Student Work

Age of Information is a metric measuring information freshness at the receiver in a remote monitoring scenario. (For details, please see: https://core.ac.uk/download/pdf/189662885.pdf) It has been used as a performance metric in communications to evaluate various medium access schemes, such as Carrier Sense Multiple Access (CSMA).

This project requires a practical implementation of CSMA and similar conventional contention-based MAC algorithms on Software Defined Radios (SDRs). The student is asked to implement CSMA algorithm with GNUradio and test on a practical setup consisting of multiple SDRs. The implementation can be done in Python or C++ based on the preference of the student.

Note: The content will be extended further in case of a Master's thesis.

The student is asked to have:

• Solid knowledge in wireless communications
• Advanced programming skills in Python or C++ programming languages
• Some experience with practical hardware in the past, e.g. wireless sensor motes, SDRs.
• Good communication & writing skills in English
• Motivation and self-organization throughout the thesis

It is beneficial to have GNUradio experience but not mandatory.

The Wireless Sensor Networks lab offers the opportunity to develop software solutions for the wireless sensor networking system, targeting innovative applications. For the next semester, a position is available to assist the participants in learning the programming environment and during the project development phase. The lab is planned to be held every Tuesday from 15:00 to 18:00.

Solid knowledge in Wireless Communication: PHY, MAC, and network layers. Solid programming skills: C/C++. Linux knowledge. Experience with embedded systems and microcontroller programming knowledge is preferable.

IEEE 802.1 TSN standards aim to provide for mechanisms and methods for deterministic delay and jitter in industrial networks. Sources of delay and jitter in a TSN are numerous, i.e., misconfigured or incorrectly dimensioned TAS schedules, congested egress ports, large clock jitter in distributed time synchronization, congestion at egress / ingress ports of the end-stations are all possible causes of poor end-to-end performance.

In this work, an existing approach to end-to-end monitoring of transmission delay and jitter will be extended to support for quick identification of source of the anomaly in a TSN scenario (i.e., the root cause). Two or more approaches will be proposed to cater for end-to-end monitoring of TSN streams, relying on:

1) intermediate, pre-placed network TAPs;

2) hardware / software data plane agents capable of telemetry and dynamic re-programming to support the quick identification of anomaly sources for individual streams.

Any resulting overhead of the intermediary monitoring solution will be quantified for consideration in the TSN stream planning. The work will hence implement and validate the monitoring mechanisms in an existing TSN testbed that relies on a Linux-based control plane, and will optionally involve P4-based middlebox devices to serve as intermediate monitoring entities. 

[1] 802.1Qbv - Enhancements for Scheduled Traffic - http://www.ieee802.org/1/pages/802.1bv.html

[2] Improving Network Monitoring and Management with Programmable Data Planes - https://p4.org/p4/inband-network-telemetry/

- Very good knowledge of networking concepts and architectures

- Very good knowledge and practical experiences with Linux networking

- Good knowledge of Python, C, C++ or Rust

- Any prior experience with TSN is a plus

- Any prior experience with P4/SDN is a plus

- Proficient English and/or German communication skills

- Ability to work and pursue solutions under limited supervision

ermin.sakic@siemens.com

blenk@tum.de

andreas.zirkler@siemens.com

Industry 4.0 scenarios have introduced the novel requirements for dynamic device-to-device interaction without strict plan-ahead interactions (typical of Industrial Ethernet of the last two decades). For example, the use cases of work piece de-attachment and interacting Automated Guided Vehicles (AGVs) in factory automation impose the need for dynamic inter-connection of distributed industrial applications at runtime [1] without a plan-ahead schedule.

Dynamic TSN stream establishment has been proposed by IEEE802.1 to address scenarios of applications requesting real-time communication services at runtime [2, 3]. TSN schedulers capable of computing the transmission schedules with guaranteed real-time properties (upper-bound constraints on latency, jitter, and guaranteed bandwidth) have accordingly been proposed in the past, however, they have traditionally focused on one-time / offline execution [4].

Combined with the requirement for dynamic and efficient resource (de-)allocation, transmission schedules must generally be adaptable at runtime in order to provide for maximum efficiency in network resource utilization. This work will investigate the runtime adaptation of TSN control and data plane to support for the dynamic stream allocation. Algorithms for adaptation of existing TAS-based schedules with minimal impact on availability (packet loss) and QoS guarantees for running TSN streams will be proposed, implemented, and validated. Novel metrics for stream consistency conservation may need to be proposed by the approach. The resulting approach will eventually be evaluated in NeSTiNg (OMNeT++/INET) simulator and/or a small-scale physical testbed comprising multiple TSN switches and end-stations.

[1] IEC/IEEE 60802 TSN Profile for Industrial Automation - https://1.ieee802.org/tsn/iec-ieee-60802/

[2] P802.1Qdd – Resource Allocation Protocol - https://1.ieee802.org/tsn/802-1qdd/

[3] 802.1Qat - Stream Reservation Protocol -/ http://www.ieee802.org/1/pages/802.1at.html

[4] Dürr et al., No-wait Packet Scheduling for IEEE Time-sensitive Networks (TSN) - https://dl.acm.org/doi/abs/10.1145/2997465.2997494

- Very good knowledge of networking concepts and architectures

- Very good knowledge and practical experiences with Linux networking

- Good knowledge of Python, C, C++ or Rust

- Any prior experience with optimization methods and and tools and OMNeT++  is a plus

- Any prior experience with TSN is a plus

- Any prior experience with SDN is a plus

- Proficient English and/or German communication skills

- Ability to work and pursue solutions under limited supervision

ermin.sakic@siemens.com

blenk@tum.de

andreas.zirkler@siemens.com

IEEE 802.1Qbv Time-Aware Shaper (TAS) standardizes the mechanisms for TDMA-based packet transmission in future Ethernet deployments, as well as the interfaces for definition and configuration of associated transmission schedules [1].

