Advanced Seminar Embedded Systems and Internet of Things

We will announce the available topics and the application process on the 06th of March 2026. You will then be able to apply for a seminar topic between the 09th of March until the 06th of April (23:59 pm).
It is mandatory to attend all the lectures of our Advanced Seminar in presence to complete the course successfully. Virtual Attendance is not possible.

Application Process

Due to the high interest in our seminar topics we use an application process to assign the topics.

If you are interested in one of the topics below, please send your application together with your CV and your transcript of records to seminar.esi@xcit.tum.de. Express your interest and explain why you want to have that specific topic and why you think that you are most suitable for the topic. This allows us to choose the most suitable candidate for the desired topic to maximize the seminar's learning outcome and to avoid dropouts.

Additionally, you can indicate a second topic that you would like to take, such that we can still find a topic for you if your primary choice is not available.

Deadline: We encourage you to apply until the 06.04.2026. Afterwards we will assign the topics and notify all applicants. Once you are given the topic, we will ask for your confirmation. You must confirm your participation until the 10th of April.

Note: We do not assign topics on a first-come-first-served basis. Even though we appreciate your interest if you have asked or applied early for a topic we can not guarantee that you get a seat. Generally we have 3-4 applicants per topic. Please think carefully if you are able to do the work required as we have to reject other students. Generally, email clients remember the people you have communicated with. You will be registered to the seminar course by the Advanced Seminar Manager after the Kick-Off meeting on 15th of April.

Kick-off meeting

This semester the seminar will be conducted in physical mode. This means that you must join the physical classes and presentation which you will find on the Moodle page. Additionally, you can schedule weekly meetings with your supervisor via Zoom or on campus. Lecture materials and videos will be available on Moodle.

The kick-off meeting will be on the 15th of April at 9:45 on Campus. We ask all successfully selected participants to be present in the kick-off meeting. Please notify us in case you can not make it to the meeting, otherwise we will assume that you are no longer interested and give your place to another applicant.

Topics

Exploring AI-Agent Driven ZK-Circuit Design
Zero-Knowledge (ZK) proofs are a powerful cryptographic tool, but transitioning them from theory to practical applications begins with a challenging step: arithmetization. This process transforms a program into an Arithmetic Circuit (also known as a ZK-Circuit), which can then be verified using ZK proofs.

However, designing these circuits is often complex, error-prone, and requires low-level operations, making it a significant barrier for developers. With the emergence of Large Language Models (LLMs) and General Purpose AI Agents, there is growing potential to automate and simplify this process. In this seminar project, the student will: Investigate how AI agents can assist in the design of ZK-Circuits. Explore existing tools and frameworks that integrate AI with ZK proof systems. Benchmark AI-assisted circuit generation against traditional methods and models. Evaluate the accuracy, efficiency, and usability of AI-driven approaches.

Literature:
[1] https://hackmd.io/@jake/plonk-arithmetization
[2] https://learn.0xparc.org/materials/circom/additional-learning-resources/r1cs%20explainer/
[3] https://tlu.tarilabs.com/cryptography/rank-1
[4] https://zcash.github.io/halo2/
[5] https://arxiv.org/abs/2407.01502
[6] https://dl.acm.org/doi/full/10.1145/3655615

Supervisor: Marco Calipari
Status: Free
Secure Multiparty Computation for Autonomous Vehicles Collaborative Perception
Collaborative perception is a cutting-edge paradigm in autonomous systems designed to enhance the perception capabilities of individual vehicles through the exchange of perception data with other vehicles. However, this data sharing poses significant privacy risks for the vehicles involved. Secure Multiparty Computation (SMC) is a privacy-preserving technology that enables parties to compute functions jointly while keeping their individual inputs private and ensuring fairness. This seminar aims to explore the potential integration of SMC within the framework of collaborative perception. We will focus on the necessary requirements for successful implementation and conduct a feasibility analysis of achieving collaborative perception in a privacy-respecting manner.

Literature:
[1] Xiong, Jinbo, et al. "Toward lightweight, privacy-preserving cooperative object classification for connected autonomous vehicles." IEEE Internet of Things Journal 9.4 (2021): 2787-2801.
[2] T. Li, L. Lin, and S. Gong, AutoMPC: Efficient Multi-Party Computation for Secure and Privacy-Preserving Cooperative Control of Connected Autonomous Vehicles. 2019.

