# Offered Theses

Please contact the doctoral researchers directly if you are interested in a Bachelor or Master thesis, a student job, an "Ingenieurspraxis" or a "Forschungspraxis". It is also usually possible to find a topic that matches your specific interests. Please include a **curriculum vitae** together with a **list of attended courses** when applying for a thesis. If your "Ingenieurspraxis" is selected to be supervised by one of our professors, please hand in the documents to Doris Dorn (Room N2401).

Quantum Games

## Quantum Games

**Keywords: **

Quantum Information, Algorithm

**Short Description: **

Analyze Quantum Games

#### Description

There are games which had a greater winning chance with entanglement. Goal of this thesis is to analyze this game and calculate examples.

#### Prerequisites

Linear Algebra, Information Theory

#### Supervisor:

Capacity Bounds for Time and Bandwidth Constraint Transmissions

## Capacity Bounds for Time and Bandwidth Constraint Transmissions

**Keywords: **

Energy concentration, Prolate Spheroidal Wave Functions, Sphere packing Bound, Sinc pulses, Raised Cosine Pulses

#### Description

In Shannon's paper [1], where the sphere packing bound is introduced, it is outlined how to calculate the finite block length capacity for a Gaussian channel if the required parameters are known. However, the transmit waveform is allowed to have infinite duration. We want to examine what happens when we introduce constraints on the energy concentration of the waveform, i.e., most of its energy is concentrated in a time interval T and a bandwidth W. The problem of the maximal energy concentration was solved in [2]. We want to find upper and lower bound for the finite block length capacity with these constraints.

[1] C. Shannon, "Probability of error for optimal codes in a Gaussian channel", *The Bell System Technical Journal*, 1959

[2] D. Slepian, H. O. Pollack, H. J. Landau, "Prolate Spheroidal Wave Functions, Fourier Analysis and Uncertainty I-V", *The Bell System Technical Journal*, 1961-1978

#### Prerequisites

- Digital Communications, Digital Communications II
- Information Theory
- Python/MATLAB

#### Contact

delcho.donev@tum.de

#### Supervisor:

Concatenated Codes for Error Correction in DNA Storage

## Concatenated Codes for Error Correction in DNA Storage

#### Description

Encoding information into synthetic DNA is a novel approach for data storage. Due to its natural robustness and size in molecular dimensions, it can be used for long-term and very high-density archiving of data. Since the DNA molecules can be corrupted by thermal processes and the writing/reading process of DNA molecules can be faulty, it is necessary to encode the data using error-correcting codes. Due to the channel model concatenated codes are a suitable candidate for efficient error correction.

The student will analyze existing schemes for error correction in DNA storage based on concatenated codes and develop improved methods using soft information in the outer code, unequal error protection and list recovery. The improvements will be analyzed anlytically and by simulations.

#### Prerequisites

- Channel coding, basic probability theory, experience in programming

- Optional: Coding theory for storage and networks

#### Supervisor:

Analysis of Deep Neural Networks using Information Theory

## Analysis of Deep Neural Networks using Information Theory

#### Description

The aim of this thesis is to take the recently introduced methods for explaining individual predictions of DNNs and adapt them to build statistical methods using information theoretic quantities that can help in understanding the internal functionality of the DNN. This can later be used to improve the performance of the DNN or to reduce the inference complexity by pruning the parts which do not play a significant role in the operation of DNN.

The work will consist of both theory and experimentation.

#### Prerequisites

- Basic knowledge information theory

- Basic knowledge of DNNs and their operation.

#### Supervisor:

Two-way MIMO Communications

## Two-way MIMO Communications

**Keywords: **

MIMO, interactive communications, channel estimation

#### Description

Massive MIMO, or equipping a huge number of co-located antennas to a base station, has been considered as a key enabling technique for 5G to fulfill the high performance requirements in terms of spectral efficiency, energy efficiency, coverage, and reliability [Larsson2014 ]. We study massive MIMO on the frequency division duplexing (FDD) mode. MIMO FDD is one of the most challenging problems in the MIMO FDD, because the resource overhead is overwhelming with a conventional closed-loop channel estimation and feedback [Caire2010]. The project aims to take a fresh look at this difficult open problem.

Let us consider two-way point-to-point MIMO FDD channels and assume that both nodes have M antennas each. W_1 to node 2 over the M * M channel H while node 2 wishes to convey a message W_2 over the M * M channel G. This model with FDD MIMO if into a special case of a two-way communication where both nodes share the same resource [Chapter 17.5, El2011]. Note that the two-way channel is used either to feedback the observations or to convey fresh information symbols. The amount of time needed for feedback depends on the size of the MIMO channels.

Possible directions are:

- Study achievable transmission strategies for the case of perfect channel.
- Characterize the capacity region.

References

[Larsson2014] EG Larsson, O. Edfors, F. Tufvesson, and TL Marzetta, `` Massive MIMO for next generation wireless systems '', IEEE Communicaions Magazine, vol. 52, no. 2, pp. 186-195, 2014

[Caire2010] G. Caire, N. Jindal, M. Kobayashi, and N. Ravindran, `` Multiuser MIMO achievable rates with downlink training and channel state feedback '', IEEE Transactions on Information Theory, vol. 56, no. 6, pp. 2845-2866, 2010

[El2011] A. El Gamal and YH Kim, `` Network Information Theory '', Cambridge University Press, 2011

#### Prerequisites

- Basic knowledge of network information theory, signal processing, linear algebra.
- Matlab programming skills.

