PROJECT (B.SC. LEVEL) - Purdue University · method and the resulting controller MATLAB/Simulink is used. Contact Melanie Schuh ... DVB-T dongles based on the Realtek RTL2832U can

PROJECT (B.SC. LEVEL)Adaptive Signal Processing for Acoustic Enhancement

Noise or

Reverberation

Estimation

Speech

Enhancement

The quality of speech signals captured, for instance, by hearing-aid micro-phones or other mobile devices, can be easily degraded by the environmen-tal noise in daily-life scenarios. Here, it will be beneficial to apply adap-tive signal processing algorithms to reduce, for instance, undesired acousticnoise or reverberation in the signals. This will help to increase sound qualityand intelligibility in speech communication or recognition systems.

In the project, we focus on implementation and verification of two-channelsignal processing related to the binaural-hearing system of human beings,e.g. [1-2]. Matlab/Simulink will be used for technical computing and to en-able a fast transition to a target, for instance PC, Speedgoat-PC, or Rasp-berry Pi. The algorithm performance will eventually be investigated withspeech in the presence of various noise types or reverberation levels.

[1] Azarpour, M., Enzner, G., "Fast Noise PSD Estimation Based on Blind Channel Identi-fication", Prof. of Intl. Workshop on Acoustic Signal Enhancement (IWAENC), Sept. 2014[2] Enzner, G., Merks, I., and Zhang, T. "Adaptive filter algorithms and misalignment criteriafor blind binaural channel identification in hearing-aids", Proc. of EUSIPCO, August 2012

PREREQUISITES: Signals and Systems, MATLAB, and Simulink

PLEASE CONTACT:Dr.-Ing. habil. Gerald EnznerRoom: ID 2 / 227Phone: +49 234 32 25392E-mail: [email protected]

Masoumeh Azarpour, M.Sc.Room: ID 2 / 255Phone: +49 234 32 27543E-mail: [email protected]

ADRESSE Universitätsstraße 150 | 44801 Bochum, Germany ANFAHRT U-Bahn: U35 | Auto: A43, Abfahrt (19) Bochum Witten

WWW.RUB.DE

Controller design for the handling system HANS

The aim of this project is to implement and test a newly developed controller design method for discrete event systems modeled by deterministic Input/Output automata using the handling system HANS. The handling system HANS is a pick-and-place system with various elements such as grippers, deflectors and deposit stations. To complete the project, an existing model for the handling system has to be extended, its controllability has to be proven and different inter-esting control tasks have to be considered. For the implementation of the controller design method and the resulting controller MATLAB/Simulink is used.

Contact Melanie Schuh Raum ID-2/539 E-Mail: [email protected] Tel.: 0234 32 26632 Handling System HANS

FAKULTÄT FÜR ELEKTROTECHNIK UND INFORMATIONSTECHNIK

PROF. DR.-ING. JAN LUNZE

Digital Communication Systems

ProjectDigital Radio Receiver with Raspberry Pi

1 Goal

To implement a Digital Audio Broadcast(DAB) receiver on Raspberry Pi[1] using RTL-SDR dongles. Thereceiver should do the following.

• Make a list of FM channels by detecting peaks in Power Spectral Density.

• Scan the DAB spectrum and make a list of channels.

• Accept an input(channel number) from the user, and play corresponding channel.

2 Details

DVB-T dongles based on the Realtek RTL2832U can be used as a cheap SDR[2], since the chip allows transfer ofraw I/Q samples to the host. The RTL2832U outputs 8-bit I/Q-samples, and the highest possible sample-rate is3.2 MS/s. This would be sufficient to receive DAB transmission. In this project we would develop a digital radioreceiver on Raspberry Pi. The Pi would accept raw I/Q data from the RTL dongle and execute the protocolstack[3] to decode the radio transmission. The GPU on Raspberry Pi has hardware MPEG decoder. We wouldalso investigate the possibility of using the GPU do decode the DAB+ audio.

3 Prerequisites

• A first course in Communication Systems

• C++/C skills

Literatur

[1] “Raspberry Pi.” https://www.raspberrypi.org/.

[2] “RTL-SDR Wiki.” http://sdr.osmocom.org/trac/wiki/rtl-sdr.

[3] “Wikipedia on DAB.” http://en.wikipedia.org/wiki/Digital_Audio_Broadcasting.

