Millions of new websites are created daily, making it challenging to determine which ones are safe. Cybersecurity involves protecting companies and users from cyberattacks. Cybercriminals exploit various methods, including phishing attacks, to trick users into revealing sensitive information. In Australia alone, there were over 74,000 reported phishing attacks in 2022, resulting in a financial loss of over $24 million. Artificial intelligence (AI) and machine learning are effective tools in various domains, such as cancer detection, financial fraud detection, and chatbot development. Machine learning models, such as Random Forest and Support Vector Machines, are commonly used for classification tasks. With the rise of cybercrime, it is crucial to use machine learning to identify both known and new malicious URLs. The purpose of the study is to compare different instance selection methods and machine learning models for classifying malicious URLs.

In this study, a dataset containing approximately 650,000 URLs from Kaggle was used. The dataset consisted of four categories: phishing, defacement, malware, and benign URLs. Three datasets, each consisting of around 170,000 URLs, were generated using instance selection methods (DRLSH, BPLSH, and random selection) implemented in MATLAB. Machine learning models, including SVM, DT, KNNs, and RF, were employed. The study applied these instance selection methods to a dataset of malicious URLs, trained the machine learning models on the resulting datasets, and evaluated their performance using 16 features and one output feature.

In the process of hyperparameter tuning, the training dataset was used to train four models with different hyperparameter settings. Bayesian optimization was employed to find the best hyperparameters for each model. The classification process was then conducted, and the results were compared. The study found that the random instance selection method outperformed the other two methods, BPLSH and DRLSH, in terms of both accuracy and elapsed time for data selection. The lower accuracies achieved by the DRLSH and BPLSH methods may be attributed to the imbalanced dataset, which led to poor sample selection.

Amid the rapid proliferation of thousands of new websites daily, distinguishing safe ones from potentially harmful ones has become an increasingly complex task. These websites often collect user data, and, without adequate cybersecurity measures such as the efficient detection and classification of malicious URLs, users’ sensitive information could be compromised. This study aims to develop models based on machine learning algorithms for the efficient identification and classification of malicious URLs, contributing to enhanced cybersecurity. Within this context, this study leverages support vector machines (SVMs), random forests (RFs), decision trees (DTs), and k-nearest neighbors (KNNs) in combination with Bayesian optimization to accurately classify URLs. To improve computational efficiency, instance selection methods are employed, including data reduction based on locality-sensitive hashing (DRLSH), border point extraction based on locality-sensitive hashing (BPLSH), and random selection. The results show the effectiveness of RFs in delivering high precision, recall, and F1 scores, with SVMs also providing competitive performance at the expense of increased training time. The results also emphasize the substantial impact of the instance selection method on the performance of these models, indicating its significance in the machine learning pipeline for malicious URL classification

Sustainability reporting is essential for organizations to report their environmental, social, and governance (ESG) performance. However, extracting, and structuring data from sustainability reports can be challenging, leading to inefficiencies and inconsistencies. This project aims to develop an integrated system for sustainability reporting by leveraging artificial intelligence (AI) techniques, particularly natural language processing (NLP), to extract and structure data from sustainability reports. Utilizing the GPT-3 model by OpenAI, the system processes unstructured text from PDF reports into a structured format compliant with European Sustainability Reporting Standards (ESRS). The system efficiently and accurately extracted crucial sustainability information through a meticulously designed pipeline, including PDF parsing, text sanitizing, batching, parallel API requests, and postprocessing. The system’s effectiveness is evaluated using cosine similarity metrics, comparing model outputs with manually extracted data. The results demonstrate high alignment between the model outputs and manual extractions, validating the system’s performance. This project contributes to advancing sustainability reporting practices, providing organizations with a robust tool for transparent and standardized disclosure of ESG impacts.

The continuously evolving nature of today's internet outdates the existing security and safety mechanisms and therefore there is an emerging need to propose robust, powerful and reliable solutions. These advancements have a great impact on the software and system architectures, resulting in a highly dynamic smart networked environment. The systems used within these complex environments have at least two things in common, namely, the need to restrict or grant access for the required resources based on security policy to face security threats, and the need to sustain resilience of the environment in face of safety hazards. Both aspects should consider trust.

