inform. Theory, vol. 66, no. 4, pp. 2375–2391, Apr. 2020, doi: 10.1109/TIT.2019.2950290.

[5] I. Smagloy, L. Welter, A. Wachter-Zeh, and E. Yaakobi, “Single-Deletion Single-Substitution Correcting Codes,” IEEE Transactions on Information Theory, pp. 1–1, 2023, doi: 10.1109/TIT.2023.3319088.

[6] W. Zhang, Z. Chen, and Z. Wang, “Limited-Magnitude Error Correction for Probability Vectors in DNA Storage,” in ICC 2022 - IEEE International Conference on Communications, Seoul, Korea, Republic of: IEEE, May 2022, pp. 3460–3465. doi: 10.1109/ICC45855.2022.9838471.

Machine learning exploits large collections of data to train parameterized models for a dedicated task, e.g., classification or regression. In mobile radio networks, valuable training data is located at the network edge and distributed among the network users. Transferring the data to a central server for the model training is often not desired due to the high communication load and privacy restrictions about the users’ data. The
federated learning paradigm [1] approaches this challenges by training the model locally at the users and sending only the model updates to the central server.

When executed over wireless channels, federated learning faces several challenges: scarce wireless radio resources, varying channel coditions, and user dropouts for instance. Moreover, mobile devices have a limited energy budget that they can invest into the federated learning procedure. Therefore, prior work has targeted the optimization of learning performance through user selection and resource allocation [2, 3] or the joint optimization of computation and communication for energy-efficiency [4, 5]. Although these solutions showed that the learning performance and energy-efficiency can be improved by sophisticated design, they rely on numerical solvers with many degrees of freedom. Furthermore, the focus in the literature lies on synchronous model updates. In particular, the exploration of asynchronous model updates as in [6] is underrepresented,
although this has been shown to outperform synchronous protocols in some cases [7].

The goal of this thesis is to formulate the system model and design, implement, and analyze analytical solutions for improving the efficiency of federated learning over wireless channels with energy-constrained devices.

[1] B. McMahan, E. Moore, D. Ramage, S. Hampson, and B. A. y. Arcas, “Communication-Efficient Learning of Deep Networks from Decentralized Data,” in Proceedings of the 20th International Conference on Artificial Intelligence and Statistics, ser. Proceedings of Machine Learning Research, vol. 54., 2017, pp. 1273–1282.
[2] M. M. Amiri, D. Gündüuz, S. R. Kulkarni, and H. V. Poor, “Convergence of update aware device scheduling for federated learning at the wireless edge,” IEEE Transactions on Wireless Communications, vol. 20, no. 6, pp. 3643–3658, 2021.
[3] M. Chen, Z. Yang, W. Saad, C. Yin, H. V. Poor, and S. Cui, “A joint learning and communications framework for federated learning over wireless networks,” IEEE Transactions on Wireless Communications, vol. 20, no. 1, pp. 269–283, 2021.
[4] Z. Yang, M. Chen, W. Saad, C. S. Hong, and M. Shikh-Bahaei, “Energy efficient federated learning over wireless communication networks,” IEEE Transactions on Wireless Communications, vol. 20, no. 3, pp. 1935–1949, 2021.
[5] X. Mo and J. Xu, “Energy-efficient federated edge learning with joint communication and computation design,” Journal of Communications and Information Networks, vol. 6, no. 2, pp. 110–124, 2021.
[6] Z. Chen, W. Yi, Y. Liu, and A. Nallanathan, “Robust federated learning for unreliable and resource-limited wireless networks,” IEEE Transactions on Wireless Communications, pp. 1–1, 2024.
[7] S. Dutta, J. Wang, and G. Joshi, “Slow and stale gradients can win the race,” IEEE Journal on Selected Areas in Information Theory, vol. 2, no. 3, pp. 1012–1024, 2021.

• Probability theory
• Prior experience in programming with PyTorch
• Federated learning

We are interested in developing SIC schemes that require only one code per stage.

The aim of this thesis is to enhance decoding algorithms for precoded polar product codes. We make use of successive cancellation list (SCL) decoding to generate reliability information and scale it based on the maximization of generalized mutual information.

In this thesis, we want to investigate the list decoding complexity of random (linear) codes in the sum-rank metric.

List decoding is a technique to decode beyond the unique decoding radius of a code at the cost of obtaining a list of candidate solutions. The sum-rank metric [1] is a relatively novel metric where the weight of a vector is given by the sum of the ranks of its component blocks.

As a starting point, the student should familiarize themselves with the concept of the sum-rank metric. Then, the list decoding behavior of a random SR code should be investigated, perhaps along the lines of these papers [2,3] that have some similar results on random rank metric codes. It would also be nice to investigate if this other technique [4] can be applied to the sum-rank metric.

Resources:

[1] https://arxiv.org/pdf/2102.02244 (this is not the paper where this metric was first studied, but it has a very nice overview of existing results)

[2] https://arxiv.org/abs/1401.2693

[3] https://arxiv.org/abs/1710.11516

[4] https://arxiv.org/abs/1704.02420

Channel coding lecture or similar (i.e., basics of linear codes and their decoding)

strong background in linear algebra

An interest in combinatorics is beneficial, it is at the core of many of the related papers

.