(a) (b)

Abbildung 1: The Pi Radio :)

Digital Communication SystemsRuhr-University-BochumUniversitatsstraße 150D-44780 Bochum, Deutschland

15. Juli 2015

www.dks.ruhr-uni-bochum.de

Kontakt: Manu Thottathil SreedharanMail: [email protected]: ID2/348Tele: +49234 32-23682

Digital Communication Systems

ProjectLDPC Codes in GNU Radio FEC API

1 Goal

To implement Low-Density Parity-Check(LDPC) Codes in GNU Radio Forward Error Correction(FEC) App-lication Programming Interface(API). Expected outcomes are:

• Incorporate LDPC encoder and decoder into FEC API framework in GNU Radio.

• Obtain bit error rate curve for the aforementioned codec over wireless channel using USRP.

2 Details

LDPC codes [1] form a class of capacity approaching codes. LDPC codes can be described by a tanner graphcorresponding to its parity check matrix as in Figure 1. LDPC codes are decoded with belief propagation [1] onthe tanner graph. A brief tutorial on this topic can be found in [2].

Some of the existing implementations of LDPC codes can be found in [3, 4]. GNU Radio provides a freeand open-source platform for implementing practical communication systems. Recently, FEC API was added toGNU Radio as a framework to implement error correcting codes. Goal of this project is to incorporate LDPCencoder and decoder into FEC API, and obtain a bit error rate curve to benchmark their performance.

3 Requirements

• A first course in Linear Algebra

• A first course in Communication Systems

• C++/C skills

Literatur

[1] T. Richardson and R. Urbanke, Modern Coding Theory. Cambridge University Press, 2008.

[2] A. Shokrollahi, “LDPC Codes: An Introduction.” https://www.ics.uci.edu/~welling/teaching/

ICS279/LPCD.pdf, April 2003.

[3] R. Neal, “Software for Low Density Parity Check Codes.” http://www.cs.utoronto.ca/~radford/ftp/

LDPC-2012-02-11/index.html, July 2013.

[4] FSF, “IT++.” http://itpp.sourceforge.net/4.3.1/group__fec.html, July 2013.

Abbildung 1: A fragment of a tanner graph

Digital Communication SystemsRuhr-University-BochumUniversitatsstraße 150D-44780 Bochum, Deutschland

15. Juli 2015

www.dks.ruhr-uni-bochum.de

Kontakt: Manu Thottathil SreedharanMail: [email protected]: ID2/348Tele: +49234 32-23682

Digital Communication Systems

Project

Information theoretic Analysis of the Board Game Mastermind

Mastermind is a deductive board game for two players, where one player (the codebreaker) tries to guess a codethe other player (the codemaker) constructed.In (n,m)-Mastermind the codemaker constructions a color-code consisting of n pins, where each pin can haveany one of m colors. The number of possible codes generated this way is mn. To break the code, the codebreakerguesses a code and obtains the following information from the codemaker after every guess:

• Number of pins with correct color and correct position

• Number of pins with correct color but wrong position.

The goal is to utilize the obtained information in a smart way and therefore breaking the code with as fewguesses as possible.[1]

One simple approach to do so is the minmax-strategy. It guarantees to break any of the 1296 possible codes inthe (4, 6)-version of Mastermind in at most 5 attempts.[2]

Tasks

This thesis consists of the following parts:

• Information theoretic analysis of the minmax-strategy for the (4, 8)-version of Mastermind

• Implementation of a Mastermind-GUI in Matlab

• Implementation of the minmax-strategy in Matlab

Prerequisites

• Basic knowledge of Information theory

• Experience with Matlab

• Motivation and determination

Literature

[1] http://en.wikipedia.org/wiki/Mastermind-board-game.

[2] http://www.dcc.fc.up.pt/ sssousa/RM09101.pdf.

Digital Communication SystemsRuhr-University-BochumUniversitatsstraße 150D-44780 Bochum, Deutschland

7/14/2015

www.dks.ruhr-uni-bochum.de/en

Contact: Dennis MichaelisMail: [email protected]: ID 2/345Fon: +49234 32-23064

Term project: Implentation of successive inter-ference cancellation on USRP devices

Supervisor: Soheil Gherekhloo

Dealing with interference is a main challenge in wireless communications.Compared with noise, interference contains information. Using this property,some techniques have been investigated which decode the interference in orderto have a cleaner version of the received signal [1]. This was shown to be optimalin some cases such as 2-user interference channel (IC) [2].