Supervisor: Marco Calipari
Status: Reserved
Image Processing Techniques for Early Camera-Based Collaborative Perception
This seminar explores image processing techniques to enable early collaborative perception in camera-based systems, particularly within autonomous vehicles. As vehicles and devices increasingly rely on shared visual data to enhance situational awareness, the ability to process and interpret images efficiently and accurately becomes critical. Fundamentals of image preprocessing for collaborative perception, including denoising, contrast enhancement, and geometric corrections. Feature extraction and matching across multiple camera feeds to enable spatial and temporal alignment. Multi-view fusion strategies, such as stitching and depth estimation, to create a unified perception model. Compression and transmission techniques optimized for low-latency sharing of visual data between agents. Real-time object detection and tracking in distributed camera networks. Challenges and solutions in synchronizing heterogeneous camera systems under varying environmental conditions.

Literature:
[1]: Hussain, Manzoor, Nazakat Ali, and Jang-Eui Hong. "Vision beyond the field-of-view: A collaborative perception system to improve safety of intelligent cyber-physical systems." Sensors 22.17 (2022): 6610.
[2]: Yushan Han, Hui Zhang, Huifang Li, Yi Jin, Congyan Lang, and Yidong Li. 2023. Collaborative perception in autonomous driving: methods, datasets, and
challenges. IEEE Intelligent Transportation Systems Magazine, 15, 6, 131–151.

Supervisior: Marco Calipari, Michael Kühr
Status: Free
Deterministic 6G Communication
This seminar topic focuses on the design and analysis of deterministic 6G communication systems operating under adversarial network conditions such as jamming, interference, and mobility-induced instability. The emphasis of this topic is on strict performance guarantees, including bounded end-to-end latency, jitter control, and guaranteed packet delivery, even in hostile or highly dynamic environments. The topic explicitly excludes time synchronization aspects and instead concentrates on scheduling, resource allocation, routing, and resilience mechanisms that preserve deterministic performance. The student will begin with the foundational concepts of 6G by reading white papers and related publications, followed by wireless delay modeling, adversarial communication scenarios, performance analysis, and deterministic scheduling under adversarial load.

Literature:
https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=10254227
https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=11103479
https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=10945893
https://arxiv.org/pdf/2602.21444

Supervisor: Rubi Debnath
Status: Free
Agentic AI for computer networks
Agentic AI is envisioned to enable self-optimizing, self-healing, and self-evolving communication systems that dynamically adapt and operate autonomously. In this seminar, the student will investigate the role of Agentic AI in modern and future communication systems, particularly in 5G and emerging 6G networks. The student should review related documents, AI-native network architectures, as well as multi-agent and multi-LLM systems in computer networks. The student will further compare centralized AI with distributed agent-based systems and evaluate agent-based 5G and 6G architectures. The seminar will begin with an examination of the requirements and architectural principles of Agentic AI, including relevant mathematical formulations, followed by an analysis of its integration into 5G, 6G, and broader computer network environments.

Literature:
https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=11339915
https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=11152698
https://arxiv.org/pdf/2502.16866
https://arxiv.org/pdf/2510.19973
https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=11370176
https://arxiv.org/pdf/2601.06640

Supervisor: Rubi Debnath
Status: Free
Deterministic Converged Satellite Communication
Deterministic Converged Satellite Communication refers to the integration of satellite networks (such as LEO constellations) with terrestrial and aerial communication systems to provide predictable, bounded-latency, and ultra-reliable communication services. Deterministic satellite communication aims to offer strict delay, jitter, and reliability guarantees suitable for mission-critical applications such as autonomous systems, military coordination, industrial control, and remote robotic operations. In this seminar topic, the student will understand the end-to-end deterministic communication requirements in a converged satellite communication system. The focus will be primarily on deterministic routing and scheduling in large scale satellite networks.

Literature:
https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=10843333
https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=11161661
https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=11017403
https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=11223261

Supervisor: Rubi Debnath
Status: Free
Authentication of sensor data
Authenticating sensor sensor data in a way that remains trustworthy even when the sensor device itself is untrusted is a niche but highly important research question. Most sensing data authentication techniques in the literature focus on forensic analysis after capture—for example, identifying the source sensor or detecting edited content based on sensor noise fingerprints such as photo-response non-uniformity (PRNU) patterns—but these do not prevent an adversary who controls the camera from spoofing or replaying frames at capture time. Another direction considers physical challenge–response via signals in the scene itself, where a trusted external challenge (e.g., dynamically changing visual patterns) is placed in view of the sensor. Capturing these physical markers can make it harder for an untrusted sensor to convincingly forge fresh footage, because the challenge sequence must physically interact with the scene light field rather than being digitally inserted later. For example, periodic visual challenges such as QR codes or text placed in the environment have been proposed to help verify that footage is fresh and genuine rather than a replay of old video. In this seminar, you will develop a literature overview of all methodologies intended to authenticate any untrusted sensor data. The key conceptual challenge for the seminar is to weigh what’s physically enforceable (scene challenges, structured light) versus what’s post-capture forensic (sensor fingerprints, PRNU), and to assess how far existing methods go toward authenticating raw captures when the internal sensor pipeline is untrusted.

