Industry 4.0 and the increasing use of artificial intelligence and machine learning have allowed the analysis of large amounts of data and improve performance across many businesses and sectors. These sectors have significantly increased their reliance on data when making decisions. This paper examines the use of data-driven decision-making on road maintenance planning in Sweden. Data related to road maintenance following the Swedish Road Maintenance Standard, were collected focusing on the International Roughness Index (IRI) and rut depth as primary features. Analyzing such data enabled the identification of maintenance needs within three separate timeframes: immediate, the next five years, and long-term. The model predicted maintenance needs based on the IRI with up to 96% accuracy. However, the model's accuracy dropped to only 67% when predicting maintenance needs over the next five years. In contrast, the model that predicted maintenance needs based on rut depth demonstrated high accuracy across all three timeframes, achieving up to 92% accuracy. The model demonstrated that modern road condition variables are crucial to prediction. In terms of predictions, 2023 IRI measurements were the most important. Based on our findings, this paper improves data-driven decision-making in Swedish road maintenance, resulting in more effective resource allocation and decreased emergency maintenance expenses. Moreover, the study highlights the value of collecting and utilizing more accurate and thorough road state data to enhance these models.

The digital transformation of the 21st century has led to a series of new possibilities and challenges, where one major concern of many major organizations and enterprises is promoting Information Security Awareness and Training (ISAT) for their employees. This aspect of Information Security (IS) can promote cybersecurity in the work environment against threats related to the human factor. Apart from traditional methods as workshops and seminars, researchers study the effect of gamification on ISAT, by proposing customized digital games to train employees regardless their IT skills. This thesis is trying to propose what techniques and approaches can be considered to train people throughout a full threat progression by studying the features of previous efforts. For this purpose, a literature study based on the principles of a systematic literature review (SLR) is essential to gather the available data and review their characteristics. More specifically, the solutions of the researchers are analyzed against the seven steps of the Lockheed Martin Cyber Kill Chain (LM CKC), where each game is classified to one or more phases, according to the training they offer. Thus, some tools can provide a wide range of training, covering many aspects of the CKC, while others are targeting a specific IS topic. The results also suggest that popular attacks involving social engineering, phishing, password and anti-malware software are addressed by many games, mainly in the early stages of the CKC and are focus on trainees without professional IT background. On the other hand, in the last two phases of the CKC, the majority of categorized games involves countermeasures that IS specialists must launch to prevent the security breach. Therefore, this study offers insight on the characteristics of serious games, which can influence an ISAT program, tailored to the enterprise’s distinct IS issue(s) and the IT background of the trainees.

Optical Character Recognition (OCR) of cursive scripts like Pashto and Urdu is difficult due the presence of complex ligatures and connected writing styles. In this paper, we evaluate and compare different approaches for the recognition of such complex ligatures. The approaches include Hidden Markov Model (HMM), Long Short Term Memory (LSTM) network and Scale Invariant Feature Transform (SIFT). Current state of the art in cursive script assumes constant scale without any rotation, while real world data contain rotation and scale variations. This research aims to evaluate the performance of sequence classifiers like HMM and LSTM and compare their performance with descriptor based classifier like SIFT. In addition, we also assess the performance of these methods against the scale and rotation variations in cursive script ligatures. Moreover, we introduce a database of 480,000 images containing 1000 unique ligatures or sub-words of Pashto. In this database, each ligature has 40 scale and 12 rotation variations. The evaluation results show a significantly improved performance of LSTM over HMM and traditional feature extraction technique such as SIFT. Keywords.

Atomic segmentation of cursive scripts into con- stituent characters is one of the most challenging problems in pattern recognition. To avoid segmentation in cursive script, concrete shapes are considered as recognizable units. Therefore, the objective of this work is to find out the alternate recognizable units in Pashto cursive script. These alternatives are ligatures and primary ligatures. However, we need sound statistical analysis to find the appropriate numbers of ligatures and primary ligatures in Pashto script. In this work, a corpus of 2, 313, 736 Pashto words are extracted from a large scale diversified web sources, and total of 19, 268 unique ligatures have been identified in Pashto cursive script. Analysis shows that only 7000 ligatures represent 91% portion of overall corpus of the Pashto unique words. Similarly, about 7, 681 primary ligatures are also identified which represent the basic shapes of all the ligatures.