As an example, suppose that the received signal is given by

xd + 2xc + z = y,

where xd, xc represent the desired and interference real signals with power P ,respectively. Moreover, z ∼ N (0, 1) represents the additive white Gaussiannoise. The receiver first detects the interference. Then, it removes the impactof the interference from the received signal. Doing this, it obtains

y′ = xd + z.

The goal of this thesis is to implement successive interference cancellationon the USRP devices.

Requirements

• The first course in communication systems

• C++/C skills

References

[1] T. Han and K. Kobayashi, “A new achievable rate region for the interferencechannel,” IEEE Trans. on Info. Theory, vol. 27, no. 1, pp. 49–60, Jan 1981.

[2] H. Sato, “The capacity of the Gaussian interference channel under stronginterference,” IEEE Trans. on Info. Theory, vol. IT-27, no. 6, pp. 786–788,Nov. 1981.

1

Security Analysis of Physical Unclonable Functions 

What is the topic? 

Physical Unconable Functions (PUF) have gained a lot of research attention in 

the last years as a new cryptographic building block. PUFs rely on the fact that 

every chip that is manufactured has some differences to other chips, due to 

manufacturing or the used materials. PUFs magnify these differences between 

physical instances of the same design to create a unique identity for every 

manufactured chip. The advantage of using PUFs compared to traditional 

cryptography is that no key needs to be programmed or stored. Furthermore, 

PUFs are often much smaller than traditional cryptography, making them very 

interesting for constrained embedded devices. 

One problem of PUFs are that they can often be modeled using Machine 

learning algorithms given enough challenge and response pairs. Different 

countermeasures have been proposed to defend against these machine 

learning attacks. In this project we want to look at the security of these 

proposals against machine learning attacks as well as implementation attacks. 

What can you do? 

There are different projects available that require different skills. One focus 

can be on efficiently implementing machine learning attacks with an emphasis 

on parallel computing. Another focus can be in hardware implementations of 

PUFs or on performing side‐channel measurements and attacks on PUFs. 

Contact: 

Chair for Embedded Security Dr. Georg T. Becker Georg.Becker@ruhr‐uni‐bochum.de 

ADDRESS Universitätsstraße 150 | 44801 Bochum, Germany WWW.RUB.DE

Reverse-Engineering on the Netlist Level

Nowadays, digital ASIC and FPGA circuits are designed and fabricated in a global, multi-vendor scenario. This emerging trend of outsourcing possesses an unfortunateattack vector regarding hardware Trojan insertion, trust validation, and integrityverification of third-party Intellectual Property (IP) cores.In this case, reverse-engineering of third-party IP cores (for which the source codeis often not available) can help to detect IP violations or hardware Trojan insertionsfor FPGA and ASIC designs.As reverse-engineering is in general a time-consuming task, algorithmic approachesto analyze digital circuitries are a crucial step towards trust in hardwareimplementations. Furthermore, these approaches can aid adversaries to insertmalicious circuitry in a third-party design.In this research thesis, the problem of algorithmic detection of various hardwarecircuitries, e.g., cryptographic modules and soft-core CPUs in FPGA and ASICnetlists shall be investigated. Additionally, the insertion of malicious circuitrycan be regarded and empirically conducted.

Contact M.Sc. Marc FyrbiakChair for Embedded SecurityRoom: ID2/645Email: [email protected]

FAKULTÄT FÜR ELEKTROTECHNIKUND INFORMATIONSTECHNIK

Chair for Embedded SecurityProf. Dr.-ing. Christof Paar

Fig. 1: Mapping of one AES S-Box column to a slice

At the „Institute of Electronic Circuits“ millimeter wave frequency synthesizers with high

bandwidth and ultra-low phasenoise based on phase-locked-loop techniques are

investigated. In such synthesizers frequency dividers with high maximum input frequencies

and wide division factor ranges are crucial components. At the institute a frequency divider

with an input frequency up to 80 GHz, division factors from 12 to 259 and a moderate

power consumption of 390 mW has been realized as a monolithic microwave integrated

circuit in a SiGe BiCMOS technology. It is based on the dual-modulus concept and emitter-

coupled logic (ECL). This project focus on the reduction of the power consumption of fast

frequency dividers. On one hand a frequency divider based on a divide-by-2/3 chain

should be designed and compared on schematic level to the dual-modulus concept in a

SiGe BiCMOS technology. On the other hand advanced circuit design techniques for

power reduction in the ECL gates should be investigated.

Contact:

Marcel van Delden, B.Sc.

