It is important to efficiently and correctly be able to detect and classify network traffic, both legitimate and malicious. The slow rate category of DoS attacks makes this task especially hard, as the generated traffic resembles legitimate traffic. This thesis proposes a specialized packet signature for the slow DoS attack Slowloris, as a result of a traffic analysis comparing legitimate traffic and malicious Slowloris traffic. The analysis was performed through packet inspection with a network protocol analyzing tool. The proposed signature focuses on different data sizes of each packet and payload inspection, specifically assessing the beginnings and endings of the payload. To evaluate the signature, it was implemented as a detection tool which was then exposed to normal rate legitimate, slow rate legitimate and malicious traffic. The tool was evaluated by measuring its accuracy and false positive rate. Each evaluation consisted of 100,000 packets and was repeated 50 times, resulting in an evaluation set of five million packets. The tool achieved an average accuracy rate of 98,3% and a false positive rate of 0,0%.

Modeling and Simulation are usually used to solve real-world problems safely and efficiently by constructing digital models of Cyber-Physical Systems. The models can be simulated and analyzed with respect to requirements, and decisions about their design can be based on this analysis. In the latest years, the field of Modeling and Simulation has grown massively and is tackling systems with increased complexity. Thus, the process of modeling and simulating Cyber-Physical systems is becoming more and more complex. This increase requires modeling languages that can express systems with increasing complexity. Modelica is an open-standard declarative equation-based object-oriented language used to model various systems expressed using equations. Modelica tools can read the models, process them, and simulate them. However, the Modelica language and tools cannot express some concepts such as structural changes to the components or behavior of Cyber-Physical Systems during Simulation. In this thesis, we propose extensions of the Modelica language to support modeling so-called variable structure systems, that is, systems where the structure of the system varies during Simulation. The full Modelica language and the new extensions are supported by a novel composable programming environment framework called OpenModelica.jl written in the Julia language. The proposed Modelica language extensions can handle explicit and implicit modeling of variable structure systems by introducing new operators and, consequently, new semantics to the Modelica language. The explicit modeling is based on extensions that switch at runtime between continuous modes of operations with operators similar to the ones used in the specification of Modelica state-machines. The implicit modeling supports reconfiguration during runtime via recompilation. A Just-in-time compiler was implemented to handle the new semantics using the symbolic-numeric programming language Julia. We investigate the performance of our new framework and compare it with existing state-of-the-art Modelica tools on models with thousands of equations and variables. The results show that our extensions and proposed runtime framework is viable for simulating both usual Modelica models and models with variable structure systems. The conclusion is that the Modelica language can be extended further to support systems with variable structures with the addition of a few operators and JIT enhanced runtime system support. Based on the result of this thesis, we propose several directions for future work. / <p><strong>Funding agencies: </strong>This work has been supported by the Swedish Government in the ELLIIT project andby Vinnova in the ITEA3 EMBRACE project. Support has also been received from the Swedish Strategic Research foundation (SSF) in the LargeDyn project. The development of OpenModelica is supported by the Open Source Modelica Consortium.</p><p>ISBN has been corrected in the PDF-version. 2022-06-03</p><p></p>

No description available.

In this thesis, the relationships between local weather and the performance of a cellular telecommunications network were investigated by means of data analysis. Models of the average daily cycles in the data were developed accounting for cyclic behaviour and seasonal trends. Several different analysis methods were then performed on data measuring deviations from these average cycles. The methods used were multiple linear regression, partial least squares regression, calculation of Spearman's correlation coefficients, and regression by artificial neural networks. Some of the results indicate that the number of calls being attempted in the network is related to the local weather conditions. Additionally, small indications were found that the percentage of failed calls in the network was related to the amount of precipitation. These findings could potentially be valuable for network operators. / I detta examensarbete undersöktes samband mellan lokala väderförhållanden och indikatorer på nätverksprestanda för ett mobilnätverk. Undersökningen genomfördes med hjälp av dataanalys. Modeller för den genomsnittliga dagliga variationen i de undersökta parametrarna utvecklades där hänsyn togs till cykliska trender och säsongsberoende. Ett flertal analysmetoder tillämpades sedan på data som mätte avvikelser från de genomsnittliga variationerna. Metoderna som användes var linjär regression, ’partial least squares’ regression, uträkning av Spearmans rangkorrelation, och regression med hjälp av artificiella neuronnät. Resultaten indikerade att antalet samtal som försökte kopplas upp i nätverket influerades av väderförhållandena. Dessutom kunde små indikationer urskiljas på att nederbörd påverkade andelen misslyckade samtalsuppkopplingar i nätverket. Resultaten kan potentiellt vara av värde för nätverksoperatörer.