#### Contact

Prof. Mari Kobayashi

Room: N406

mari.kobayashi@tum.de

#### Supervisor:

Unmanned Aerial Vehicle (UAV)-aided Cellular Networks

## Unmanned Aerial Vehicle (UAV)-aided Cellular Networks

**Keywords: **

UAV, multi-cell broadcast channel, relay channel, feedback

#### Description

Recently, UAVs (ie, known as drones) have recently been viewed as highly selective (see eg [Zeng2016]).

Considering the network scenario depicted in Figure 1. One macro BS communicating with groups of users through relaying UAVs due to non-line of sight (NLoS) between the macro BS and each group of users. By assuming that the macro BS is equipped by radar, we consider a UAVs UAVs while estimating the channel state of each UAV. The channel estimation is performed by generalized feedback, ie a round-trip channel output available at the BS. The channel at hand to multi-cell broadcast channels with radar-aided backhaul links, or hierarchical downlink channels. Note that multi-cell broadcast channels have been extensively studied in literature (see [Gesbert2010] and references therein). In our setting, the BS-UAV backhaul link evolves in time due to mobility of UAVs.

Possible directions are:

- For the case of a single and static relaying UAV, study relay strategies to maximize the network throughput.
- Study the tradeoff between the quality of channel estimation and the resulting network throughput.
- Generalize to the case of multiple static relaying UAVs or / and moving UAVs.

This thesis can be done in collaboration with Prof. David Gesbert at EURECOM, Sophia-Antipolis, France.

References

[Zeng2016] Y. Zeng, R. Zhang, TJ Lim, `` Wireless communications with unmanned aerial vehicles: opportunities and challenges '', IEEE Communications Magazine, vol. 54, no. 5, pp. 36-42, 2016

[Gesbert2010] D. Gesbert, S. Hanly, H. Huang, S. Shamai, O. Simeone, and W. Yu, `` Multi-cell MIMO Cooperative Networks: A New Look at Interference '', IEEE Journal on Selected Areas in Communications, vol. 28. no. 9, pp. 1380-1408, 2010.

#### Prerequisites

- Basic knowledge of network information theory, signal processing.
- Matlab programming skills.

#### Contact

Prof. Mari Kobayashi

Room: N406

mari.kobayashi@tum.de

#### Supervisor:

Polar Coding with Non-Binary Kernels

## Polar Coding with Non-Binary Kernels

#### Description

This thesis will focus on polar codes with non-binary kernels on GF(q). Some of the following tasks might be covered:

- Kernel selection
- Decoder implementation
- Efficient construction
- Comparison of binary and non-binary polar codes

#### Prerequisites

- Channel Coding
- Information Theory
- Matlab/C++

#### Supervisor:

Adaptive List Decoding for Polar Codes

## Adaptive List Decoding for Polar Codes

#### Description

The finite-length performance of polar codes can be improved by using successive cancellation list decoding. In this thesis, decoder design/implementation and performance prediction are investigated.

#### Prerequisites

- Information Theory
- Channel Coding
- Channel Codes for Iterative Decoding
- Matlab/C++

#### Supervisor:

Simulation of Improved Staircase Code Decoding

## Simulation of Improved Staircase Code Decoding

#### Description

Staircase codes, as introduced in 2011 by Smith et. al. [1], are a hardware friendly code design for error correction in optical communication systems. However, the choices of parameters such as block size and code rate that achieve a desired output bit error rate are limited by the error floor of the decoder. A new and improved decoder has been devised [2], allowing for staircase codes with a scope of new parameters to be considered for optical communication. While estimations show the significant improvements, the high throughput required to simulate the error floor can only be achieved with an efficient and parallelisable implementation. The main goal of the thesis is the implementation in VHDL and simulation on an FPGA of the new decoder in order to provide further evidence for the estimated performance.

[1] https://arxiv.org/abs/1201.4106

[2] https://arxiv.org/abs/1704.01893

#### Prerequisites

interest in channel coding, knowledge in VHDL

#### Supervisor:

Capacity Bounds for Time and Bandwidth Constraint Transmissions

## Capacity Bounds for Time and Bandwidth Constraint Transmissions

**Keywords: **

Energy concentration, Prolate Spheroidal Wave Functions, Sphere packing Bound, Sinc pulses, Raised Cosine Pulses

#### Description

In Shannon's paper [1], where the sphere packing bound is introduced, it is outlined how to calculate the finite block length capacity for a Gaussian channel if the required parameters are known. However, the transmit waveform is allowed to have infinite duration. We want to examine what happens when we introduce constraints on the energy concentration of the waveform, i.e., most of its energy is concentrated in a time interval T and a bandwidth W. The problem of the maximal energy concentration was solved in [2]. We want to find upper and lower bound for the finite block length capacity with these constraints.

[1] C. Shannon, "Probability of error for optimal codes in a Gaussian channel", *The Bell System Technical Journal*, 1959

[2] D. Slepian, H. O. Pollack, H. J. Landau, "Prolate Spheroidal Wave Functions, Fourier Analysis and Uncertainty I-V", *The Bell System Technical Journal*, 1961-1978

#### Prerequisites

- Digital Communications, Digital Communications II
- Information Theory
- Python/MATLAB

#### Contact

delcho.donev@tum.de

#### Supervisor:

Analysis of Deep Neural Networks using Information Theory

## Analysis of Deep Neural Networks using Information Theory

#### Description

The aim of this thesis is to take the recently introduced methods for explaining individual predictions of DNNs and adapt them to build statistical methods using information theoretic quantities that can help in understanding the internal functionality of the DNN. This can later be used to improve the performance of the DNN or to reduce the inference complexity by pruning the parts which do not play a significant role in the operation of DNN.

The work will consist of both theory and experimentation.

#### Prerequisites

- Basic knowledge information theory

- Basic knowledge of DNNs and their operation.