For this thesis the algorithm of a Particle Filter has been partly implemented on a Zynq All-programmable SoC allowing for speed ups of up to 12 times in parts of the code. The implementation is used to track a robot in an environment known by a 2D map. Since some parts have not been implemented in hardware lots of communication is done between the hardware and software parts of the code which creates a performance bottleneck.

Cross-domain intrusion detection, a critical component of cybersecurity, involves evaluating the performance of neural networks across diverse datasets or databases. The ability of intrusion detection systems to effectively adapt to new threats and data sources is paramount for safeguarding networks and sensitive information. This research delves into the intricate world of cross-domain intrusion detection, where neural networks must demonstrate their versatility and adaptability. The results of our experiments expose a significant challenge: the phenomenon known as catastrophic forgetting. This is the tendency of neural networks to forget previously acquired knowledge when exposed to new information. In the context of intrusion detection, it means that as models are sequentially trained on different intrusion detection datasets, their performance on earlier datasets degrades drastically. This degradation poses a substantial threat to the reliability of intrusion detection systems. In response to this challenge, this research investigates potential solutions to mitigate the effects of catastrophic forgetting. We propose the application of continual learning techniques as a means to address this problem. Specifically, we explore the Elastic Weight Consolidation (EWC) algorithm as an example of preserving previously learned knowledge while allowing the model to adapt to new intrusion detection tasks. By examining the performance of neural networks on various intrusion detection datasets, we aim to shed light on the practical implications of catastrophic forgetting and the potential benefits of adopting EWC as a memory-preserving technique. This research underscores the importance of addressing catastrophic forgetting in cross-domain intrusion detection systems. It provides a stepping stone for future endeavours in enhancing multi-task learning and adaptability within the critical domain of intrusion detection, ultimately contributing to the ongoing efforts to fortify cybersecurity defences.

The reliable internet stream transport (RIST) protocol is used for streaming video over the internet and requires time synchronisation to ensure that each frame is played out at the correct time. One method to ensure time synchronisation today in RIST productions is by utilizing the network time protocol (NTP). NTP has several issues and requirements that makes it a not ideal time synchronisation method in RIST productions and this thesis proposes a dynamic average time synchronisation (DATS) method as an option for using NTP in RIST productions. The DATS method uses a two-way synchronisation scheme to estimate the time offset between a video sender node and a video receiver node which is added to an average used as the time offset value. Additionally, it is explored whether or not a Kalman filter can further increase performance in DATS. Furthermore, the Kalman filter parameters were attempted to be optimized with a genetic algorithm. With a simulated \textit{testsrc FFmpeg} stream, the performance of DATS is evaluated and compared with the NTP implementation \textit{Chrony}. The metrics used in the evaluation was the playout delay and the error in time synchronisation which were calculated by periodical messages between a sender and receiver node. The results reveal that DATS performs slightly better without a Kalman filter but is slightly less robust without it. The results also indicated that DATS is comparable to NTP in an unloaded network, but is outperformed by NTP in a congested network.

Captioning has long been used in media to help D/deaf and hard-of-hearing persons. Captioning however is difficult and time-consuming manual work. With the rapid evolution of automated speech recognition (ASR) systems, live captioning of everyday speech will soon be a practical reality. A proof of concept Android application for use with a specific headset has been created using the built-in Android SpeechRecognizer, a free open-source API (application programming interface) available for Android phones. This application unlike many existing solutions focuses on two major features, communication with in-situ microphones and hardware via bluetooth and long-duration speech recognition. Long-duration speech recognition was made possible using the segmented session mode of the SpeechRecognizer which was recently added in API version 33 (March2023). The results while not complete show promise for future development. Some initial testing shows a word error rate (WER) of 8% but further testing is required. Tests with noise also show that the system is surprisingly resistant to static noise. The application shows promise and development will continue in the coming weeks. This project was financed by Hörselforskningsfonden in project FA21-0017 and was performed under the supervision of Amin Saremi. / Undertexter har länge använts i media som hjälp för hörselskadade och döva personer. Skrivandet av dessa undertexter kräver tyvärr mycket svårt och manuellt arbete. Den snabba utvecklingen av automatiska taligenkänningssystem (ASR) pekar dock på en framtid där undertexter kan genereras i realtid för vardagliga situationer. En prototyp av en Androidapplikation har skapats för en särskild hörlur. Applikationen använder Androids inbyggda SpeechRecognizer-system, ett gratis API (applikationsprogrammeringsgränssnitt) med öppen källkod tillgängligt för Android-telefoner. Till skillnad från många existerande lösningar så fokuserar denna applikation på två specifika huvudområden: in-situ mikrofoner och hårdvara via bluetooth samt långvarig röstigenkänning. Långvarig röstigenkänning möjliggjordes segmenterade sessioner is SpeechRecognizer-systemet som lades till i API version 33 (mars 2023). Resultaten är inte kompletta me när lovande för vidare utveckling. Enkel preliminär testning visar på ordfelsfrekvens (WER) på ungefär 8% men vidare testning behövs i framtiden. Ytterligare test med bakgrundsljud visar även att systemet är förvånandsvärt resistant mot statiskt brus. Applikationen är lovande och kommer att fortsätta utvecklas under de kommande veckorna. Detta projekt finansierades av Hörselforskningsfonden i projekt FA21-0017 och utfördes under uppsikt av Amin Saremi.