In this work, a state-of-the-art end-to-end TAS-based scheduler will be extended with schemes enabling for data plane redundancy under consideration of worst-case delay and jitter constraints. Support for TSN streams with 1:1 replication for proactive redundancy, as well as an approach to efficient schedule reservation with reactive 'fast-failover' redundancy will be investigated, implemented, and subsequently evaluated in terms of adherence to jitter, delay, and packet loss constraints, as well as the computation complexity and acceptance rates of the approach.

The joint routing/scheduling task will first be formulated as an optimization problem and implemented using an optimization solver. The student will then extend the existing TAS scheduler relying on metaheuristics for faster execution compared to using a solver.

[1] IEEE 802.1Qbv - Enhancements for Scheduled Traffic - http://www.ieee802.org/1/pages/802.1bv.html

[2] IEEE P802.1CB – Frame Replication and Elimination for Reliability - https://1.ieee802.org/tsn/802-1cb/ 

- Very good knowledge of networking concepts and architectures

- Very good knowledge and practical experiences with Linux networking

- Good knowledge of Python, C, C++ or Rust

- Any prior experience with optimization methods and tools is a plus

- Any prior experience with TSN is a plus

- Any prior experience with SDN is a plus

- Proficient English and/or German communication skills

- Ability to work and pursue solutions under limited supervision

- Motivation to work with real hardware deployments

ermin.sakic@siemens.com

blenk@tum.de

andreas.zirkler@siemens.com

In future, more data will be transmitted wirelessly, especially in case of aircraft cabins. However, cabin channel models are scarce and outdated for radio frequencies. Therefore, we developed a 3D structure of a single aisle aircraft in blender. The next step is to develop a statistical description of channel model with the help of a raytracing software like Wireless InSite or adapting an existing raytracing engine as LuxCore.

The goal of this thesis is to extend the single aisle aircraft to a two aisle in blender and to derive a small scale (multipath) and large scale (pathloss) model. The overall goal is to compare the channels with exisiting channel models and for different positions of accesspoint.

Tasks:

•     Review of related literature
•     Extend the model
•     Compare the new channel effects
•     Evaluate the simulation results

Related Work:

Blender

https://www.blender.org/

Raytracing

Vitucci, E. M., et al. "Ray tracing RF field prediction: An unforgiving validation." International Journal of Antennas and Propagation 2015 (2015).
A. Navarro, D. Guevara, N. Cardon and J. Gimenez, "Using 3D game engines and GPU for ray launching based channel modeling in indoor," 2014 XXXIth URSI General Assembly and Scientific Symposium (URSI GASS), Beijing, 2014, pp. 1-4, doi: 10.1109/URSIGASS.2014.6929301.

• Some experience in python
• Interest in 3D modeling
• Interest in channel models

<p>The course "Data Networking" (replaces Broadband Communication Networks) of Prof. Wolfgang Kellerer teaches network-related methods of mobile communication: WIFI, 2G to 5G cellular networks, etc. In order to bridge the gap between the methods and real life, innovative teaching concepts will be applied and developed to present Data Networking in a blended learning format based on Podcasts (the students learn and experiment in short episodes about wireless communication and networking in day to day scenarios), online quizzes, feedback channels, recording of lecture, tutorial and discussions.</p>

<p>In order to setup and support the course Data Networking in WS 20/21, Prof. Kellerer is looking for students experienced in programming and in video tools for recording and other online tools (incl. moodle) to help him. Work can start in Sept./Oct 2020 already for preparations.</p>

Note: This work can also be be done in parallel to taking the course Data Networking in WS 20/21

<p>Programming skills; experience with online teaching tools, video tools and recording; knowledge of communication networks </p>

Arled Papa (arled.papa@tum.de)

Anna Prado (anna.prado@tum.de)

Software-Defined Radio Access Network (SD-RAN) is receiving a lot of attention in 5G networks, since it offers means for a more flexible and programmable mobile network architecture.

The heart of the SD-RAN architecture are the so called SD-RAN controllers. Currently, initial prototypes have been developed and used in commercial and academic testbeds. However, most of the solutions only contain a single SD-RAN controller. Nonetheless, a single controller becomes also a single point of failure for a system, not only due to potential controller failures but also due to a high load induced from the devices in the data plane.

To this end a multi-controller control plane often becomes a reasonable choice. However, a multi-controller control plane renders the communication among the controllers more challenging, since they need to often exchange control information with each other to keep an up to date network state. Unfortunately, currently there is no protocol available for such a communication.

The aim of this work is the development and implementation of an East/West API for SD-RAN controller communication according to 5G stardardization. The protocol should enable the exchange of infromation among the SD-RAN controllers regarding UEs, BSs, wireless channel state and allow for control plane migration among controllers.

• Experience with programming languages Python/C++.
• Experience with socket programming.
• Knowledge about SDN is a must.
• Knowledge about 4G/5G networks is a plus.

Current SD-RAN platforms are only evaluated for small network dimensions due to expensive wireless hardware. However, in order to test the real scalability of the existing solutions a large number of devices needs to be deployed. To this end, an emulation mode for the control and data plane is seen as a potential solution. 