In recent years, the emerging trends in cloud computing technologies have given rise to different computing services through the Internet. Organizations across the globe have seized this opportunity as a critical business driver for computing resource access and utilities that will indeed support significant business operations. Embracing SaaS as a crucial business factor enhances corporate business strategy through economies of scale, easy manageability, cost-effectiveness, non-geographical dependence, high reliability, flexible resources, and fast innovation. However, this has also come with various risks due to the limitation of traditional user identity and access control solutions’ inability to effectively identify and manage cloud users’ authorization process when interacting with the cloud. The limit can result in a legitimate user account's impersonation to carry out malicious activities after the user account is compromised to go undetected since traditional solutions seldom function based on user behavior trust level behind any account.

Furthermore, the limitation is a significant vulnerability to the cloud environment. This vulnerability is known to be exploited by threats that can eventually lead to substantial unacceptable risks that can undermine security principles or requirements such as confidentiality, integrity, and availability. Significant consequences of this risk are categorized into financial damages, legal implications, reputational damages, and regulatory implications to the cloud environment. As a result, a solution that could contribute to the remediation of these potential risks incurred due to the limitation of user identity and access control management was proposed and designed as User Behavior Trust-Based Adaptive Security framework. The design aims to enhance how cloud users' identity and access control might be managed effectively based on a user behavior trust context and adaptation of corresponding access control measures through adaptive security. The design capability was manifested by integrating it into the standard ISO/2705:2018 Risk Management process. Although, there have been several good information security frameworks such as ISO/IEC 27005:2018 and other technical countermeasures such as SaaS Identity & Access Management (IDaaS) to deal with this risk on the public cloud services. However, they are based on static mitigation approaches, so there is a solid need to shift towards a more dynamic strategical approach.

The presented design work, User Behavior Trust-Based Adaptive Security framework, intends to serve as a proposed guideline for risk mitigation that would enhance user identity and access control limitations across the cloud. The solution functions by a trust modeling process that evaluates cloud user activities to compute a user behavior comprehensive trust degree. The resulting data is further used as input feeds parameters into a policy decision point process. The policy decision point process adapts the input parameters to user behavior trust level and behavior risk rating to determine the appropriate access control decision. Ultimately, the adaptive security solution consults the policy decision points to dynamically enforce the corresponding controls measures based on the access control decision received as input feed. The report also conducts a risk assessment process to identify vulnerabilities, threats, and risks related to user behavior trust level and risk rating regarding SaaS resources. Then adapt the mitigation solution, User Behavior Trust-Based Adaptive Security framework, as a possible risk treatment within the risk management process ISO/2705:2018.

This report uses a design methodology derived from User Behavior Trust Modelling scientific research work, Gartner Adaptive Security Architecture Model, and eXtensible Access Control Markup Language's policy decision point concept. The design evaluates user behavior trust level by the trust modeling, while the integrated policy decision point processes the trust level to make the access control decision which is later enforced by the adaptive security solution. The report further adapts the risk management procedure ISO/2705:2018 to identify risk from user behavior and trust level, then implements the design solution as a possible risk treatment. The research findings were documented as Results and Discussion, where the functional and operational aspects of the designed framework were provided. In addition, the effects of applying the framework as a possible risk treatment solution were observed through conducting an ISO/2705:2018 risk management procedure. The notable outcome of a reduction of identified risk levels was an improvement in user attitude or behavior, which eventually increased user behavior trust level and reduced associated behavior risk. At the same time, the discussion detailed the interpretation of the results, implications, and limitation of the research, why the framework could be considered a remediation solution beyond the state-of-the-art for cloud user identity and access management—precisely by integrating user behavior, trust, policy decision making with adaptive security into risk management process to reduce IDM-associated risk in the SaaS.