Literature:
Junia Valente & Alvaro Cárdenas (2015): “Using Visual Challenges to Verify the Integrity of Security Cameras”
Valente (2017): “Remote Proofs of Video Freshness for Public Spaces”
Maier et al. (2020): "Camera Fingerprinting Authentication Revisited"

Supervisor: Maximilian Lüdecke
Status: Free
Key Management in International Rail Traffic Management Systems
Key management in international rail traffic management systems represents one of the most critical cybersecurity challenges facing modern railway infrastructure. This research area focuses on the secure generation, distribution, storage, and lifecycle management of cryptographic keys that protect communications between train control systems operating across national borders.
The European Rail Traffic Management System (ERTMS) serves as the primary example of international rail interoperability, where the European Train Control System (ETCS) enables trains to operate seamlessly across different countries using unified signalling and communication protocols. Central to ERTMS security is the Key Management Centre (KMC), which generates and distributes 3DES keys to On-Board Units (OBUs) in trains and Radio Block Centres (RBCs) along railway lines. These cryptographic keys enable message authentication through the EuroRadio protocol, ensuring that only authorized entities can communicate within the rail network.
The current key management approach faces significant operational challenges. Tens of thousands of keys must be physically distributed using portable media devices such as USB drives and CDs, creating a prohibitively high management burden and introducing security vulnerabilities. This manual distribution process becomes particularly complex for international operations, where trains crossing borders require access to keys from multiple national KMCs. The existing system requires extensive pre-planning and coordination between infrastructure managers from different countries, often leading to delays and increased operational costs.
As part of this assignment, you will deep-dive the different key generation/negotiation/transportation/storage schemes used by different national railway (infrastructure) operators, concluding your research with a comparison of the different schemes and standardization proposals and highlighting shared strength/weaknesses.

Literature:
Thomas, R. J., Ordean, M., Chothia, T., & de Ruiter, J. (2017). “TRAKS: A Universal Key Management Scheme for ERTMS.”
European Union Agency for Railways. (2020). “Railway Cybersecurity Report.”
Prof. Dr. Katzenbeisser. "Why railway is safe but not secure (Talk)" - 37c3 Chaos Communication Congress
Franeková et al.. "Approaches to a solution of key management system for cryptography communications within railway applications"

Supervisor: Maximilian Lüdecke
Status: Free
Creating a Camera System Model for Physical Adversarial Attacks
For the development of automated driving algorithms, cameras are, among Radar and LiDAR, the most important sensors to perceive the environment. While these sensors are used for safety-critical applications such as object detection or lane keeping, multiple physical adversarial attacks are known. A wide majority of these physical attacks already compensate for real-world effects, such as differences in brightness across viewing angles. Yet, the camera itself is often overlooked when creating physical adversarial attacks, although different cameras perceive their environment significantly differently. In this topic, the student shall investigate the image processing pipeline to derive improvement options for physical adversarial examples to compensate for camera effects.

Literature:
N. Akhtar, A. Mian, N. Kardan, and M. Shah, “Advances in Adversarial Attacks and Defenses in Computer Vision: A Survey,” IEEE Access, vol. 9, 2021.
A. Guesmi, M. A. Hanif, B. Ouni, and M. Shafique, “Physical Adversarial Attacks for Camera-Based Smart Systems: Current Trends, Categorization, Applications, Research Challenges, and Future Outlook,” IEEE Access, vol. 11, 2023.
M. Kühr, M. Hamad, P. MohajerAnsari, M. D. Pesé, and S. Steinhorst, “SoK: Security of the image processing pipeline for camera-based sensing in autonomous vehicles,” arXiv [cs.CR], 2026.

Supervisor: Michael Kühr
Status: Reserved
Semantic Description of object manipulation with robot arms
With Physical AI, robots promise to manipulate many different objects in open, dynamic environments. However, practical deployment is still constrained by factors such as a robot’s reach, available force, and softer constraints like energy consumption and time required for a manipulation, all of which are crucial for planning. At the same time, users or higher-level agents must be able to specify what object should be manipulated and how this manipulation should be carried out, in a way that is both machine-interpretable and interoperable across platforms.
In this seminar, the student will systematize existing approaches to the semantic description of robot manipulation tasks, object affordances, and robot capabilities, and investigate to what extent the Web of Things (in particular WoT Thing Description and semantic extensions) can serve as a unifying, web-native interface for describing what a robot can do with which object, under which constraints, and how a user/agent can ask for it.