In this work, the student shall deploy a data plane user based simulator that is able to connect to an SD-RAN controller in real time through a control plane channel and report user specific statistics. The controller should be able to send singaling information back to the simulator and the effect on the user perfromance should be measured. 

The main goal of this work is to investigate the real time deployment of an SD-RAN architecture and identify potential bottlenecks regarding the control plane signaling and its effects on the users in the data plane.

• Mandatory programming skills in Python, C++ is a plus.
• Mandatory knowledge about socket programming.
• Knowledge about mobile networks 4G/5G.
• Knowlegde about event based simulators is a plus.
• Experience with google protobuffers is a plus.

In future, data will be transmitted over light, especially in case of aircraft cabins. However, cabin channel models are scarce and outdated for radio frequencies and non existent for communication based on light. Therefore, we developed a 3D structure in blender to simulate light propagation in aircrafts. However, the model is lacking passengers.

The goal of this thesis is to extend the model by adding passengers, with the help of pre-existing models and automated by a script. After extending the simulator, different scenarios should be simulated and analysed. The overall goal is to show the influence of adding passengers and that a lightbased communication is a reliable approach to connect multiple devices in the cabin.

Tasks:

• Review of related literature
• Extend the model with people
• Compare the new channel effect
• Evaluate of the simulation results

Related Work:

Blender

https://www.blender.org/

LiFi

H. Haas, L. Yin, Y. Wang and C. Chen, "What is LiFi?," in Journal of Lightwave Technology, vol. 34, no. 6, pp. 1533-1544, 15 March15, 2016, doi: 10.1109/JLT.2015.2510021.

• Some experience in python
• Interest in 3D modeling
• Interest in new communication technologies

Problem Statement

The human hand is one of the most versatile instruments to interact with the environment. Consequently, hand interactions play a key role for realistic interactivity in virtual environments. Most existing hand tracking systems today are vision based which require a complex instrumentalization of the environment. This constrains portability and ad-hoc usage, limiting application use cases for industry and consumers.

The goal of this thesis is to implement a glove to track a hand position in space and visualize it in real time in VR to simulate the trajectory of an object released from the hand. The overall goal of this research project is to implement a physically realistic VR simulation of the popular game beer pong to investigate whether practice in VR environments results in training effects in real life.

Tasks

- Review of related literature

yle="margin-left: 18.0pt; mso-add-space: auto; text-indent: -18.0pt; mso-list: l0 level1 lfo1;">- Prototypical development of a sensory glove

yle="margin-left: 18.0pt; mso-add-space: auto; text-indent: -18.0pt; mso-list: l0 level1 lfo1;">- Implementation of real-time hand tracking and visualization in VR (Unity)

yle="margin-left: 18.0pt; mso-add-space: auto; text-indent: -18.0pt; mso-list: l0 level1 lfo1;">- Implementation of a real-time trajectory simulation

yle="margin-left: 18.0pt; mso-add-space: auto; text-indent: -18.0pt; mso-list: l0 level1 lfo1;"> - Evaluation of the implemented trajectory simulation in VR in a user study

Preferred Qualifications

- Hardware Prototyping (IMUs, Microcontrollers, e.g. nRF52 DK)
- Experience with signal processing and filtering (i.e. Kalman filtering)
- Interest in VR (Unity, C++ or C#)
- Independent thinking and creative problem solving
Related Work

Real Time VR

https://doi.org/10.1109/38.963459

Hand Tracking and Gloves

https://ieeexplore.ieee.org/document/6223225

https://doi.org/10.1109/TIM.2008.925725

https://doi.org/10.1145/1531326.1531369

Trajectory Prediction

https://www.3delement.com/?p=610

https://web.mit.edu/16.unified/www/FALL/systems/Lab_Notes/traj.pdf

Michael Fröhlich, CDTM (TUM & LMU)

http://www.cdtm.de

Email: froehlich@cdtm.de

Future IoT will need wireless links with high data rates, low latency and reliable connectivity despite the limited radio spectrum. Connected lighting is an interesting infrastructure for IoT services because it enables visible light communication (VLC), i.e. a wireless communication using unlicensed light spectrum. This work will aim at developing a tool to perform an economic evaluation of the proposed solution in the particular case of a smart office.

For that purpose, the following tasks will have to be performed:

• Definition of a high-level framework specifying the different modules that will be implemented as well as the required inputs and the expected outputs of the tool.
• Development of a cost evaluation Excel-VBA tool. This tool will allow to evaluate different variations of the selected case study and if possible, to compare different alternative models (e.g., dimensioning) or scenarios (e.g., building types).

- Excel and VBA

The advent of programmable switches and its high-level programming language such as P4 enabled many new innovative applications. Network telemetry is one of these promising applications.

In this work, the student will build a deep understanding of the state-of-the-art work related to network telemetry and data analytics using P4 programmable hardware. Mainly the focus will be on the "StarFlow" system in order to reproduce this work, analyze its strong features, and identify its limitations.

Here are more references related to StarFlow project:

1- Blog with project description: https://blog.apnic.net/2018/10/02/packet-level-network-telemetry-and-analytics-without-compromises/

2- Published Paper: https://www.usenix.org/system/files/atc18-sonchack.pdf

3-Git Repository: https://github.com/jsonch/StarFlow

-Background on P4

- C++ Programming

- Critical Thinking

P4 is a novel language designed for programming packet processors in a high-level language. Currently, it can be used to program different types of networking devices. https://p4.org/

Our goal in this thesis is to investigate the optimal placement of different network functions (or P4 programs) into different P4 devices based on the characteristics of the P4 programs and the P4 devices. 