Finally, this study has outlined the significance of adopting the designed framework as a possible mitigation solution to enhance the shortcomings of user identity and access control management in the cloud. It has demonstrated that SaaS identified risk can be reduced to an acceptable level when user behavior and activities are taken seriously. Insight into the current trust state and associated risk level of cloud users are vital for continuous risk monitoring and reduction. The solution is to be used as a recommended guideline that might significantly contribute to the research community and information security field of cloud security. Future research direction to consider the possibility of simulating and transforming this conceptual and abstract framework into a real-world working solution due to research work limitations. The framework was designed based on recognized and accepted scientific and technological principles and concepts, from user behavior trust modeling, eXtensible access control markup language, and adaptive security architecture. In addition, to extend this concept to a future research area that will focus exclusively on application-processes behavior.

In recent years, mobile devices have become commonly used not only for voice communications but also to play a major role in our daily activities. Accordingly, the number of mobile users and the number of mobile applications (apps) have increased exponentially. With a wide user base exceeding 2 billion users, Android is the most popular operating system worldwide, which makes it a frequent target for malicious actors. Adware is a form of malware that downloads and displays unwanted advertisements, which are often offensive and always unsolicited. This paper presents a machine learning-based system (AdStop) that detects Android adware by examining the features in the flow of network traffic. The design goals of AdStop are high accuracy, high speed, and good generalizability beyond the training dataset. A feature reduction stage was implemented to increase the accuracy of Adware detection and reduce the time overhead. The number of relevant features used in training was reduced from 79 to 13 to improve the efficiency and simplify the deployment of AdStop. In experiments, the tool had an accuracy of 98.02% with a false positive rate of 2% and a false negative rate of 1.9%. The time overhead was 5.54 s for training and 9.36 µs for a single instance in the testing phase. In tests, AdStop outperformed other methods described in the literature. It is an accurate and lightweight tool for detecting mobile adware.

The emergence of unmanned aerial vehicles (also referred to as drones) has transformed the digital landscape of surveillance and supply chain logistics, especially in terrains where such was previously deemed unattainable. Moreover, the adoption of drones has further led to the proliferation of diverse drone types and drone-related criminality, which has introduced a myriad of security and forensics-related concerns. As a step towards understanding the state-of-the-art research into these challenges and potential approaches to mitigation, this study provides a detailed review of existing digital forensic models using the Design Science Research method. The outcome of this study generated in-depth knowledge of the research challenges and opportunities through which an effective investigation can be carried out on drone-related incidents. Furthermore, a potential generic investigation model has been proposed. The findings presented in this study are essentially relevant to forensic researchers and practitioners towards a guided methodology for drone-related event investigation. Ultimately, it is important to mention that this study presents a background for the development of international standardization for drone forensics.

In a world were data / network infrastructures grows by the day, it is not unfamiliar for tools that streamline the work and provides a great scalability is put into production. In this paper I will describe one of these solutions called ACI (Application Centric The infrastructure). Cisco ACI is possible thanks to a combination of different components that interact with dedicated hard / software, in this paper the fundamentals behind these different components will be described what ACI is and how it works. This will be compared to a traditional data solution implemented by myself. For four weeks, I worked at Axians IT in Solna researching automation / streamlining solutions and what pros / cons they have on real IT companies and data centers. During these four weeks I worked parallel to my research conducting an implementation of a data solution to the company EEHunddagis to give them their own production environment and then compare my hand configured solution with an automation solution and what pros / cons they both bring with them Automation is something that has risen up in the later years and is still on the rise. I will therefore in this paper explain why these types of solutions are necessary and who could use them in their own network environment. 

Decentralization is essential when trust and performance must not depend on a single organization. Distributed Ledger Technologies (DLTs) and Decentralized Hash Tables (DHTs) are examples where the DLT is useful for transactional events, and the DHT is useful for large-scale data storage. The combination of these two technologies can meet many challenges. The blockchain is a DLT with immutable history protected by cryptographic signatures in data blocks. Identification is an essential issue traditionally provided by centralized trust anchors. Self-sovereign identities (SSIs) are proposed decentralized models where users can control and manage their identities with the help of DHT. However, slowness is a challenge among decentralized identification systems because of many connections and requests among participants. In this article, we focus on decentralized identification by DLT and DHT, where users can control their information and store biometrics. We survey some existing alternatives and address the performance challenge by comparing different decentralized identification technologies based on execution time and throughput. We show that the DHT and machine learning model (BioIPFS) performs better than other solutions such as uPort, ShoCard, and BBID.