This thesis explores possible improvements, using parallel computing, to the PSF-alignment and image subtraction algorithm found in HOTPANTS. In time-domain astronomy the PSF-alignment and image subtraction algorithm OIS is used to identify transient events. hotpants is a software package based on OIS, the software package ISIS, and other subsequent research done to improve OIS. A parallel GPU implementation of the algorithm from HOTPANTS – henceforth known as BACH –was created for this thesis. The goal of this thesis is to answer the questions: “what parts of HOTPANTS are most suited for parallelisation?” and “how does bach perform compared to HOTPANTS and SFFT?”, another PSF-alignment and image subtraction tool. The authors found that the parts most susceptible to parallelisation were the convolution and subtraction steps. However, the subtraction did not display a significant improvement to its sequential counterpart. The other parts of HOTPANTS were deemed too complex to implement in parallel on the GPU. However, some parts could probably either be partly parallelised on the GPU or parallelised usingthe CPU. BACH was always as fast as or faster than HOTPANTS; it was generally 2 times faster, but was up to 4.5 times faster in some test cases. It was also faster than SFFT, but this result was not equivalent to the result presented in [15], which is why the authors of this thesis believe something was wrong with either the installation of SFFT or the hardware used to test it.

Point clouds are a widely used system for the collection and application of 3D data. But most timesthe data gathered is too scarce to reliably be used in any application. Therefore this thesis presentsa GAN based upsampling method within a patch based approach together with a GCN based featureextractor, in an attempt to enhance the density and reliability of point cloud data. Our approachis rigorously compared with existing methods to compare the performance. The thesis also makescorrelations between input sizes and how the quality of the inputs affects the upsampled result. TheGAN is also applied to real-world data to assess the viability of its current state, and to test how it isaffected by the interference that occurs in an unsupervised scenario.

Video analysis and visualization are widely used in various applications, including sports analysis, video surveillance and medical imaging. This study investigates the use of video visualization as a tool in the field of shooting. The study aims to answer the question: "What can be a good visualization strategy for video-based shooting analysis?". To answer this question, a software, using video visualization to analyze shooting videos, was created. The visualized videos were then tested by both inexperienced individuals and shooting experts. The implementation steps of the project consisted of recording shooting videos followed by implementing the visualizations. Object detection was used to track the shooting target and extract data. Displayed on the visualized videos were colored markings to show the locations of the aimpoint of the gun and the target. The data extracted was used to show graphs of relevant shooting metrics. The user tests focused on collecting quantitative and qualitative data from the users. The testers reacted positively to the use of visualized videos as a tool for analysing shooting performance. The analysis of the testers responses suggests that video visualization is an effective tool for analyzing shooting videos and holds great promise for future research in the field of shooting. The program implemented in the study has good potential for being used as a tool for improving shooting strategies.

