Lunds Tekniska Högskola

Denna sida på svenska This page in English

Computer Science Master Thesis Seminars


Schema för presentationer den 29 augusti.

Program för dagen med populärvetenskapliga sammanfattningar: 



 Schema och abstracts:


Henrik Gyllensvärd, Niklas Welander: Variant Handling for Mobile Web Applications

Examiner: Lars Bendix 

Erik Karlsson: Mapping and Scheduling DAGs on a Many-Core NoC Architecture

Examiner: Flavius Gruian


Oguz Taskin: Compression of Surface Data on Dynamic Topologies

Examiner: Michael Doggett

Jesper Lekland, Asmir Sabanovic: Gesture Recognition With The Sony Smartband

Examiner: Pierre Nugues


Gustaf Waldemarson: Packet Ray Tracing with the ARM NEON Architecture

Examiner: Michael Doggett

Carl Fagerlin: Analysis of Large Data Streams in Support of Automated Decision Making

Examiner: Pierre Nugues

12-13Lunch Break

Dennis Andersson, Astrid Stenholm: Machine CheckUp (A prestudy on developing condition-based maintenance)

Examiner: Klas Nilsson



Magnus Walter, Linnea Johansson Problems experienced in regression testing and how these interact

Examiner: Emelie Engström 



Marie Versland: Assessment Model for System Integration Solutions

Examiner: Per Runeson



Vill du opponera?

Anmäl dig som opponent till den examinator som finns angiven vid den presentation som du är intresserad av, maila till Antagning sker i mån av plats. Ta med dig en skriftlig opponering till presentationen, att dela med examinatorn och den som presenterar. Oppositionen examineras individuellt.

Presentationerna äger rum i E-huset, Ole Römersväg 3. Lokaler: E:2116, E:2405


Obs! Två andra presentationer är på gång, om du vill opponera en annan dag:

  • Point-to-point autentisering för internet of things, Martin Larsson, André Ahlfors Dahl. Datum: 1/9 Examinator: Görel Hedin
  • The Role of Stakeholder Inertia in the Requirements Management Process, Krystian Hoffman. Datum: v.36 (?) Examinator: Martin Höst


Kl 09.15: Variant handling for mobile web applications

Författare/Authors: Henrik Gyllensvärd, Niklas Welander

Handledare/Supervisor: Thomas Hermansson

Examinator/Examiner: Lars Bendix (LTH) - OBS! Denna presentation är inte öppen för opposition.

Lokal/Room: E:2116


Developing native third-pary applicaitons for mobile platforms today generally creates the same applicaiton, but for different systems. Web technology, which refers to HTML5, CSS and JavaScript, has become a possible solution in writing native code on some mobile platforms. If several platforms adopts the same technology it will ease the developmet overall, but does not solve the problem about several different code bases that each require individual attention.

We want to investigate if it is possible to combine several code bases into one, for platforms supporting web applications. This should reduce or eliminate the double maintenance problem but instead introduces a problem with variants. That is another part we want to investigate, is it possible to handle the variants created? We also want to investigate if and how we can minimize the amount of platform specific code for communicating with the platforms hardware.


Kl 09.15: Mapping and Scheduling DAGs on a Many-Core NoC Architecture

Författare/Authors: Erik Karlsson

Handledare/Supervisor: Aneta Vulgarakis (ABB)

Examinator/Examiner: Flavius Gruian (LTH)

Lokal/Room: E:2405

Current multicore architecture has a relatively low upper limit of how many cores can be feasibly put onto one chip before the communication bus becomes the bottleneck of the system. To solve the scaling problems with multicore architectures manycore with Network on Chip architecture has been introduced. Real-time scheduling algorithms need to be adapted to be able to schedule applications onto manycore systems. To address this, we have implemented an offline real-time scheduler into the PASA framework that is under development at ABB Corporate Research. To evaluate our scheduler we have run it with various hardware and software specifications as input. We have also run a multicore scheduling algorithm that we know performs well with the same input parameters to compare the resulting schedules. The results show that our way of calculating network communication costs is in many cases unacceptable as it yields too great of a cost. While our base algorithm for scheduling the nodes are shown to scale well.


Kl 10.15: Compression of Surface Data on Dynamic Topologies

Författare/Authors: Oguz Taskin

Handledare/Supervisor: Jan Schmid (DICE)

Examinator/Examiner: Michael Doggett (LTH)

Lokal/Room: E:2116

Mapping of textures onto surfaces has for a long time been the preferred way of providing detail to a surface. Problems with this technique such as discontinuities and seams appearing in the textured 3D model has lead to the development of different solutions to eliminate these artifacts. “Mesh Colors” proposed by C,Yuksel (2010) presents a solution where surface data is directly associated with the mesh in a topologically independent way removing the need for a map.

The purpose of this study is to develop an image codec for mesh colours data, data directly associated with the mesh. It focuses on compression of colour data but supports compression of textures usually of interest in 3D renderings such as normal maps and specular maps.

The result is a JPEG like image codec with comparable compression ratios and quality of reconstructed images. Because of the similarities to the JPEG standard arising compression artifacts resemble those that appear in reconstructed images compressed with JPEG codecs.