Commercial-Off-The-Shelf (COTS) embedded devices rely on vendor-specific firmware to perform essential tasks. These firmware have been under active analysis by researchers to check security features and identify possible vendor backdoors. However, consistently unpacking newly created filesystem formats has been exceptionally challenging. To thwart attempts at unpacking, vendors frequently use encryption and obfuscation methods. On the other hand, when handling encrypted, obfuscated, big endian cramfs, or custom filesystem formats found in firmware under test, the available literature and tools are insufficient. This study introduces DUDE, an automated framework that provides novel functionalities, outperforming cutting-edge tools in the decryption, unpacking, deobfuscation, and endian conversion of firmware. For big endian compressed romfs filesystem formats, DUDE supports endian conversion. It also supports deobfuscating obfuscated signatures for successful unpacking. Moreover, decryption support for encrypted binaries from the D-Link and MOXA series has also been added, allowing for easier analysis and access to the contents of these firmware files. Additionally, the framework offers unpacking assistance by supporting the extraction of special filesystem formats commonly found in firmware samples from various vendors. A remarkable 78% (1424 out of 1814) firmware binaries from different vendors were successfully unpacked using the suggested framework. This performance surpasses the capabilities of commercially available tools combined on a single platform.

Nowadays, there is a mobile application for almost every need, for instance, training, games, and news applications. Almost all the companies have also their own mobile applications that facilitate the daily communication between their employees and supervisors. In this thesis, the development process of an Android application for the company Askås is created, where the employees can for example send a Flex application or sick leave application, see their registered personal information and edit them. The final result of the project is almost what Askås asked for, An Android application that facilitates their employees’ daily needs. The development process, the results, and the encountered problems are also represented in a more detailed manner in the thesis. 

This report concerns the implementation of a simulation framework to evaluate an emerging peer-to-peer network topology scheme using 8-point hypercircles, entitled HyperCircle. This topology was proposed in order to alleviate some of the drawbacks of current P2P systems evolving in an uncontrolled manner, such as scalability issues, network overload and long search times. The framework is supposed to be used to evaluate the advantages of this new topology. The framework has been built on top of an existing simulator software solution, the selection of which was an important part of the development. Weighing different variables such as scalability and API usability, the selection fell on OverSim, an open-source discreet-event simulator based on OMNET++.

After formalizing the protocol for easier implementation, as well as extending it for better performance, implementation followed using C++ with OverSim’s API and simulation library. Implemented as a module (alongside other stock modules providing their own protocols such as Chord and Kademlia), it can be used in OverSim to simulate a user-defined network using one of the simulation routine applications provided (or using a custom application written by the user). For the purposes of this thesis, the standard application KBRTestApp was used; an application sending test messages between randomly selected nodes, while adding and removing nodes at specific time intervals. The adding and removing of nodes can be configured with probability parameters.

Tentative testing shows that this implementation of the HyperCircle protocol has a certain performance gain over the OverSim implementations of the Chord and Kademlia protocols, measurable in the time it takes a message to get from sender to recipient. Further testing is outside the scope of this thesis.

Inom hälso- och sjukvården är IT-säkerhet avgörande för att skydda både personlig information och patientsäkerheten. För närvarande genomförs implementering av säkerhetsåtgärder och tester efter mjukvaruutvecklingen, vilket kan minska effektiviteten och utgöra en potentiell risk för patienternas integritet. Detta arbete undersökte implementeringen av DevSecOps-metodiken inom hälso- och sjukvården med fokus på utvecklingsfasen. Genom att intervjua anställda och använda säkerhetsverktyg som SAST, kodgranskning, penetrationstestning och DAST identifierades fördelar och utmaningar. Utmaningarna inkluderade brist på säkerhetskunskap och svårighet att integrera verktyg kostnadsfritt. Trots detta visade resultatet på möjligheten att förbättra säkerheten, effektivisera arbetet och spara pengar genom att använda gratis verktyg och implementera säkerhet redan i utvecklingsfasen. Utbildning och anställning av säkerhetskompetent personal betonades också som viktigt för att upprätthålla höga säkerhetsstandarder

Hälso- och sjukvårdsbranschen har länge varit en sektor som hanterar stora mängder känsliga patientdata och personuppgifter. Integriteten och säkerheten hos patientdata har blivit allt viktigare som en följd av ökad datavolym och digitalisering. Detta examensarbete fokuserade på att utforma och implementera en säker datadelning infrastruktur för att skydda integritet och sekretess för patientdata. Syntetisk data användes för att möjliggöra tillgång för forskare och studenter i reglerade miljöer utan att riskera patienters privatliv. Projektet lyckades genom att utvärdera olika integritetsbevarande mekanismer och skapa en maskininlärningsbaserad applikation för att visa den säkra datadelningsinfrastrukturens funktionalitet. Trots vissa utmaningar visade de valda algoritmerna lovande resultat i fråga om integritetsbevarande och statistisk likhet. Slutligen kan användningen av syntetiska data främja rättvisa beslutsprocesser och bidra till säkra datadelningspraxis inom hälso- och sjukvårdsbranschen.