Immersion is a subject studied in various areas and especially in games, since a game that immerses a player, counts as a game of great value. Many studies have been performed to understand how to measure immersion through various forms of questionnaires, but how can one with the help of gamification and loud reading make reading more immersive? To see if gamification and loud reading can make reading more immersive an application is developed, the application provides a paragraph, and when read loud correctly creates sound and shows images. Then the application is tested by a group of people to see if gamification can be used to bring life to reading and further immerse the reader in the world of books.

Endpoint detection and response is an integral part of the security of large-scale networks. Embedded hardware, such as those found at Ericsson Radio Access Network endpoints, have strict performance requirements that need to be met. This fact makes implementing intrusion detection non-trivial, as intrusion detection software often generate a lot of processing overhead. Wazuh, an established open-source distributed and centralized intrusion detection and response system, shows a lot of promise as a large-scale intrusion detection system. It is very modular and has various capabilities that can be utilized in different ways to minimize processing overhead. One of these capabilities is native support for the native Linux syscall monitoring tool AuditD. While AuditD is very capable, it can introduce severe performance penalties in certain scenarios. Falco is another syscall monitoring tool that shows promise with regards to performance, and also has more features than AuditD; which is why Falco is included as a direct comparison to AuditD. This study evaluates Wazuh, AuditD, and Falco based on a set of requirements set by Ericsson, including flexibility, scalability and reliability, by enacting performance benchmarks with normal background operations active. The results of this study show that, with the correct configuration, Wazuh can be used as an intrusion detection system in embedded systems with limited hardware, where AuditD and Falco can serve as a great addition to detecting indicators of compromise. The solution is to use a minimal intrusion detection ruleset, and in the event of suspicious activity, activate more modules to increase threat detection at the cost of CPU overhead and execution time for normal system operation.

The purpose of this project was to build an accessible website for Inclusive Web Solutions, an IT consulting company, The website consists of a public part, which presents the company, and a private admin site where employees can create, edit, update and delete the dynamic content of the site. Subscribers may only create, edit, update and delete their own blog posts. A REST API is used as a data source. All dynamic data is stored in a database. The website, the REST API and the database have been implemented using React, Node.js, Express.js, MongoDB and SASS. Web design, flowcharts and a target group analysis have been part of the planning phase. The website has been tested with regard to accessibility, responsiveness, compatibility and code validity using automated testing tools and user testing. The website was supposed to be accessible according to WCAG level AA. The website, the REST API and the database have been deployed. A custom domain has been added to the website. Överall, the purpose of the project has been achieved. There are a few issues with regard to the website’s responsiveness and accessibility. There are also a few usability issues related to the site’s multilingualism. The accessibility goal has not been fully achieved. Validation errors pertaining to the website’s HTML code can be explained by the validation method that was used. / Syftet med detta projekt har varit att skapa en tillgänglig webbplats a t konsultföretaget Inclusive Web Solutions. Webbplatsen består av en publik, tvåspråkig del som presenterar företaget, och en lösenordsskyddad administrationssida da r medarbetare kan skapa, redigera, uppdatera och radera webbplatsens dynamiska innehåll. Gästskribenter kan endast skapa, redigera, uppdatera och radera egna blogginlägg. Webbplatsen använder en REST-webbtjänst som datakälla och alla data som hanteras på den icke-publika webbplatsen lagras i en databas. React, Node.js, Express.js, MongoDB och SASS har använts för att implementera webbplatsen, REST-webbtjänsten och databasen. Arbetet har planerats med hjälp av webbdesign, målgruppsanalyser och flo desscheman. Webbplatsens tillgänglighet, responsivitet, kompatibilitet och validitet har testats med automatiska testverktyg och användartester. När det gäller tillgängligheten, sa har målet varit att uppfylla kraven på WCAG-nivån AA. Webbplatsen, REST-webbtjänsten och databasen har publicerats. Webbplatsen har dessutom publicerats under egen domän. Överlag har syftet med projektmålet uppnåtts. Alla delar har skapats och det mesta av funktionaliteten har implementerats enligt plan. Webbplatsen uppvisar vissa brister när det gäller responsivitet och tillgänglighet. Vissa brister i användbarheten finns, vilket rör bl.a. flerspråkigheten. Tillgänglighetsmålet har inte uppnåtts fullt ut. De valideringsfel som förekommer i HTML-koden har visat sig bero på hur valideringen har gjorts.