In this work, the student will formulate the placement problem as an optimization problem and solve it for different objectives and under the relevant constraints.

- Good theoretical background on optimization.

- Background on P4 language is a plus.

The 5G cellular networks are the state-of-the-art cellular networks for the coming 10 years. On the other hand, P4 is a promising language for programming packet processors. It can be used to describe the packet processing pipelines of different devices (FPGAs/ASICs/...).

In this work, we will explore and analyze the potentials of offloading specific 5G Network Functions into P4 programmable devices. The performance gains of this approach will be evaluated.

- Good Critical Thinking

- Basic Programming Skills

- Knowledge about P4 language and 5G is a plus.

Networked Control Systems, i. e. the interconnection of a control system over a communication network, are a fundamental building block of future industrial automation systems. To study NCS, the NCSbench platform [1,2] was developed allowing performance measurements of a real NCS. The platform is built using Lego, is developed in python, communicates over standard IP network interfaces and is fully open-source.

In order to further develop the platform, a working position is available.

[1] https://github.com/tum-lkn/NCSbench/wiki

[2] S. Zoppi, O. Ayan, F. Molinari; Z. Music, S. Gallenmüller, G. Carle, W. Kellerer, NCSbench: Reproducible Benchmarking Platform for Networked Control Systems. IEEE Consumer Communications & Networking Conference, 2020

This work will consist of a significant amount of programming and also of processing the measurement data.

Requirements:

• Good knowledge of wireless systems and protocols (WLAN, WSN).
• Basic knowledge of LTI control systems.
• Strong C and Python programming skills.
• Experience with Linux systems is recommended.
• Experience with python data processing tools is beneficial.

yle="text-align: justify;">Flexible optical networks achieve a significant increase of network capacity by assigning as much spectrum to the demands as needed. With advances in transponders supporting software tunable channel configurations, network planners are able to select the best combination of data rate, modulation format and forward error correction (FEC) for each light-path. However, the impact of failures in these networks is higher due to the closeness between channels as well as the low OSNR margins.

yle="text-align: justify;">Hence, in this thesis we will look into the monitoring capabilities of flexible optical networks provided by existing components as well as extra monitoring equipment. All the gathered monitoring information should be processed by machine learning based classification tools such that fast and efficient failure diagnosis can be achieved.  

Programming skills (Python)

Network slicing is a new emerging technique introduced in 5G to boost network performance and accomodate the new heterogeneous requirements of upcoming applications. To this end this has attracted a lot of attention in the state-of-the-art where there is vast of ongoing research in resource allocation and network efficiency in the RAN domain.

The focus of this thesis will be on development and implementation of heuristic approaches for increasing efficiency in resource allocation and network performance. 

• Strong mathematical background.
• Solid programming skills (python, matlab, c++).
• Knowledge of optimization techniques is a plus.

In this work, we plan to implement several Virtual Network Functions (VNFs) for NAT, Firewall, IDS, etc. on Virtual Machines and Contrainers. We plan to characterize the VNFs in terms of resource usage. Thereafter, we study the reconfigurability and its overhead on the performance of the VNFs. We would like to measure the metrics such as throughput, latency, etc.

programming, linux

amir.varasteh@tum.de

The goal is to investigate the impact of topology, addressing scheme and communication pattern on the learnability of a routing protocol.

Wireless resource allocation for networked control systems. An extension of the SotA with Gilbert-Elliot Model is the aim of this work.

Communication networks form the backbone of our digital society and have become a critical infrastructure. However, the operation of networks today is still a manual and error-prone process, which is worrysome.

We have recently started exploring the vision of self-repairing networks and developed a first approach and software which allows for a fast what-if analysis of MPLS networks [1]: our approach relies on a classic result in formal methods and automatically tests whether network configurations are policy compliant (e.g., "can host A reach host B even under up two 2 link failures?").

While our first results are promising and the approach fast in theory, it is still relatively slow in practice. 

In this project, we would like to explore the factors that impact the time of our verification tool. For this, we conduct extensive time measurements of our tool and analyze the performance based on various data attributes. Moreover, we want to ask the question of how much is the performance of such tool predictable when using machine learning.

This is a joint work with University of Vienna, Faculty of Computer Science. A short research visit abroad in Austria might be included.

[1]  Jesper Stenbjerg Jensen, Troels Beck Krøgh, Jonas Sand Madsen, Stefan Schmid, Ji?í Srba, and Marc Tom Thorgersen. 2018. P-Rex: fast verification of MPLS networks with multiple link failures. In Proceedings of the 14th International Conference on emerging Networking EXperiments and Technologies (CoNEXT ’18). Association for Computing Machinery, New York, NY, USA, 217–227. DOI:https://doi.org/10.1145/3281411.3281432

The efficient embedding of virtual networks is an integral part of future communication infrastructures such as 5G/6G. In order to gain the highest revenue from their infrastructure, operators need to solve the so-called virtual network embedding problem. As this problem combines variants of well-known optimization problems, e.g., the bin-packing problem and the set covering problem, the development of fast and efficient algorithms seems to be an endless endeavor. While research has proposed a countless number of algorithms, less attention has been paid to simple problem aspects. For instance, although most work focuses on algorithms for the online embedding problem, less attention has been paid to answering questions such as how and why online and offline embedding problem solutions differ at all. Hence, this thesis targets at the question of whether time really matters. How different or how equal do existing algorithms behave when faced with online or offline variants. This study should try to reveal potential similarities or diversities between the online and offline virtual network embedding problem. To achieve this, a simple offline embedding simulator should be developed in C++ and compared to an existing online embedding simulator. The offline embedding simulator should implement at least two state of the art algorithms and compare their performance with respect to online and offline use cases.