Nowadays, optical network-on-chip is accepted as a promising alternative solution for traditional electrical interconnects due to lower transmission delay and power consumption as well as considerable high data bandwidth. However, silicon photonics struggles with some particular challenges that threaten the reliability of the data transmission process.The most important challenges can be considered as temperature fluctuation, process variation, aging, crosstalk noise, and insertion loss. Although several attempts have been made to investigate the effect of these issues on the reliability of optical network-on-chip, none of them modeled the reliability of photonic network-on-chip in a system-level approach based on basic element failure rate. In this paper, an analytical model-based simulator, called Reliability Assessment of Photonic Network-on-Chips (RAP-NoC), is proposed to evaluate the reliability of different 2D optical network-on-chip architectures and data traffic. The experimental results show that, in general, Mesh topology is more reliable than Torus considering the same size. Increasing the reliability of Microring Resonator (MR) has a more significant impact on the reliability of an optical router rather than a network.

In today's ever-automating world, businesses are increasingly focused on optimizing network operations, including configuration, management, and troubleshooting. This study scrutinizes the comparison between manual and automated methodologies within the framework of the cloud-based Cisco Meraki platform. Our objective is to establish decisive criteria for identifying the most advantageous circumstances for automation, particularly in network configuration. To achieve this, we constructed a network consisting of three Meraki devices. Automation of network configuration was facilitated through the development of Python scripts leveraging the Meraki API library in conjunction with configuration files. The configuration files were constructed in various file formats–Excel, CSV, JSON, and YAML– in order to assess their impact on automation effectiveness. Moreover, manual configuration tasks were undertaken by three network technicians with varying levels of Meraki proficiency. A comprehensive series of tests, encompassing both automated and manual methods, was subsequently conducted, focusing on metrics such as execution time and keystroke count. Through meticulous analysis, we provided valuable insights into the effectiveness of automation in network configuration processes. Results demonstrated that the use of JSON files in the automation process yielded the shortest completion time and required the fewest keystrokes compared to other automation methods. Moreover, manual configuration via the Meraki dashboard GUI exhibited comparable efficiency to automation methods in networks with a small number of devices. Nonetheless, owing to the capacity to reuse configuration files with minor adjustments for similar setups, the true advantage of automation over manual configuration becomes apparent as the device count increases.

Parkinsons sjukdom (PD) är en progressiv, kronisk neurodegenerativ sjukdom som påverkar patienternas livskvalitet. Handtremor är ett framträdande motoriskt symptom på PD. Dock är nuvarande kliniska tremorbedömningsmetoder tidskrävande och dyra och kanske inte fångar hela omfattningen av tremorfluktuationer. Inbyggda sensorer i smartphones erbjuder ett tillgängligt och kostnadseffektivt alternativ för objektiv tremorbedömning.

Denna studie presenterar en systematisk metod för att utveckla en kvantitativ algoritm för bedömning av Parkinsons sjukdomstremor med hjälp av data från Inertial Measurement Unit (IMU). Denna studie börjar med en omfattande data visualisering och förståelsefas, vilket leder till designbeslutet att implementera en multipel linjär regressionsmodell för förutsägelse av tremorseveritet. IMU-data, insamlad från 10 patienter, förbehandlas och normaliseras för att säkerställa konsekvens och ta hänsyn till varierande grad av tremorseveritet.

Funktionsextraktion genomförs baserat på insikter från litteraturen, vilket resulterar i 16 unika funktioner. Dessa unika funktioner extraheras för var och en av accelerations- och rotationshastighetsdata, vilket resulterar i totalt 582 funktioner över båda händerna och alla tre tremortyper. Rekursiv funktionseleminering med korsvalidering (RFECV) används för funktionsval, vilket identifierar de mest relevanta funktionerna som bidrar till förutsägelse av tremorseveritet. En multipel linjär regressionsmodell implementeras och tränas med hjälp av Leave-One-Out med Cross-Validation (LOOCV) metoden.

Modellens prestanda utvärderas vilket resulterar i ett genomsnittligt MSE på 0,88, ett genomsnittligt MAE på 0,69 och ett R² på 0,88. Resultaten indikerar en stark korrelation mellan förutsagd och faktisk tremorseveritet, vilket tyder på modellens höga validitet. De valda funktionerna visar en hög korrelation med patienternas MDS-UPDRS-poäng, vilket ytterligare validerar deras relevans i förutsägelse av tremorseveritet. Större resultat kunde uppnås, men provstorleken var den största begränsningen under denna studie.