We investigate code construction algorithms for SC list decoding of polar codes [1] based on methods like [2].

• [1] https://doi.org/10.1109/TIT.2015.2410251
• [2] https://doi.org/10.1109/TCOMM.2021.3120274

This work shall deal with post-quantum secure schemes utilizing the Lee metric. The student's task is to get familiar with this metric and to design a signature scheme. The student shall show that known attacks in the Hamming metric are not applicable for the Lee metric.

Channel Coding

Compared with traditional frame-based cameras, an event-based camera has the advantages of low latency, high dynamic range, (almost) no motion blur, etc., and can respond fast to a brightness change at the image plane with thresholds determined by the previous state of brightness. It can generate event data of structural features without signal processing, such as edges and corners, which saves time, energy and computing effort.

However, it still lacks standard processes for analyzing, characterizing and evaluating event-based camera information. Moreover, data acquisition of this camera demands changes in brightness on a respective pixel. This can be introduced artificially by relative movement of the camera to the object. This might miss out part of the data due to linear translation along structural elements like edges or introduce some error due to error motion or vibration.

This thesis aims to ensure safety and quality when implementing event-based cameras in the field of industrial inspection, by dedicated experiments and related discussion of results. Event based camera imageswill be characterized and evaluated. New methods for event generation and signal processing will be proposed which will make use of the special characteristics of event-based cameras and their special characteristics arising from their working principle.

In theory, the check nodes operations of belief propagation rely on tanh() and arctanh() functions which require high computational power. Hence, in most of the practical applications, an approximation called “MinSum” is used. This method aims to exploit the structure of tanh() function where the absolute value of output sufficiently converges to 1 with increasing absolute value of input. Hence, in a series multiplication of absolute outputs, the most dominant effect comes from the minimum element and other contributions are considered negligible. However, neglection of other attenuation factors cause overcalculated outputs in “MinSum” algorithm which can accumulated by multiple iterations. This drawback can be compensated through adding attenuation or/and offset factors. These factors are mostly iteration specific and intuitively determined, which means one factor which is determined by educated guess is applied to all leaving edges. However, every edge in an unfolded Tanner graph has its own unique identity corresponding to the previous nodes and edges that the message is transmitted.

In addition to approach aiming to close the performance gap between main algorithm and “MinSum ” approximation, we can intend to improve the qualities of main algorithm. Even though belief propagation decoding in LDPC codes is considered as highly successful, it is still a “suboptimal” method compared to very expensive but accurate Maximum A posteriori Probability (MAP) estimation. It means there might be some room for improvement in performance. Additionally, belief propagation requires multiple iterations to converge and the required number of iterations can dramatically increase by decreasing signal-to-noise ratio (SNR). Additional correction weights imposed on iterated messages can be a candidate to improve performance in overall.

5G specification for channel coding is using protograph based LDPC codes. Every node duplicated from same base matrix node is keen to show similar properties, it may be possible to use same weights for these nodes by preserving the good decoding results. This detail can help us to using additional correction weights by minimum additional memory burden.

In this project the student should calculate the zero-error capacity for 

a multi-user model with feedback.

Information theory

A homomorphic encryption (HE) scheme enables performing operations on plaintexts in their encrypted form. [1] A somewhat HE scheme supports both additions and multiplication, however up to a certain number of either of them. In comparison, a fully HE scheme supports an infinite number of additions and multiplications to be performed.

Nowadays, many of the efficient HE schemes are based on lattices, i.e. the Euclidean metric. Although different metrics can be used to construct HE schemes, such as the rank metric and the Hamming metric.

The goal of this research internship is to investigate the somewhat HE scheme presented in [1], which is based on the rank metric. Following that, the student should compare the scheme to other schemes based on other metrics, such as the one in [2] based on the Hamming metric, and the ones in [3], [4], [5], based on the Euclidean metric.

References:

[1] C. Aguilar-Melchor, V. Dyseryn, and P. Gaborit, ‘Somewhat Homomorphic Encryption based on Random Codes’, 2023, 2023/1798. Accessed: Oct. 09, 2024. [Online]. Available: https://eprint.iacr.org/2023/1798

[2] H. Corrigan-Gibbs, A. Henzinger, Y. Kalai, and V. Vaikuntanathan, ‘Somewhat Homomorphic Encryption from Linear Homomorphism and Sparse LPN’, 2024, 2024/1760. Accessed: Nov. 08, 2024. [Online]. Available: https://eprint.iacr.org/2024/1760

[3] Z. Brakerski and V. Vaikuntanathan, ‘Efficient Fully Homomorphic Encryption from (Standard) LWE’, in 2011 IEEE 52nd Annual Symposium on Foundations of Computer Science, Palm Springs, CA, USA: IEEE, Oct. 2011, pp. 97–106. doi: 10.1109/FOCS.2011.12.