No description available.

Recommendation systems is an area within machine learning that has become increasingly relevant with the expansion of the daily usage of technology. The most popular approaches when making a recommendation system are collaborative filtering and content-based. Collaborative filtering also contains two major sub approaches memory-based and model-based. This thesis will explore both content-based and collaborative filtering to use as a recommendation system on a sparse boolean dataset. For the content-based filtering approach term frequency-inverse document frequency algorithm was implemented. As a memory-based approach K-nearest neighbours method was conducted. For the model-based approach two different algorithms were implemented, singular value decomposition and alter least square. To evaluate, a cross-approach evaluator was used by looking at the recommendations as a search, a search that the users were not aware of. Key values such as the number of test users who could received a recommendation, time consumption, F1 score (precision and recall) and the dataset size were used to compare the methods and reach conclusions.  The finding of the study was that collaborative filtering was the most accurate choice when it comes to sparse datasets. The implemented algorithm for the model-based collaborative filtering that performed most accurate was Singular value decomposition without any regularization against overfitting. A further step of this thesis would be to evaluate the different methods in an online environment with active users, giving feedback in real time. / Rekommendationssystem är ett område inom maskininlärning som har blivit allt vanligare i och med expansionen av den dagliga användningen av teknik. Det mest populära metoder när du gör ett rekommendationssystemet, “collaborative filtering” och “content-based filtering”. Collaborative filtering innehåller också två sub kategorier, “memory-based” och “model-based”. Denna avhandling kommer att undersöka både “content-based” och “collaborative filtering” för användning som ett rekommendationssystem för ett glest boolesk dataset. Som “content-based” strategi implementerades term frekvens omvänd dokument frekvens (TF-IDF) algoritmen. Som en “memory-based” strategi implementerades K-närmast grannarna (K-NN) metoden. För “model-based” angripsättet implementerades två olika algoritmer, singulärvärdesuppdelning (SVD) och altenerande minsta kvadrat metoden (ALS). För att kunna utvärdera metoderna mot varandra sågs rekommendationer som en sökning, en sökning som användarna inte var medvetna om att det gjort. Viktiga värden som antalet testanvändare som kunde fått en rekommendation, tidsåtgång, “F1 score” (precision och recall) och dataset storlek användes för att jämföra det olika metoderna och dra slutsatser. Resultatet av studien visar att “collaborative filtering” var den högst presterande när det gäller en gles datamängd. Den implementerade algoritmen för “model-based collaborative filtering“ som visat sig vara den mest exakta var SVD utan reglering mot “overfitting”. En framtida påbyggnad av denna rapport är att utvärdera olika metoder i en online-miljö med aktiva användare som kan ge respons i realtid.

Protein localization is an important part in understanding the functionality of a protein. The current method of localizing proteins is to manually annotate microscopy images. This thesis investigates the feasibility of using deep artificial neural networks to automatically classify subcellular protein locations based on immunoflourescent images. We investigate the applicability in both single-label and multi-label classification, as well as cross cell line classification. We show that deep single-label neural networks can be used for protein localization with up to 73% accuracy. We also show the potential of deep multi-label neural networks for protein localization and cross cell line classification but conclude that more research is needed before we can say for certain that the method is applicable.