Denna studie visar potentialen att använda IMU-data och multipel linjär regressionsmodellering för noggrann PD-tremorbedömning inom Mobistudy, vilket bidrar till fältet för kvantitativ PD-analys.

The goal of parameterized verification is to prove the correctness of a system specification regardless of the number of its components. The problem is of interest in several different areas: verification of hardware design, multithreaded programs, distributed systems, and communication protocols. The problem is undecidable in general. Solutions for restricted classes of systems and properties have been studied in areas like theorem proving, model checking, automata and logic, process algebra, and constraint solving. In this introduction to the special issue, dedicated to a selection of works from the Parameterized Verification workshop PV '14 and PV '15, we survey some of the works developed in this research area.

We propose a formal model of concurrent systems in which the history of a computation is explicitly represented as a collection of events that provide a view of a sequence of configurations. In our model events generated by transitions become part of the system configurations leading to operational semantics with historical data. This model allows us to formalize what is usually done in symbolic verification algorithms. Indeed, search algorithms often use meta-information, e.g., names of fired transitions, selected processes, etc., to reconstruct (error) traces from symbolic state exploration. The other interesting point of the proposed model is related to a possible new application of the theory of well-structured transition systems (wsts). In our setting wsts theory can be applied to formally extend the class of properties that can be verified using coverability to take into consideration (ordered and unordered) historical data. This can be done by using different types of representation of collections of events and by combining them with wsts by using closure properties of well-quasi orderings.

We study decidability and undecidability results for parameterized verification of a formal model of timed Ad Hoc network protocols. The communication topology is defined by an undirected graph and the behaviour of each node is defined by a timed automaton communicating with its neighbours via broadcast messages. We consider parameterized verification problems formulated in terms of reachability. In particular we are interested in searching for an initial configuration from which an individual node can reach an error state. We study the problem for dense and discrete time and compare the results with those obtained for (fully connected) networks of timed automata.

We compare the expressive power of a class of well-structured transition systems that includes relational automata (extensions of), Petri nets, lossy channel systems, constrained multiset rewriting systems, and data nets. For each one of these models we study the class of languages generated by labeled transition systems describing their semantics. We consider here two types of accepting conditions: coverability and reachability of a fixed a priori configuration. In both cases we obtain a strict hierarchy in which constrained multiset rewriting systems is the most expressive model.

We consider timed Petri nets, i.e., unbounded Petri nets where each token carries a real-valued clock. Transition arcs are labeled with time intervals, which specify constraints on the ages of tokens. Our cost model assigns token storage costs per time unit to places, and firing costs to transitions. We study the cost to reach a given control-state. In general, a cost-optimal run may not exist. However, we show that the infimum of the costs is computable.

Att erhålla data från olika sensorer som finns på olika mobila enheter kan vara användbart inom exempelvis sport och medicin. För att data från de olika sensorerna ska kunna kombineras för analys måste de mobila enheterna tidssynkroniseras först. Denna rapport presenterar en applikation som kan användas för att beräkna skillnaden mellan de interna klockorna på två Android enheter med en kombination av Cristian- och Marzullo -algoritmerna. Olika metoder för att ansluta enheterna via både Wi-Fi och internet testas för att bestämma den optimala metoden för tidssynkronisering. Rapporten validerar också synkroniseringen genom att testa olika sensorer på två identiska Android -smartphones. Resultaten visar att klocksynkronisering mellan två mobila enheter kan uppnås med en round-trip time på 2 millisekunder eller mindre med Wi-Fi Direct. Validering av synkroniseringen visar att en fördröjning på 7 millisekunder eller mindre kan uppnås mellan två sensorer av samma typ på två identiska Android -smartphones. Det visar också att den minst möjliga fördröjningen mellan sensorer av olika typer är 16 millisekunder. Slutsatsen är att när två Android smartphones är tidssynkroniserade kan endast data från sensorer av samma typ kombineras, med undantag för kamerasensorn. Ytterligare tester med mer robust utrustning behövs för att eliminera mänskliga fel vilket kan möjligen ge mer önskvärda resultat.

In this article, the Addis Ababa city road safety policies are examined and analysed based on the Vision Zero approach to road safety work. Three major policy documents are explored and assessed in terms of how they compare with Vision Zero policy in Sweden, concerning how road safety problems are conceptualised, the responsibility ascriptions promoted, the nature of goal setting concerning road safety objectives, and the specific road safety interventions promoted. It is concluded that there is a big difference between the Swedish Vision Zero approach to road safety work and the Addis Ababa road safety approach in terms of how road safety problems are framed and how responsibility ascriptions are made. In Addis Ababa, policy documents primarily frame road safety problems as individual road user problems and, hence, the responsibility for traffic safety is mainly left to the individual road users. The responsibility extended to other system components such as the vehicles, road design, and the operation of the traffic is growing but still very limited. It is argued that in order to find and secure long-term solutions for traffic safety in the city, a paradigm shift is needed, both regarding what are perceived to be the main causes of road safety problems in the city and who should be responsible for ensuring that road fatalities and injuries are prevented.