Programming skills, interests in algorithms, strong motivation.

Goal is to implement a probabilistic graphical model to represent weighted graphs with overlapping communities.

...

Software-Defined Networking (SDN) is proposed as at technic to increase the network performance by decoupling the control from the data plane of commodity hardware. Recently, SDN has been proposed as a solution to further improve the Radio Access Network (RAN) side of the network. To this aim the notion of SD-RAN controllers has emerged and there is a vast of ongoing research on the performance, evaluation and use of such entities.

In this thesis, the student will have a hands-on experience with FlexRAN controller which is one of the main state-of-the-art SD-RAN controllers. The goal is to enable control of FlexRAN over multiple cells (i.e., eNB/gNB) and ensure a logical flow among the messages exchanged among them. The final evaluation will require a demonstration of the Proof-of-Concept and a set of measurements regarding the cell and SD-RAN controller performance.

• Programming skills in C/C++ is a must, Python knowledge is a plus.
• Knowlegde about SDN, RAN, network slicing is a plus.

...

Das Thema dieser Bachelorthesis ist die Entwicklung, Optimierung und Evaluation des proprietären Brandmeldebussystems LSNi1 der Firma Bosch Sicherheitssysteme GmbH in Grasbrunn.

Brandmeldeanlagen sind ein bestehender notwendiger Teil der Gebaudetechnik und sollen fur neue vernetzte IoT-Dienste verwendet werden. Aufgrund immer höheren Anforderungen dieser Dienste soll die Datenrate uber den Brandmeldebus gesteigert werden. Zur Zeit wird das Bussystem zur verhaltnismäßig datenarmen Alarmabfrage und Antwort der Netzelemente verwendet. Ein wichtiger Punkt ist dabei, die Systemcharakteristika des Brandmeldesystems nicht zu verschlechtern. Solche Netzwerke sind als Sicherheitssysteme deklariert und mussen daher äußerst zuverlässig in Sachen Ausfallsicherheit und Fehlererkennung sein.

Die Aufgabe besteht darin, eine neue Übertragungstechnik mit höherer Datenrate auf der physikalischen Schicht des Busses zu finden und diese mit Hilfe eines Prototypen umzusetzen. Die Optimierung des Prototypen soll durch Berechnung und Experimentieren mit Systemelementen ausgearbeitet werden. Daraufhin werden verschiedene Testaufbauten, im firmeneigenen Labor getestet. Die dabei aufgenommenen Daten werden über ein entwickeltes Matlab-Evaluierungsprogramm ausgewertet, um eindeutige Aussagen über die Funktionalitat des Systems zu geben. Eine besondere Herausforderung bei Systemen dieser Art sind die hohen Kabellängen und die große Anzahl an Systemelementen auf dem Bus. Mit den ausgewerteten Ergebnissen soll überpruft werden, in welchem Umfang die Datenrate auf dem Bus gesteigert werden kann.

In this work, the student will derive analytical equations for different performance metrics related to P4 devices. The modeling will be based on queuing theory and previously conducted measurements on P4 devices. Simulations will be used to verify the derived performance model. 

Knowledge about P4, Queuing Theory, and Matlab. 

Today's Data Center (DC) networks are facing increasing demands and a plethora of requirements. Factors for this are the rise of Cloud Computing, Virtualization and emerging high data rate applications such as distributed Machine Learning frameworks.
Recently several new architectures have been proposed that rely on reconfigurable optics to evolve the topology over time, such as Helios[3], Projector[5], c-Through[1] or RotorNet[6].
Some of these approaches learn attributes of the traffic distribution in the network to determine topology configurations, e.g., xWeaver[4] or DeepConf[2].
A different stream of research focuses on generating adversarial input to networks or networking algorithms [7,8] to identify weak spots or improve robustness and solution quality of data-driven approaches.
The goal of this thesis is to apply the latter one to reconfigurable topology designs with the main focus on the data-driven variants (xWeaver) to benchmark their performance and to improve their solution quality.
The thesis builds upon an existing flow-level simulator in Python and initial algorithms that generate adversarial inputs for networking problems.

[1] G. Wang et al., “c-Through: Part-time Optics in Data Centers,” presented at the ACM SIGCOMM, 2010, p. 12.
[2] S. Salman, C. Streiffer, H. Chen, T. Benson, and A. Kadav, “DeepConf: Automating Data Center Network Topologies Management with Machine Learning,” in Proceedings of the 2018 Workshop on Network Meets AI & ML, New York, NY, USA, 2018, pp. 8–14, doi: 10.1145/3229543.3229554.
[3] N. Farrington et al., “Helios: A Hybrid Electrical/Optical Switch Architecture for Modular Data Centers,” pp. 1–12, 2010.
[4] M. Wang et al., “Neural Network Meets DCN: Traffic-driven Topology Adaptation with Deep Learning,” Proceedings of the ACM on Measurement and Analysis of Computing Systems, vol. 2, no. 2, pp. 1–25, Jun. 2018, doi: 10.1145/3224421.
[5] M. Ghobadi et al., “ProjecToR: Agile Reconfigurable Data Center Interconnect,” 2016, pp. 216–229, doi: 10.1145/2934872.2934911.
[6] W. M. Mellette et al., “RotorNet: A Scalable, Low-complexity, Optical Datacenter Network,” 2017, pp. 267–280, doi: 10.1145/3098822.3098838.
[7] S. Lettner and A. Blenk, “Adversarial Network Algorithm Benchmarking,” in Proceedings of the 15th International Conference on emerging Networking EXperiments and Technologies, Orlando FL USA, Dec. 2019, pp. 31–33, doi: 10.1145/3360468.3366779.
[8] J. Zerwas et al., “NetBOA: Self-Driving Network Benchmarking,” in Proceedings of the 2019 Workshop on Network Meets AI & ML  - NetAI’19, Beijing, China, 2019, pp. 8–14, doi: 10.1145/3341216.3342207.