Att förutspå aktiepriser är något som undersökts i hundra- tals år och kan användas som grund för investeringsbeslut. Under senare år har maskininlärning applicerats på detta problem vilket gjort att ett flertal metoder föreslagits och den elektroniska handelns införande har gjort att högupp- löst data med nanosekundsprecision blivit tillgänglig. I den- na rapport presenteras resultat av en jämförande studie av klassificeringsmetoder som föreslagits för att förutspå ak- tieprisernas rörelser. Högupplöst data för fyra olika aktier från Stockholmsbörsen har använts för att jämföra tre olika metoder. Resultaten visar att det på grund av datas fördelning är svårt att uppnå ett tillfredsställande resultat. Priserna rör sig relativt sett mer sällan när högupplöst data används, vilket gör att ett fåtal felklassificeringar kan göra modellen otillräcklig. Skillnaderna mellan metodernas resultat anty- der dock att valet av indikatorer har stor betydelse och för vidare forskning kan det vara relevant att undersöka såväl innehåll i indata som metoder för att hantera den ojämna fördelningen. Vidare har en litteraturstudie genomförts för att un- dersöka den elektroniska handels bakgrund, syfte och på- verkan. Även om detta inte tycks ha varit det ursprungliga syftet i alla fall verkar den elektroniska handeln ha lett till mer effektiva marknader med lägre transaktionskostnader. / The prediction of stock prices has been attempted for hun- dreds of years and can be used for making investment deci- sions. In later years machine learning has been appliced to this problem, leading to the proposal of several dif- ferent methods. Furthermore, the recent introduction of electronic market places has made available high-resolution data with nanosecond precision. This report presents the results of a study comparing classification methods that have been proposed for predicting the movement of stock prices. High-resolution data from four different stocks on the Stockholm Stock Exchange has been used to compare three different methods. The results show that the distribution of the data makes it difficult to reach a satisfying result. Prices move rela- tively more rarely when high-resolution data is used, mak- ing even a few misclassifications significant. However, the differences between the methods’ results suggest that the choice of indicators is of great importance and for future research it may be relevant looking at both the content of input data and methods of handling the skewed distribu- tion. Furthermore, a literature review has been performed re- searching the electronic trading’s background, purpose and effects. Although it may not have been the intended pur- pose it seems that electronic trading has made the markets more efficient and transaction costs lower.

In this thesis, the performance of a number of models used to predict the result in soccer games has been investigated, using data on the previously played games in the league. Established models were implemented, and tested on a wide set of soccer leagues during several years. The performance of each model was measured using the likelihood ratio against a simple baseline distribution. The performance of these models was then analyzed to find systematic differences correlating with some properties of a soccer league, such as average number of goals in the league, and determine which models overall performed best. The results showed that such differences do exist, correlating with the average number of goals in a league as well as the variance in performance among teams in the league. Additionally, statistically significant differences in the performance of some models were established. / I denna rapport undersöktes prestandan hos ett antal modeller för att förutsäga resultat i fotbollsmatcher, tränade på resultaten i tidigare matcher i ligan. Etablerade modeller implementerades och testades sedan på flera årgångar av ett brett urval av fotbollsligor. Prestandan för varje modell mättes som en likelihood-ratio mot en enkel basdistribution. Modellernas prestanda analyserades sedan för att hitta systematiska skillnader som korrelerar med någon viss egenskap hos en fotbollsliga, t.ex. genomsnittligt antal mål i ligan, samt att avgöra vilka modeller som presterade bäst. Resultaten visade att sådana skillnader finns, och att prestandan korrelerar med dels genomsnittligt antal mål i ligan, men även variansen hos prestandan för lagen i ligan. Dessutom hittades statistiskt signifikanta skillnader i prestanda mellan några av modellerna.

Message queues are data structures that can be distributed over a network to send messages between different end points. These message queues are used for example in micro service based architectures for communication. Some message queues sends one message at a time and some queues batch messages together and sends them in a chunk. This study focused on batch file based message queues and how compression affects latency and throughput versus no compression. The use of the right compression method is vital for reducing network traffic, CPU and memory usage. This study performed tests where data of different sizes were sent from a producer to the queue and then consumed by a consumer. These three components were placed on a locked local machine. Tests were done with different number of messages sent and different sizes of each message. The data that were sent was a map of the Faroe Islands represented as a XML file to create fair compression ratios. The study concluded that at no point had no compression lower latency or higher throughput compared to a tested compression method.