Teleoperation is increasingly valued for its costeffectiveness and safety benefits in managing workflows acrossindustries like harbor automation, construction, transportationand logistics, mining, and forestry. Ensuring the safe operation ofheavy machinery involves effectively transferring and visualizingthe vehicle’s state and surroundings to remote operators for realtime tasks. Multi-camera video-based environment perceptionand state knowledge are crucial but bandwidth-intensive. Toaddress this, integrating computing functions at the edge isvital. This paper presents ”edge on wheels,” a system designusing the YOLO model for object detection/depth perceptiontasks. We divide the design into edge and cloud components,where offloading decisions on edge and cloud are based onworkload, task priority, and network quality. Finally, we discusschallenges and future enhancements for real-time object detectionin teleoperated driving scenarios.

The accelerated move towards the adoption of the industrial Internet of Things (IIoT) paradigm has resulted in numerous shortcomings as far as security is concerned. One of the IIoT affecting critical security threats is what is termed as the ” False Data Injection” (FDI) attack. The FDI attacks aim to mislead the industrial platforms by falsifying their sensor measurements. FDI attacks have successfully overcome the classical threat detection approaches. In this study, we present a novel method of FDI attack detection using Autoencoders (AEs). We exploit the sensor data correlation in time and space, which in turn can help identify the falsified data. Moreover, the falsified data are cleaned using the denoising AEs. Performance evaluation proves the success of our technique in detecting FDI attacks. It also significantly outperforms a support vector machine (SVM) based approach used for the same purpose. The denoising AE data cleaning algorithm is also shown to be very effective in recovering clean data from corrupted (attacked) data.

Projektets mål har varit att skapa en webbplats med boknings-funktionalitet för Kristinehamnsortens biodlareförening, en förening för biodlare i Kristinehamns kommun med omnejd. Webbplatsen har skapats i PHP, SASS och JavaScript, med en REST-webbtjänst för bokning av biodlingsredskap skriven i PHP och kopplad till MySQL-databas. Via REST-webbtjänsten skapades funktionalitet för att medlemmar i föreningen ska kunna boka redskap samt att styrelsemedlemmar kan logga in i ett administrationsgränssnitt för att se bokningar som har kommit in. Webbplatsen har även designats modulärt där innehåll enkelt kan läggas till i moduler av olika storlekar, för att ska vara smidigt och fungera sömlöst att lägga till, ändra eller ta bort innehåll. Arbetet med projektet utvärderades via teams med ett par medlemmar ifrån föreningen, där mindre justeringar gjordes i form av ändring av färg och storlek på vissa element. Valet att skapa en REST-webbtjänst för ändamålet visade sig vara överflödigt, och valet föll på att skapa inloggningsfunktionalitet med direktanslutning till databas i stället. Det finns även tankar på att på längre sikt konvertera bokningssystemet till en direktanslutning också.

Denna rapport går igenom det arbete som gjorts under projektarbetet som förrättades hos företaget Learn2Esport. Företaget har utvecklat en plattform för utbildning inom e-sport vid namn Gameplan. Projektets syfte har varit att implementera ny design för plattformens sidor som innehåller de produktpaket som erbjuds. Målet är att göra dessa sidor mer överskådliga och användarvänliga, vilket är av vikt för administratörer och företagets effektivitet. Utveckling har gjorts i React med skriptspråket TypeScript. Arbetet har inneburit en omkonstruktion av index-sidan som listar alla paket samt implementering av undersidor för varje paket. För varje undersida har fyra tabbar skapats där olika data relaterat till paketet läses ut under varje tabb.

This thesis investigates architectures for multiplying elements in Galois rings of the size 4^m, where m is an integer.

The main question is whether known architectures for multiplying in Galois fields can be used for Galois rings also, with small modifications, and the answer to that question is that they can.

Different representations for elements in Galois rings are also explored, and the performance of multipliers for the different representations is investigated.

The Internet traffic volume continues to grow at a great rate, now driven by video and TV distribution. For network operators it is important to avoid congestion in the network, and to meet service level agreements with their customers.  This thesis presents work on two methods operators can use to reduce links loads in their networks: traffic engineering and content caching.