Decentralized channel access strategies with Reinforcement Learning when the agents are networked control systems. We assume slot based centralized scheduling is not feasible.

Sensors need to observe the channel outcome and also plant state to decide on action.

Python programming, Basics of Communication Networks, Reinforcement Learning Understanding

Current mobile phones are equipped with additional Artificial Intelligence (AI) chipsets for performing different tasks on mobile such as Augmented Reality (AR). Advanced tasks like object recognition can still be handled on the mobile, but seldom run in real-time. An alternative to the computation on the mobile would be to offload the AI part to a nearby edge cloud. This might, however, require a huge bandwidth and increases the latency.

In this thesis, the trade-off between offloading to the edge cloud and local AI computation on the mobile should be evaluated also considering a third option, a pre-computation on the mobile, and a detailed analysis on the edge cloud. An example use case for this thesis is a real-time detection and classification of construction sites while driving through a city. 

The Boarder Gateway Protocol (BGP) lies at the heart of the Internet. Internet routers use BGP to exchange routing information with neighbors. Based on their routing policy, current updates, and internal network routes, routers populate their local routing information base. The final paths are pushed to the forwarding information base (FIB). For fast matching, FIBs are implemented with expensive TCAM. In order to use TCAM space efficiently, aggregation algorithms compress the RIB before pushing the final paths to the FIB. However, each BGP update may demand a change of the FIB, leading to expensive compression and decompression tasks. Hence, this work focusses on the possibility of machine learning to enhance the FIB compression routine.

Recently, a scalable load balancing algorithm in the dataplane has been proposed that leverages P4 to estimate the utilization in the network and assign flows to the least utilized path. This approach can be interpreted as a form of Graham's List algorithm.

In this thesis, the student is tasked to investigate how a different online scheduling algorithm called IMBAL performs compared to HULA. A prototype of IMBAL should be implemented in NS3. The tasks of this thesis are:

1. Literature research and overview to online scheduling and traffic engineering in data center networks.
2. Design how IMBAL can be implemented in NS3.
3. Implementation of IMBAL in NS3.
4. Evaluation of the implementation in NS3 with production traffic traces and comparison to HULA (a HULA implementation is provided from the chair and its implementation not part of this thesis).

Modeling of NetFPGA Programmed with P4

Network slicing is a new emerging technique introduced in 5G to boost network performance and accomodate the new heterogeneous requirements of upcoming applications. To this end this has attracted a lot of attention in the state-of-the-art where there is vast of ongoing research in resource allocation and network efficiency in the RAN domain.

The focus of this thesis will be on implementation of an event based slicing simulator and built-up of a deep reinforcement learning framework that interacts with the schedulers of the event based simulator. The final goal is the ability to predict the scheduler type and adjust the scheduling decisions accordingly to boost the network performace.

Next generation mobile networks are envisioned to cope with heterogeneous applications with diverse requirements. To this end, 5G is paving the way towards more scalable and higher performing deployments. This leads to a revised architecture, where the majority of the functionalities are implemented as network functions, which could be scaled up/down depending on the application requirements. 

3GPP has already released the 5G architecture overview, however there exists no actual open source deployment of RAN functionalities. This will be crucial towards the evaluation of the Core network both in terms of scalability and performance. In this thesis, the student shall understand the 5G standardization, especially the control plane communication between the RAN and 5G Core. Further, an initial RAN function compatible with the 5G standards shall be implemented and evaluation of control plane performance will be carried out. 

• Strong knowledge on programming languages Python, C++ or Java.
• Knowledge about mobile networking is necessary.
• Knowlegde about 4G/5G architecture is a plus.

Network slicing is a novel approach used in 5G networks to enable the deployment and coexistence of heterogeneous network with various Quality of Service (QoS) requirements in the same physical infrastructure. In the Radio Access Network (RAN) the network slicing concept is complex is due to the stochastic nature of wireless channels and scarce resources. Vast of ongoing research deal with optimization approaches for RAN slicing, however most of the focus is on the downlink scenario. Nonethess, a joint optimization of downlink and uplink scenarios can render the problem more complex, but on the other hand can procude better results for the overal system behavior.

In this master's thesis the student will focus on a combined downlink/uplink scenario of a cellular system and provide a joint optimization algorithm to enhance the network performance. 

• Basic knowledge of cellular systems 4G/5G.
• Good knowledge of programming languages Python, Matlab.
• Good mathematical background, especially on optimization techniques.
• Knowlegde of tools such as Gurobi is a plus.