Literature:
1: Y. Liu et al., "Aligning Cyber Space With Physical World: A Comprehensive Survey on Embodied AI," in IEEE/ASME Transactions on Mechatronics, Dec. 2025, doi: 10.1109/TMECH.2025.3574943, https://ieeexplore.ieee.org/abstract/document/11098567
2: S. Kaebisch et al, "Web of Things (WoT) Thing Description 1.1", 2023, https://www.w3.org/TR/wot-thing-description11/
3: M. Freund et al., “The SPA Ontology: Towards a Web of Things Ready for Robotic Planning.”, in ESWC Workshops, 2024, https://ceur-ws.org/Vol-3749/akr3-02.pdf

Supervisor: Roman Binkert
Status: Free
Interfaces for LLM‑Based Agents: Natural Language, APIs, and Semantic Descriptions in Cyber‑Physical Systems
As LLM-based agents become capable of calling tools, composing APIs, and coordinating with each other, the design of their interfaces—especially to the physical world—becomes a central question, particularly given the constraints of cyber-physical systems (CPS) such as low-power devices with limited computational resources and unreliable connectivity. In this seminar, the student will systematically analyze different interface paradigms for agent–device and agent–agent interaction in CPS, including but not limited to natural‑language prompts, traditional APIs, and machine‑readable semantic descriptions such as W3C Web of Things Thing Descriptions, and shall compare them regarding their strengths, limitations, reliability, and interoperability in practical CPS scenarios.

Literature:
1: X. Li et al. A survey on LLM-based multi-agent systems: workflow, infrastructure, and challenges. Vicinagearth_ 1, 9 (2024). https://doi.org/10.1007/s44336-024-00009-2
2: S. Kaebisch et al, "Web of Things (WoT) Thing Description 1.1", 2023, https://www.w3.org/TR/wot-thing-description11/
3: R. Binkert et al. 2025. Interoperable Cyber-Physical Multi-Agent Systems Through Web of Things. In Web Engineering: 25th International Conference, ICWE 2025, https://doi.org/10.1007/978-3-031-97207-2_24

Supervisor: Roman Binkert
Status: Free
Zero-Knowledge Proofs for Privacy-Preserving Healthcare Applications
Modern healthcare systems face a fundamental tension: sensitive patient data must remain private, yet increasingly needs to be shared — for insurance verification, cross-institutional care, clinical trial eligibility, and AI-assisted diagnostics. Existing approaches such as pseudonymization or access control are often insufficient, as they require exposing raw data to a verifying party. Zero-knowledge proofs (ZKPs) offer a cryptographic solution: a prover can convince a verifier that a statement is true without revealing any underlying data. ZKPs have seen recent advances with schemes such as zk-SNARKs (Groth16, PLONK) and zk-STARKs moving from theoretical constructs to deployable systems. However, their application to healthcare remains an underexplored research area. This project surveys and critically analyzes the landscape of ZKP applications in healthcare across two dimensions:
1) Use case analysis: A central goal is to identify and categorize concrete medical scenarios where ZKPs provide meaningful privacy guarantees — such as anonymous credential systems for vaccination or prescription records, privacy-preserving genomic queries, diagnostic inference without data disclosure, and patient-controlled access delegation to EHR data.
2) System and feasibility evaluation: Relevant ZKP schemes (zk-SNARKs, zk-STARKs, Bulletproofs) are examined against the practical constraints of healthcare deployments — proof generation time, verifier overhead, circuit complexity for medical predicates — and how they compose with complementary technologies such as homomorphic encryption and federated learning.

Literature:
Goldwasser, S., Micali, S., & Rackoff, C. (1989). The Knowledge Complexity of Interactive Proof Systems. SIAM Journal on Computing, 18(1), 186–208.
Ullah, I., et al. (2022). A Review of zk-SNARKs. arXiv:2202.06877.
Sallal, M., et al. (2024). Evaluating the Efficiency of zk-SNARK, zk-STARK, and Bulletproof in Real-World Scenarios: A Benchmark Study. Information, 15(8), 463. https://doi.org/10.3390/info15080463
Luong, D. A., & Park, J. H. (2022). Privacy-Preserving Blockchain-Based Healthcare System for IoT Devices Using zk-SNARK. IEEE Access, 10.
Bharath Babu, S., & Jothi, K. R. (2024). A Secure Framework for Privacy-Preserving Analytics in Healthcare Records Using Zero-Knowledge Proofs and Blockchain in Multi-Tenant Cloud Environments. IEEE Access. https://doi.org/10.1109/ACCESS.2024.3509457

Supervisor: Marco Calipari, Miro Moffet (TUM Create Singapore)
Status: Free