[4] L. Ducas and D. Micciancio, ‘FHEW: Bootstrapping Homomorphic Encryption in Less Than a Second’, in Advances in Cryptology -- EUROCRYPT 2015, E. Oswald and M. Fischlin, Eds., Berlin, Heidelberg: Springer, 2015, pp. 617–640. doi: 10.1007/978-3-662-46800-5_24.

[5] I. Chillotti, N. Gama, M. Georgieva, and M. Izabachène, ‘TFHE: Fast Fully Homomorphic Encryption over the Torus’, 2018, 2018/421. Accessed: Nov. 08, 2024. [Online]. Available: https://eprint.iacr.org/2018/421

We explore alternative encoding schemes for polar shaping.

Non-volatile memories (NVMs) are electronic data-storage technologies that do not require a continuous power supply to retain data; unlike traditional magnetic or optical media, they do not utilize mechanically movable components and can therefore offer better performance, and allow for three-dimensional scaling of storage devices. Under most realistic workloads, they also offer better energy efficiency.

However, these technologies also feature imbalances in behavior, performance and consequences, between the processes of reading data and writing it. To wit, in memory cells which represent data by the level of held charge (traditionally allowing for representation of several logical levels), the process of charge-injection is a simple and efficient, whereas charge-depletion is both technically complex (requiring the depletion of entire blocks of cells) and destructive, a main driver of cell-degradation over the device's life cycle.

Different coding theoretic approaches have been explored to alleviate this imbalance, including coding schemes that delay charge-depletion cycles [1], [2]. This project will build on the work done in [2], by calculating the asymptotic behavior of the number of realizable permutation sequences, when those are restricted to belong to a single Kendall-tau error-detecting code. Possible extensions will be general error-correction capabilities, as well as general window-lengths.

[1] A. Jiang, R. Mateescu, M. Schwartz and J. Bruck, "Rank Modulation for Flash Memories," in IEEE Transactions on Information Theory, vol. 55, no. 6, pp. 2659-2673, June 2009, doi: 10.1109/TIT.2009.2018336.

[2] M. Horovitz and T. Etzion, "Local Rank Modulation for Flash Memories," in IEEE Transactions on Information Theory, vol. 65, no. 3, pp. 1705-1713, March 2019, doi: 10.1109/TIT.2018.2859403.

Design of the hardware-software interface of a real-time intradyne FSOC transceiver to evaluate different signal processing architectures to enable logging and visualization of high-speed data streams from the FPGA as well as configuration and calibration of the signal processing algorithms.

Federated learning is a machine learning paradigm that aims to learn collaboratively from decentralized private data owned by entities referred to as clients. However, due to its decentralized nature, federated learning is susceptible to model poisoning attacks, where malicious clients try to corrupt the learning process by modifying local model updates. Moreover, the updates sent by the clients might leak information about the private data involved in the learning. The goal of this work is to investigate and combine existing robust aggregation techniques in FL with differential privacy techniques.

References:

[1] - https://arxiv.org/pdf/2304.09762.pdf

[2] - https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=9757841

[3] - https://dl.acm.org/doi/abs/10.1145/3465084.3467919

- Basic knowledge about machine learning and gradient descent optimization

- First experience with machine learning in python

- Undergraduate statistics courses

- Prior knowledge about differential privacy is a plus

Modern vehicles boast advanced infotainment systems, delivering a range of services encompassing entertainment, navigation, comfort, and connectivity. These applications require extensive real-time data, necessitating efficient and reliable data processing from vehicle sensors and control units. With every vehicle launch and OS release, new sensors and APIs are introduced, which serve as important data points for high-level analytics and applications. Currently, these changes must be monitored and integrated manually, which is a demanding task prone to errors. The objective of this internship is to develop tools that automatically generate a data collector access layer for the platform APIs in every major OS release. This is done by analyzing the available APIs, followed by the creation of a data dependency tree based on the JSON format. An appropriate graph reduction algorithm is then applied to generate a data structure with minimum edge dependencies while preserving analyzability of the data points. An appropriate lookup schema is then created based on the reduced data points, which calls the requested API based on the supplied configuration. Python and C++ are the languages used within the scope of this project.

Whenever an automobile manufacturer develops a new car model, the aim is to surpass its predecessors in terms of performance, efficiency, safety, and overall user experience. In order to do so, logged data has to be analyzed to find out where improvements can be made. Whenever a vehicle returns from a test drive, the collected data will be stored as an MDF (Measurement Data Format) file. These files are then read and analyzed with a software tool, but as of now, only a few files can be processed simultaneously due to hardware limitations. This research internship explores how data can be efficiently organized to maximize analyzability. Eventually, a program should be created which allows users to analyze all MDF files a vehicle has produced in parallel. 

During the internship, the student will be researching the application of Neural Network-based signal predistortion to mitigate the effects of fiber chromatic dispersion in direct detection systems.

• basic Python skills beneficial

Forschungspraxis

.

.

.

.

the students should translate functions from Matlab and C to julia. the functions involve calculation of infomation theoretic quantities to basic function of a discrete time transmission chain.

The students task is to learn julia and matlab to a extend that the translation from one to the other language is possible. We furthermore expect the student to learn how to use git and gitlab for managing larger projects.

basic programming knowledge

.