Kl 10.15: Gesture Recognition With The Sony Smartband

Författare/Authors: Jesper Lekland, Asmir Sabanovic

Handledare/Supervisor: Håkan Jonsson

Examinator/Examiner: Pierre Nugues (LTH)

Lokal/Room: E:2405

This thesis takes a single wrist worn tri-axial accelerometer and applies to the ever growing world of personal fitness. Being able to get instant feedback and keep good track of your progress at the gym is for many a taxing and often expensive ordeal with personal trainers and the burden of handwriting their progress in a training diary.

In this paper we strive towards automating the training diary and research how this thesis can be further developed to resemble a digital personal trainer. We explore if and to which extent a simple, commercially available,  tri-axial accelerometer can provide enough information to correctly categorize what a user is currently doing at the gym.

The thesis involves developing an Android application that in collaboration with the Sony Smartband can accurately predict the muscle group(s) a user is engaging at the gym.


Kl 11.15: Packet Ray Tracing with the ARM NEON Architecture

Författare/Authors: Gustaf Waldemarson

Handledare/Supervisor: Johan Grönqvist (ARM)

Examinator/Examiner: Michael Doggett (LTH)

Lokal/Room: E:2116

The ray tracing algorithm has long been used to create near photorealistic images and even simple ray tracers can simulate effects such as shadows, reflections and refractions where other methods struggle. Today, the state-of-the-art ray tracers are typically those that are able to leverage instruction level parallelism by using SIMD extensions to trace multiple rays simultaneously. Within the ARM processor the SIMD architecture is known as NEON and it is used to speed up applications such computer graphics. Thus, in this thesis we have investigated the implementation and performance of a ray tracer that utilizes the NEON architecture to trace packets of rays similar to modern ray tracing frameworks. To determine its efficiency, we also developed a single-ray tracer as a reference and compared them in regard to runtime and power consumption. In the end, we found that the optimized ray tracer performs around 150 - 300% better than the reference.


Kl 11.15: Analysis of Large Data Streams in Support of Automated Decision Making

Författare/Authors: Carl Fagerlin

Handledare/Supervisor: Jacek Malec (LTH)

Examinator/Examiner: Pierre Nugues (LTH)

Lokal/Room: E:2405


"Big data" is becoming an increasing concern for many companies and corporations relying on increasingly larger volumes of data streams and analysis of said streams to make correct decisions. In this paper, the author attempts to cluster transaction requests, originating from data streams, according to set parameters to identify users with unwanted or unsavory behavior. Two strategies are employed: sequence alignment, primarily used for large bulks of transaction requests, comparing a parameterized transaction request to a gold standard, and secondly a sliding window, for real-time evaluation of transaction requests using pattern recognition with the state machine-based API Esper. Using sequence alignment a small subset of users are correctly identified with relatively high probability as unwanted users, though the outcome will heavily depend on the sequence alignment algorithm used as well as the input parameters.

For the sliding window, the patterns implemented correctly identify all cases for very large perpetual data streams. However it places a heavy emphasis on correctly identifying existing patterns in said data streams.


Kl 13.15: Machine CheckUp (A prestudy on developing condition-based maintenance)

Författare/Authors:  Dennis Andersson, Astrid Stenholm 

Handledare/Supervisor: Niklas Svanberg (Tetra Pak)

Examinator/Examiner: Klas Nilsson (LTH)

Lokal/Room: E:2116

Condition-based maintenance (CBM) is a method which is used to make decisions whether or not to perform maintenance on a machine based on information gathered from the machine or components of interest. This master’s thesis was carried out at a company within the packaging industry and the goal was to implement a CBM tool with the possibility to estimate remaining useful life (RUL) for a set of critical components which could serve as a base for further development within the company. Selecting the components was also a part of this thesis. The process of implementing condition-based maintenance was more difficult than expected and the goal of estimating RUL was not met for any of the components, but the work that has been done can be useful for further development. Thus, this thesis will serve as a prestudy on developing condition-based maintenance and contains information of what is required in order to be successful.



Kl 14.15: Problems experienced in regression testing and how these interact

Författare/Authors: Linnea Johansson, Magnus Walter

Handledare/Supervisor: Magnus C Ohlsson (System Verification)

Examinator/Examiner: Emelie Engström (LTH)

Lokal/Room: E:2116


The aim of this study was to reveal a wide range of regression testing problems experienced in a real-life industrial context. A case study was performed at a software consultant agency which identified a total of 90 project-specific and 14 general problems. The main variations identified among these was a division between manual and automated testing; problems not linked to a role were also independent of time; architectural and environmental problems were most difficult to influence; architectural, environmental and strategic problems had the most profound effect on the regression testing effort. It was concluded that there is a need for test strategies designed for a regression specific context; the lack of test strategies has a large effect on the overall quality of the regression testing. It was also revealed that some types of problems could not be amended despite being severe, for example architectural problems in the SUT.



Kl 15.15: Assessment Model for System Integration Solutions

Författare/Authors: Marie Versland

Handledare/Supervisor: Elizabeth Bjarnason (LTH)

Examinator/Examiner: Per Runeson (LTH)

Lokal/Room: E:2116


In today’s society there are a high number of systems that need to integrate with each other. Systems integration solutions are not always optimal and some are in need of being modified. But to evaluate the status of an integration solution the current solution first needs to be assessed. This report presents an assessment model that is meant to help with assessments of system integration solutions. Interviews and background reading was done in order to design the model. The model has been discussed with veterans within systems integration and has been applied on a company’s integration solution in order to make the model as accurate and reliable as possible. The results from the model describes an integration solution and the model makes the assessment less person depending.