This thesis studies access patterns for TV and video and the potential for caching.  The investigation is done both using simulation and by analysis of logs from a large TV-on-Demand system over four months.

The results show that there is a small set of programs that account for a large fraction of the requests and that a comparatively small local cache can be used to significantly reduce the peak link loads during prime time. The investigation also demonstrates how the popularity of programs changes over time and shows that the access pattern in a TV-on-Demand system very much depends on the content type.

For traffic engineering the objective is to avoid congestion in the network and to make better use of available resources by adapting the routing to the current traffic situation. The main challenge for traffic engineering in IP networks is to cope with the dynamics of Internet traffic demands.

This thesis proposes L-balanced routings that route the traffic on the shortest paths possible but make sure that no link is utilised to more than a given level L. L-balanced routing gives efficient routing of traffic and controlled spare capacity to handle unpredictable changes in traffic.  We present an L-balanced routing algorithm and a heuristic search method for finding L-balanced weight settings for the legacy routing protocols OSPF and IS-IS. We show that the search and the resulting weight settings work well in real network scenarios.

Today video and TV distribution dominate Internet traffic and the increasing demand for high-bandwidth multimedia services puts pressure on Internet service providers. In this paper we simulate TV distribution with time-shift and investigate the effect of introducing a local cache close to the viewers. We study what impact TV program popularity, program set size, cache replacement policy and other factors have on the caching efficiency. The simulation results show that introducing a local cache close to the viewers significantly reduces the network load from TV-on-Demand services. By caching 4% of the program volume we can decrease the peak load during prime time by almost 50%. We also show that the TV program type and how program popularity changes over time can have a big influence on cache hit ratios and the resulting link loads

IPTV, where television is distributed over the Internet Protocol in a single operator network, has become popular and widespread. Many telecom and broadband companies have become TV providers and distribute TV channels using multicast over their backbone networks. IPTV also means an evolution to time-shifted television where viewers now often can choose to watch the programs at any time. However, distributing individual TV streams to each viewer requires a lot of bandwidth and is a big challenge for TV operators. In this paper we present an empirical IPTV workload model, simulate IPTV distribution with time-shift, and show that local caching can limit the bandwidth requirements significantly.

Today increasingly large volumes of TV and video are distributed over IP-networks and over the Internet. It is therefore essential for traffic and cache management to understand TV program popularity and access patterns in real networks.

In this paper we study access patterns in a large TV-on-Demand system over four months. We study user behaviour and program popularity and its impact on caching. The demand varies a lot in daily and weekly cycles. There are large peaks in demand, especially on Friday and Saturday evenings, that need to be handled.

We see that the cacheability, the share of requests that are not first-time requests, is very high. Furthermore, there is a small set of programs that account for a large fraction of the requests. We also find that the share of requests for the top most popular programs grows during prime time, and the change rate among them decreases. This is important for caching. The cache hit ratio increases during prime time when the demand is the highest, and aching makes the biggest difference when it matters most.

We also study the popularity (in terms of number of requests and rank) of individual programs and how that changes over time. Also, we see that the type of programs offered determines what the access pattern will look like.

The purpose of this thesis is to document the development of the graphics part of an extremely pluggable game engine/lab environment for a course in advanced game programming. This thesis is one out of five, and concerns indoor, realtime computer 3D graphics. It covers state-of-the-art techniques such as GLSL - the OpenGL Shading Language - and more well known techniques such as portal based rendering.

Kravhantering vid agil mjukvaruutveckling är en utmaning som allt fler företag ställs inför.Prototyper, modeller som liknar tilltänkta slutprodukter, kan användas för att inhämtaviktig information om det som ska utvecklas. För att beskriva hur lik en prototyp är dentilltänkta slutprodukten används begreppet verklighetsfaktor. Studiens syfte är dels attöka kunskapen kring prototypanvändning i agil mjukvaruutveckling, dels att undersökavilken effekt en prototyps verklighetsfaktor har då prototyper används i diskussioner inomkravhantering. En fallstudie görs på företaget Exsitec där personal intervjuas angående prototypanvändning i mjukvaruprojekt. Två prototyper utvecklas sedan med låg respekti-ve hög verklighetsfaktor och används som diskussionsunderlag i intervjuer. Studien visar att användning av prototyper i mjukvaruprojekt kan bidra till ökat förtroende hos kun-der, förbättrad kommunikation med kunder och kan förenkla att uppnå konsensus mellan olika intressenter. Vidare kan de, beroende av hur de används, bidra till helhetsbilden avprodukten och fungera som dokumentation. Studien påvisar även några, om än subtila, skillnader i den information som samlas in med hjälp av prototyper med låg respekti-ve hög verklighetsfaktor. Hög verklighetsfaktor tycks medföra att fler krav samlas in, men göra respondenter mindre benägna att vilja komma med förslag på mer omfattandeförändringar.