Decentralized identification is an interesting topic for Internet-based systems. Although the use of centralized systems for identification is prevalent, there is still a need for decentralized identification systems that do not rely on a centralized entity. We propose using Distributed Ledger Technology (DLT) to fulfill the needs. Although DLT and Distributed Hash Table (DHT) can be used as technologies in Decentralized Identifiers (DIDs), immutability and the growth of transactions are still challenging. This paper investigates DIDs under the W3C standard and DLT-based technology as one of the solutions for verifiable data registries in DIDs. We evaluate our proposed solution by comparing three models wrt. query time for verifiable data registries. Our proposed architecture for data registry achieves better query time by concealing old transactions from the main DLT.

Video conferencing applications help people communicate via the Internet and provide a significant and consistent basis for virtual meetings. However, integrity, security, identification, and authentication problems are still universal. Current video conference technologies typically rely on cloud systems to provide a stable and secure basis for executing tasks and processes. At the same time, video conferencing applications are being migrated from centralized to decentralized solutions for better performance without the need for third-party interactions. This article demonstrates a decentralized smart identification scheme for video conferencing applications based on biometric technology, machine learning, and a decentralized hash table combined with blockchain technology. We store users' information on a distributed hash table and transactional events on the distributed ledger after identifying users by implementing machine learning functions. Furthermore, we leverage distributed ledger technology's immutability and traceability properties and distributed hash table unlimited storage feature to improve the system's storage capacity and immutability by evaluating three possible architectures. The experimental results show that an architecture based on blockchain and distributed hash table has better efficiency but needs a longer time to execute than the two other architectures using a centralized database.

Purpose - The aim of this paper is to investigate corporate financial disclosure via Twitter among the top listed 350 companies in the UK as well as identify the determinants of the extent of social media usage to disclose financial information.

Design/methodology/approach – This study applies an unsupervised machine learning technique, namely, Latent Dirichlet Allocation (LDA) topic modeling to identify financial disclosure tweets. Panel, Logistic, and Generalized Linear Model Regressions are also run to identify the determinants of financial disclosure on Twitter focusing mainly on board characteristics.

Findings – Topic modeling results reveal that companies mainly tweet about 12 topics, including financial disclosure, which has a probability of occurrence of about 7 percent. Several board characteristics are found to be associated with the extent of Twitter usage as a financial disclosure platform, among which are board independence, gender diversity, and board tenure.

Originality/value – Extensive literature examines disclosure via traditional media and its determinants, yet this paper extends the literature by investigating the relatively new disclosure channel of social media. This study is among the first to utilize machine learning, instead of manual coding techniques, to automatically unveil the tweets’ topics and reveal financial disclosure tweets. It is also among the first to investigate the relationships between several board characteristics and financial disclosure on Twitter; providing a distinction between the roles of executive versus non-executive directors relating to disclosure decisions.

Ongoing advancements in technology have changed dramatically the disclosure media that companies adopt. Such disclosure media have evolved from the traditional paper-based ones, to the internet as the new platform to disclose information via companies’ designated websites. However, currently the new media for disclosures are the social media. The aim of this paper is to investigate corporate social media accounts for financial disclosure, as well as, identify its determinants. The sample of the study is comprised of the tweets posted on the Twitter accounts belonging to the FTSE 350 constituents. Topic modeling is applied to identify financial disclosure tweets and logistic regression is run to identify the determinants of financial disclosure on Twitter. Results show that companies use Twitter to make corporate disclosures and some board characteristics are found to have a significant relationship with financial disclosure.