[email protected]

Project

Institute of Electronic CircuitsProf. Dr.-Ing. Thomas Musch

LEHRSTUHL FÜR

Elektronische Schaltungstechnik

Investigations on fast, low powerfrequency dividers

P/P+1 A

fin fps

B

fout

dual-modulus concept

2/3

fin2/3

f12/3

f2 fN

P0 P1 PN-1fout

divide-by-2/3 chain

Delay+Sync.

OutputBuffer

Divider A

Div

ider

B

Logic Interface

Pre

scal

er

100 μm

D-Latch 2Emitter

FollowerD-Latch 1 Emitter

FollowerSet/Reseton Load

L

XOR

Ground Penetrating Radar (GPR) is a metrological method for nondestructive, subsurface imaging. It is

used in a broad variety of applications such as archeology, geophysics, glaciology, or humanitarian

demining. Therefore, electromagnetic waves are radiated into soil by means of GPR antennas (Fig. 1).

Soil inhomogeneities, like buried objects or soil layers, cause reflections which are monitored and

evaluated. Gathering numerous measurements from different positions allow for the calculation of a two

dimensional profile image as shown in Fig. 2. In order to increase the information content of the

measured ground profile, advanced radar signal processing methods, like Synthetic Aperture Radar

(SAR) algorithms, are applied to the raw data. Fig. 3 shows the processed data set from Fig. 2. For

improving GPR methods, test measurements are performed under predefined laboratory conditions by

using scanning frames and stepping motors as shown in Fig. 4. The efficient and reliable detection of

shallow buried objects like anti-personal land mines (Fig. 5), demands for detection algorithms, which

generally rely on 3D electromagnetic radar cross section (RCS) simulations.

Within this main project, we offer three different sub-projects:

1) Simulation and characterization of GPR-antennas

2) GPR test measurements and signal processing

3) RCS simulation and detection algorithms

Contact:

Dr.-Ing. Christoph Baer

[email protected]

Project

Institute of Electronic CircuitsProf. Dr.-Ing. Thomas Musch

LEHRSTUHL FÜR

Elektronische Schaltungstechnik

Ground Penetrating Radar:Antennas, Simulation, and Signal Processing

(5)

(3)(1) (4)

(2)(3)(5)

1/2

Aging studies on automotive lithium-ion cells: Circuit topologies, measurement procedures, modeling and parameter estimation techniques

The research group of Automotive Electronics conducts long-term aging studies on lithium-ion cells for use in battery electric vehicles. The tests are performed in close cooperation with leading German automobile manufacturers and are used to evaluate cycle life under varying load conditions and to analyze the dynamic battery performance. For this purpose specialized single-cell test equipment has been developed by our research group, allowing charge and discharge currents up to 600 A combined with wide bandwidth and high measurement resolution (Fig. 1). More than 120 testers and 10 temperature chambers are currently in use at our test facility, evaluating the next generation of automotive lithium-ion cells (Fig. 2). The obtained measurement data is used for cell modeling and development of real-time parameter estimation techniques for use in onboard battery management systems. Special interest is taken in the inclusion of aging effects and advanced wideband analysis techniques.

Fig. 1 Single cell tester Fig. 2 Test facility

An assortment of possible topics for a research project is listed below.

1. Circuit topologies for high-frequency stimulation of lithium-ion cells with low distortion Advanced frequency-domain measurements require sinusoidal stimulation of the cell with high amplitude and low distortion. Different approaches for a suitable circuit topology are to be compared and evaluated by simulation and measurement.

2/2

2. Methods for low-frequency cell stimulation and application to estimation algorithms Low frequency stimulation of lithium-ion cells offers further insight into electrochemical aging processes, but state-of-the art measurement procedures are very time consuming. In this project optimized time-domain stimulation approaches are to be developed, implemented and compared to existing methodologies, especially with regard to applicability in estimation algorithms.

3. Internal thermal behavior of lithium-ion cells – measurement procedures and modeling The internal cell temperature is an important factor for cell aging, but cannot be measured directly. By exploiting the well-known temperature dependency of internal impedance, a suitable measurement concept is to be developed to accurately track impedance variations during arbitrary load profiles and correlate them to internal cell temperature.

4. Bridging the gap between laboratory parameter estimation and real life application In this project, various real-life driving cycles from our cell excitation database are to be used to evaluate and improve the performance of existing estimation algorithms in both simulation and measurement.