A trend seen on the web today is to create a platform where externally developed applications can run inside some kind of main application. This is often done by providing an API to access data and business logic of your service and a sandbox environment in which third-party applications can run. By providing this, it is made possible for external developers to come up with new ideas based on your service. Some good examples on this are Spotify Apps, Apps on Facebook and SalesForce.com.

Ipendo Systems AB is a company that develops a web platform for intellectual properties. Currently most things on this platform are developed by developers at Ipendo Systems AB. Some interest has though risen to enable external developers to create applications that will in some way run inside the main platform.

In this thesis an analysis of already existing solutions has been done. These solutions were Spotify Apps and Apps on Facebook. The two have different approaches on how to enable third-party applications to run inside their own service. Facebook’s solution builds mainly on iframe embedded web pages where data access is provided through a web API. Spotify on the other hand hosts the third-party applications themselves but the applications may only consist of HTML5, CSS3 and JavaScript.

In addition to the analysis a prototype was developed. The purpose of the prototype was to show possible ways to enable third-party applications to run inside your own service. Two solutions showing this were developed. The first one was based on Facebook’s approach with iframing of external web pages. The second was a slightly modified version of Spotify’s solution with only client-side code hosted by the main application. To safely embed the client side code in the main application a sandboxing tool for JavaScript called Caja was used.

Of the two versions implemented in the prototype was the Iframe solution considered more ready to be utilized in a production environment than Caja. Caja could be seen as an interesting technique for the future but might not be ready to use today. The reason behind this conclusion was that Caja decreased the performance of the written JavaScript as well as adding complexity while developing the third-party applications.

Creating native mobile application on multiple platforms generate a lot of duplicate code. This thesis has evaluated if the code quality attribute modifiability improves when migrating to React Native. One Android and one iOS codebase existed for an application and a third codebase was developed with React Native. The measurements of the codebases were based on the SQMMA-model. The metrics for the model were collected with static analyzers created specifically for this project. The results created consists of graphs that show the modifiability for some specific components over time and graphs that show the stability of the platforms. These graphs show that when measuring code metrics on applications over time it is better to do this on a large codebase that has been developed for some time. When calculating a modifiability value the sum of the metrics and the average value of the metrics between files should be used and it is shown that the React Native platform seems to be more stable than native.

Microscope cameras possess high zoom capabilities and are frequently used for quality control in various industries. Visus Technology is a company that manufactures and sells these cameras, along with control software. The problem is that these cameras can only be set to one focus depth at a time. Therefore, we were tasked with implementing a fast focus stacking algorithm that combines multiple input images with different focus depths, resulting in a composite image with consistent focus throughout. This will later be integrated into the control software.

We initially approached the problem by choosing pixels with the highest Laplacian intensity. This resulted in very noisy output images since the blurry images could sometimes produce high-intensity pixels where they were not supposed to be. To address this, we implemented a grid-based focus stacking algorithm. The algorithm divides the input images into tiles, filtering out the blurriest tiles to exclude them from the stacking process. The result is a composite image with significantly reduced blur.

This paper describes the system used by the Machine Learning Group of LTU in subtask 1 of the SemEval-2022 Task 4: Patronizing and Condescending Language (PCL) Detection. Our system consists of finetuning a pretrained text-to-text transfer transformer (T5) and innovatively reducing its out-of-class predictions. The main contributions of this paper are 1) the description of the implementation details of the T5 model we used, 2) analysis of the successes & struggles of the model in this task, and 3) ablation studies beyond the official submission to ascertain the relative importance of data split. Our model achieves an F1 score of 0.5452 on the official test set.

Among the more salient accelerator technologies to continue performance scaling in High-Performance Computing (HPC) are Coarse-Grained Reconfigurable Arrays (CGRAs). However, what benefits CGRAs will bring to HPC workloads and how those benefits will be reaped is an open research question today. In this work, we propose a framework to explore the design space of CGRAs for HPC workloads, which includes a tool flow of compilation and simulation, a CGRA HDL library written in SystemVerilog, and a synthesizable CGRA design as a baseline. Using RTL simulation, we evaluate two well-known computation kernels with the baseline CGRA for multiple different architectural parameters. The simulation results demonstrate both correctness and usefulness of our exploration framework.

The most widely used protocol for providing reliable service and congestion control in the Internet is the Transmission Control Protocol (TCP). When the Internet is moving towards more use in mobile applications it is getting more important to know how TCP works for this purpose.