Introduction to Software Defined Networking (SDN) and Network Virtualization paradigms has brought new challenges to tackle. In this environment, network switches communicate to SDN controllers through a hypervisor. Therefore, considering a network graph, three questions may come up: 1) Where to deploy SDN controllers? 2) Where to deploy SDN hypervirors? 3) How to connect network switches to the respective SDN controllers to meet the traffic dynamicity and QoS requirements?

In this work, we plan to design and evaluate an optimization framework to optimally answer the above questions.

Mathematical Optimization, Algorithms, Java / Python

amir.varasteh@tum.de

Optical core networks deploy fixed grid in order to accommodate their demands. The trend is however, migrate towards flexgrid networks able to use as much spectrum as required by the demand allowing a better use of the spectrum. However, the Routing and Wavelength Assignment (RWA) paradigm becomes more complex as it has to find the suitable spectrum of the demand.

This thesis will consider different multiplexing alternatives: space division, band division multiplexing, etc.

The objective is to compare the performance, throughput, BoM, etc. of the different alternatives for different networks and demand evolution matrices.

Java programming skills

Basic knowledge of flexible optical networks

The goal of this thesis is the classification of packet-level traces using Markov- and Hidden Markov Model. The scenario is open-world: Traffic of specific web applications should be distinguished from all possible web-pages (background traffic). In addition, several pages should be differentiated. Examples include: Google Maps, Youtube, Google Search, Facebook, Google Drive, Instagram, Amazon Store, Amazon Prime Video, etc.

This research internship looks at Quanktum Key Distribution exchange in a wide area pan-European network. Different architectures, routing and spectrum allocation methods need to be considered, keeping in mind both technological and economic constraints of such a combined deployment.

This work is helpful for Optical Networking as well as satellite network providers in order to provide secure data transmissions over large region multi-national governmental organizations, banks and security agencies.

Optical Networking Devices can now be controlled using Software Defined Networking (SDN). With the advent of software configurable channel cards, network operators have the option to extract physical layer information and make short term changes to the network in order to achieve higher throughput or to save costs.

An offline network planning agent is used to calculate Routing, Wavelength and Spectrum Allocation of optical lightpaths between various demands. However, to integrate this to an SDN based orchestrator, it needs to be made compliant to OpenROADM.

In this internship/bachelor thesis, a student will first research about OpenROADM YANG models and identify the areas where it should be extended. Then they will create new north bound and south bound mediators and finally simulate a network planning scenario using NetConf and observe various elements in order to draw useful conclusions.

Must Have

• Basic knowledge of graph theory, communication networks and optical networks
• Independent worker
• Working knowledge of Python
• Willingness to learn

Good to Have

• Knowledge of Java
• Familiarity with NetConf/YANG

To apply, please send your CV and grade transcripts to kireet.patri@tum.de

...

Flexibility is an indispensable feature of future communication networks and technologies. Only flexible networks allow coping with the fast-changing demands of future network applications and use cases like 5G/6G. In order to quantify the flexibility of networks, we have recently presented a new flexibility measurement framework. As many problems in network communication are studied on graphs, this thesis should investigate the impact of topologies on network flexibility by applying the concepts proposed in our framework. In detail, different networking use cases (like network programmability, network failures, and network verification) should be analyzed for a variety of network topologies. As graph measures are one way to characterize network graphs, the thesis should analyze the relation between graph features and network flexibility. It will be expected that the studies will be data-intensive; hence, the student should bring preliminaries in carrying out automated and data-intensive analysis. Finally, this thesis should study the predictability of flexibility given topologies. For this, it should apply supervised machine learning models to answer the question of whether we can predict the flexibility of given problems for specific networking problems.

Flexibility is an indispensable feature of future communication networks and technologies. Only flexible networks allow to cope with the fast-changing demands of future network applications and use cases like 5G/6G. In order to quantify the flexibility of networks, we have recently presented a new flexibility measurement framework. Many problems in network communication are studied on graphs. This internship centers around a particular use case of Controller Placement Problem and investigate the impact of topologies on flexibility of dynamic control plane by applying the concepts as proposed in our framework. As graph measures are one way to characterize network graphs, the thesis should analyze the relation between graph features and flexibility. It will be expected that the studies will be data intensive; hence, the student should bring preliminaries in carrying out automated and data intensive analysis. Finally, this internship should study the predictability of flexibility given topologies. For this, it should apply supervised machine learning models to answer the question whether we can predict the flexibility of dynamic control plane.

The student will analyze the performance of an existing SDN testbed and, if required, adapt the controller software (OpenDaylight) himself / herself. Furthermore, he/she will setup and extend the testbed with readily-available TSN forwarding devices. Finally, the student will evaluate the suitability of the TSN devices for deployment in a number of pre-selected industrial applications. Subsequently, he/she will take part in formulation and deployment of configurations of the physical hardware (i.e., time schedules, forwarding tables and features such as pre-emption and frame replication).

The position is in industry with Siemens AG. The student will be supervised by an university contact as first supervisor, with industrial contact as second supervisor. Extensions into obligatory (research) internships / thesis are not excluded.

- Linux Networking Knowledge

- Ideally first-hand experiences with SDN as technology or development experience with common SDN platforms

- Some programming experience is required (Java, Python, NETCONF/YANG)

- Proficient English and/or German communication skills

- Ability to work and pursue solutions under limited supervision

- Motivation to work with real hardware deployments

Ermin Sakic - ermin.sakic@tum.de

Thanks to the ease of deployment and low cost, Wireless Sensor Networks (WSN) is a good candidate technology for many applications.