5. Cell modeling and aging effects Simulation of the electrical power train requires accurate battery models with inclusion of aging effects. In this project different aging mechanisms are analyzed and an adaptive cell model valid for different states of aging is developed. The performance of the model is evaluated using synthetic and real-life driving cycles.

Interested students are kindly asked to contact us for detailed project descriptions or additional topics in our area of research.

Contact:

Research Group of Automotive Electronics – Institute of Electronic Circuits Dipl.-Ing. Patrick Wesskamp [email protected]

At the „Institute of Electronic Circuits“ Radar-Systems are investigated and developed.

This project focus on a MIMO-Radar-System with transmitters and receivers based on

SiGe chips at a centerfrequency of 80 GHz and with a bandwidth of 25 GHz. As the

coherence in such MIMO-Systems is key to high precision measurements, concepts for

the distribution of the reference signal, the feeding of the multiple antennas as well as the

synchronized analog to digital conversion for the different channels have to be investigated

and realized. Among the coherence also high dynamic and low noise are main issues

which have to be considered.

Contact:

Marcel van Delden, B.Sc.

[email protected]

Project

Institute of Electronic CircuitsProf. Dr.-Ing. Thomas Musch

LEHRSTUHL FÜR

Elektronische Schaltungstechnik

Investigations on coherent signalingin MIMO-Radar-Systems

TX-2

TX-1

transmitter

RX-2

TX-n

RX-n

reference

TX

AP 4.2.2

receiver

RX

AP 4.2.3

...

RX-1

PFD

/M

fIF

PFD

/M

TX-1

RX-1

ADC

fref

Modern Microscopy (Photonics and Terahertz Technology)

A microscope is a typical tool to analyze tiny structures. Widely used in science and industry, microscopy is important in research of typically small modern devices and objects. Because of the on-going miniaturization of semiconductor and mechanical devices, the need for high resolution microscopes is still growing. It is our aim to improve known microscopy techniques and research new ones. We realized a flexible confocal laser scanning microscope (CLSM) to analyse semiconductor materials in reflection and in micro-photoluminescence mode. The capability to quickly change the laser sources, objectives and filters allows a wide range of experiments with different kinds of sam-ples. Current projects implement phase based methods to improve the axial resolution, using synergy effects of the research projects in the field of holography and optical coherence tomography (OCT). Using various combinations of pinhole, objective and light source, the following specifications can be archieved:

Figure 1: CLSM

500nm lateral resolutions 4µm up to 100µm axial resolution Laser sources: Diode lasers (405nm and 660nm), HeNe Laser and Ti:Sa Laser (700nm

to 900nm)

An important subject is the analysis of semiconductor materials, especially the III/V semicon-ductors for optical integrated circuits. The micro-photoluminescence mode allows spectral imaging in the range of 400nm to 1700nm.

Ultra short pulse generation (Photonics and Terahertz Technology)

Semiconductor lasers are a key technology in the telecommunications sector. The feature of direct electrically pumping allows for a fast modulation of the laser in combination with a very compact setup. Since the end of the 1980’s there is an intense ongoing research of semiconductor lasers as a source for ultrashort laser pulses. But the complex carrier dynamics in the semiconductor material doesn’t allow the direct generation of femtosecond pulses with high average output power from an electrically pumped semiconductor laser. Thus, we have developed a modular system concept as de-picted in figure 1: in an oscillator based on a two section laser diode in an external cavity pulses with a duration of 5 ps are generated. The resonator geometry allows the compensation of the carrier dyna-mics in a fashion that the spectral bandwidth is greatly increased, while the pulse duration stays in the ps-regime.

Fig. 1: Modular setup of high power femtosecond diode laser system

This enables the amplification by an optical amplifier which is also an electrically pumped semicon-ductor device. In a second step the pulse is compressed by an external pulse compressor to some 100 fs. The usage of semiconductor devices enables the possibility to build the whole laser with a foot print as small as 300 * 300 mm². Figure 2 shows the compact femtosecond diode laser system in front of a commercial Ti:Sapphire laser system. All components to generate high power laser pulses are in-cluded.

Fig. 2: Compact semiconductor laser system in front of a commercially available Ti:Sapphire laser

The resonator internal dispersion control allows to generate pulses with a duration as short as 158 fs. These are to our knowledge the shortest pulses which were generated by an electrically pumped semi-conductor laser. This closes the gap to fibre and solid state laser systems. After the pulse amplification pulse energies up to 2 nJ are generated. This leads to a peak power of 6.5 kW.