One of the technologies used for mobile Internet is the Enhanced General Packet Radio Service (EGPRS) extension to the popular GSM system. This thesis presents a low-level analysis of TCP performance in an EGPRS system and an overview of existing TCP, GSM and EGPRS technologies.

The bottleneck in an EGPRS system is the wireless link – the connection between the mobile phone and the GSM base station. The data transfer over the wireless link is mainly managed by the complex RLC/MAC protocol.

In this thesis, simulations were made to identify some problems with running TCP and RLC/MAC together. The simulations were made using existing EGPRS testing software together with a new TCP module. The simulation software is also briefly described in the thesis.

Additionaly, some suggestions are given in order to enhance performance, both by changing the EGPRS system and by modifying the TCP algorithms and parameters.

Projektets mål är en, för kund, optimerad webbplats byggd i Wordpress med tillägget Beaver Builder och Beaver Themer samt en grunduppsättning av valda tillägg för en modern, säker och kostnadseffektiv Wordpressajt. Projektet ska även innefatta en utvärdering och analys av hur funktionalitet och design kan prioriteras för optimal leverans och kundnöjdhet. Syftet är en nöjd kund samt att införskaffa tillräcklig med kunskap för att bygga webbplatser i Beaver Builder med en sund prioritering av funktionalitet och design. Marknadsundersökning har gjorts med frågor om webbplatser inom svensk besöksnäring. Utefter resultatet har designen gjorts för att vara tydlig, enkel och för att, med hjälp av bilder, få rätt känsla. Användartester på både stora skärmar och små skärmar har utförts i form av enkäter. Resultaten visar på en tydlig, snygg och lättnavigerad webbplats.Vemdalen caterings webbplats har publicerats på domänen vemdalencatering.se. Webbplatsen är byggd i Wordpress med valda tillägg. Designen är responsiv och fungerar bra för olika skärmstorlekar samt är enligt användbarhetstester även användarvänlig. Validering har utförts med resultat som följer de utsatta målen. Sökmotoroptimering har utförts. Utvärdering och analys med hjälp av användartester, marknadsundersökning, SE ranking och Google Analytics visar att prioritet för design och upplevelse i många fall kan vara högre än tillgänglighet, beroende på målgrupp, och att mer arbete behövs för att ranka högre i Googles sökmotor och få fler besökare till webbplatsen.

Resource efficient approaches are of great importance for resource constrained embedded systems. In this paper, we present an approach targeting systems where tasks of a critical application are partitioned on a multi-core platform and by using resource reservation techniques, the remaining bandwidth capacity on each core is utilized for one or a set of non-critical application(s). To provide a resource efficient solution and to exploit the potential parallelism of the extra applications on the multi-core processor, global scheduling is used to schedule the tasks of the non-critical applications. Recently a specific instantiation of such a system has been studied where tasks do not share resources other than the processor. In this paper, we enable semaphore-based resource sharing among tasks within critical and non-critical applications using a suspension-based synchronization protocol. Tasks of non-critical applications have partial access to the processor bandwidth. The paper provides the systems schedulability analysis where blocking due to resource sharing is bounded. Further, we perform experimental evaluations under balanced and unbalanced allocation of tasks of a critical application to cores.

The old methods used for images inpainting of the Depth Image Based Rendering (DIBR) process are inefficient in producing high-quality virtual views from captured data. From the viewpoint of the original image, the generated data’s structure seems less distorted in the virtual view obtained by translation but when then the virtual view involves rotation, gaps and missing spaces become visible in the DIBR generated data. The typical approaches for filling the disocclusion tend to be slow, inefficient, and inaccurate. In this project, a modern technique Generative Adversarial Network (GAN) is used to fill the disocclusion. GAN consists of two or more neural networks that compete against each other and get trained. This study result shows that GAN can inpaint the disocclusion with a consistency of the structure. Additionally, another method (Filling) is used to enhance the quality of GAN and DIBR images. The statistical evaluation of results shows that GAN and filling method enhance the quality of DIBR images.

This paper studies the utility of using data analytics and machine learning techniques for identifying, classifying, and characterizing the dynamics of large-scale parallel (MPI) programs. To this end, we run microbenchmarks and realistic proxy applications with the regular compute-communicate structure on two different supercomputing platforms and choose the per-process performance and MPI time per time step as relevant observables. Using principal component analysis, clustering techniques, correlation functions, and a new “phase space plot,” we show how desynchronization patterns (or lack thereof) can be readily identified from a data set that is much smaller than a full MPI trace. Our methods also lead the way towards a more general classification of parallel program dynamics.