In the context of industrial applications, WSN provides the necessary communication infrastructure for sensors and actuators to wirelessly operate in closed-loop control systems called Networked Control Systems (NCSs).

WSN devices, however, are subject to packet loss due to the low-power communication technology and external disturbances.

This effect severely degrades the quality of control and can lead to the instability of the system under control [1].

In this work:

• An industrial wireless sensor network will be deployed - using Zolertia Re-Motes and the Contiki-NG firmware.
• A measurement-based model of the WSN channel quality will be obtained.
• A simulation of an LTI NCS will be implemented [3].
• The transient QoC of NCS will be evaluated following the NCS benchmarking methodology in [2].

[1]: Zhang, Lixian, Huijun Gao, and Okyay Kaynak. "Network-induced constraints in networked control systems—A survey." IEEE transactions on industrial informatics 9.1 (2012): 403-416.

[2]: Zoppi, S., Ayan, O., Molinari, F., Music, Z., Gallenmüller, S., Carle, G., & Kellerer, W. "NCSbenchReproducible Benchmarking Platform for Networked Control Systems."

[3]: Music, Z., Molinari, F., Gallenmüller, S., Ayan, O., Zoppi, S., Kellerer, W., ... & Raisch, J. (2018). Design Of a Networked Controller For a Two-Wheeled Inverted Pendulum Robot. arXiv preprint arXiv:1812.03071

This work will consist of a significant amount of programming for embedded devices, and for a large part also of processing the measurement data.

Requirements:

• Strong C and Python programming skills.
• Good knowledge of wireless systems and protocols (WSN).
• Basic knowledge of LTI control systems.
• Experience with python/Matlab data processing tools is beneficial.
• Experience with embedded systems is beneficial but not required.

Networked control systems are feedback control loops communicating over communication networks. Their dynamics can be modeled in discrete time in a practical, digital setting.

The goal of this project to setup a graphical user interface (GUI) based testbed that emulates such applications, e.g., inverted pendulum, cruise control etc. The packets will be sent/received through a UDP socket. The real-time performance of the control application should be animated and illustrated nicely. For instance we should see a pendulum balancing on the monitor.

Example: https://www.youtube.com/watch?v=qjhAAQexzLg

This work requires:

• Fundamentals of control theory (e.g. Regeulngssysteme I in Bachelor EI)
• Fundamentals of communication systems (e.g.Kommunikationsnetze in Bachelor EI, Broadband Communication Networks in Master EI / MSCE)
• Experience in GUI programming preferably QT Gui with C++ (1st preference) or Python GUI (2nd preference)
• Experience in socket programming is an advantage but can be learned quite fast

The Wireless Sensor Networks lab offers the opportunity to develop software solutions for the wireless sensor networking system, targeting innovative applications. For the next semester, a position is available to assist the participants in learning the programming environment and during the project development phase. The lab is every Tuesday from 15:00 to 18:00.

Solid knowledge in Wireless Communication: PHY, MAC, and network layers. Solid programming skills: C/C++. Linux knowledge. Experience with embedded systems and microcontroller programming knowledge is preferable.

Internetkommunikation offers the opportunity to develop interesting software solutions for technical questions about internet protocols and mechanisms. For the next semester, a position is available to assist the teaching assistants for the tutorials and class project.

• Knowledge in computer networking

• Good programming skills in Python

Network Function Virtualization (NFV) is becoming a promissing technology in modern networks. A challenging problem is determining the placement of Virtual Network Functions (VNFs). In this work, we plan to implement existing algorithms for embedding VNFs chains in NFV-enabled networks. 

Experience in Python or Java, object oriented programming

amir.varasteh@tum.de

The lecture Broadband Communication Networks (Prof. Wolfgang Kellerer) teaches network-related methods of mobile communication: WIFI, 2G to 5G cellular networks, etc. In order to bridge the gap between the methods and real life, innovative teaching concepts shall be developed in form of short podcasts where the students learn in short episodes about wireless communication and networking in day to day scenarios. In the podcasts the student should also be introduced to short exercises they can perform on their own.

In order to support designing these podcasts Pro. Kellerer is looking for a student experienced in programming and maybe in podcasts to help him.

Very good programming skills; experience with podcasts; experience with app programming (on android/ios)

The Wireless Sensor Networks lab offers the opportunity to develop software solutions for the wireless sensor networking system, targeting innovative applications. For the next semester, a position is available to assist the participants in learning the programming environment and during the project development phase. The lab is every Tuesday from 15:00 to 18:00.

Solid knowledge in Wireless Communication: PHY, MAC, and network layers. Solid programming skills: C/C++. Linux knowledge. Experience with embedded systems and microcontroller programming knowledge is preferable.

The yle="color: #333333; font-family: Arial, sans-serif; font-size: 14px;">Software-Defined Networking (SDN) lab offers the opportunity to work with real hardware on interesting and entertaining projects related to the SDN network paradigm. For the next semester, a position is available to assist the teaching assistants for the lab (definition and preparation of the assignments, preparation of the hardware, etc.).

yle="list-style-type: square; margin-left: 0px; padding-left: 20px; color: #333333; font-family: Arial, sans-serif; font-size: 14px;">
• yle="padding-left: 0.3em;">Solid knowledge in computer networking (TCP/IP, SDN)
• yle="padding-left: 0.3em;">Solid knowledge of networking tools and Linux (iperf, ssh, etc)
• yle="padding-left: 0.3em;">Good programming skills: C/C++, Python