Ongoing advancements in technology have changed dramatically the disclosure media that companies adopt. Such disclosure media have evolved from the traditional paper-based ones, to the internet as the new platform to disclose information via companies’ designated websites. However, currently the new media for disclosures are the social media. The aim of this paper is to investigate corporate social media accounts for financial disclosure, as well as, identify its determinants. The sample of the study is comprised of the tweets posted on the Twitter accounts belonging to the FTSE 350 constituents. Topic modeling is applied to identify financial disclosure tweets and logistic regression is run to identify the determinants of financial disclosure on Twitter. Results show that companies use Twitter to make corporate disclosures and some board characteristics are found to have a significant relationship with financial disclosure. 

This thesis demonstrates a technique for developing efficient applications interpreting spacial deep learning output using Hyper Dimensional Computing (HDC), also known as Vector Symbolic Architecture (VSA). As a part of the application demonstration, a novel preprocessing technique for motion using state machines and spacial semantic pointers will be explained. The application will be evaluated and run on a Google Coral edge TPU interpreting real time inference of a compressed object detection model.

When developing applications for iOS devices it is very common to use Apple’s own Core Data database system which is a framework handling database persistence for iOS devices among other things, but since there so many different kinds of applications it might not be the best option to use the same database system every time. Realm is another database system for iOS devices, it is very lightweight and a big rival to Core Data. This work was conducted with the goal of finding differences between the database systems Core Data and Realm that might show that one or the other is better used in some cases. The comparison between the systems was divided into two different parts, one theoretical comparison focused on reading and analyzing documentation and development of a test application. The test application tested time of create, read, update and delete operations in relation to increasing number of objects and increased number of properties in each object. The tests on Core Data were made with two different implementations to get the aspect of time difference based on implementation included. The results were fairly similar on the different operations with a slight advantage to Core Data. The big difference was seen in implementation difficulty and usability. The included features in the database system were also considered. Realm included more of commonly used and important features but Core Data gives the user the ability to add most of these in the implementation, this results in a question of user case.

Data Ductus is a Swedish IT-consultant company, their customer base ranging from small startups to large scale cooperations. The company has steadily grown since the 80s and has established offices in both Sweden and the US.

With the help of machine learning, this project will present a possible solution to the errors caused by the human factor in the logistic business.A way of preprocessing data before applying it to a machine learning algorithm, as well as a couple of algorithms to use will be presented.

This study examines the design and trustworthiness factors of recommendation systems forexperience goods in the e-commerce industry. Experience goods are products that involvesensory experiences and pose challenges for consumers to assess and select online. Theresearch adopts a mixed method approach, combining exploratory and interpretive researchmethods to gain insights into users' interpretations and meanings attached to their experiences.The methodology includes analyzing publications, conducting a survey, and objectivelydocumenting recommendation systems in the alpine industry. The survey collects opinions fromparticipants who have used various recommendation systems, covering aspects such as usermodel, item model, recommendation algorithm, user interface, evaluation, and trustworthiness.A thematic analysis is employed to identify patterns and meaningful themes in the data. Thefindings emphasize the importance of understanding user preferences, balancingrecommendations, improving accuracy, enhancing interface usability, incorporating feedback,and addressing recommendation diversity to enhance trustworthiness. A hybrid filteringapproach with feature-based systems and integrated behavior-based techniques is identified aseffective. While the survey's convenience sampling and limited sample size may limitgeneralizability, the findings provide insights for designing effective recommendation systems forexperience goods in e-commerce. By considering the strengths and limitations of differenttechniques, vendors can create systems that assist customers in purchasing these uniqueproducts. However, recommendation systems should be viewed as a valuable tool rather thanthe sole determinant in purchase decisions for alpine equipment. Further research with a largerand more diverse sample is recommended to validate the findings and improve generalizability.

With so many developers writing code, so many choose to become a developer every day, using tools to aid in the work process is needed. With all the testing being done for multiple different devices and sources there is a need to make it better and more efficient. In this thesis connecting the variety of different tools such as version control, project management, issue tracking and test systems is explored as a possible solution. A possible solution was implemented and then analyzed through a questionnaire that were answered by developers. For an example results as high as 75\% answering 5 if they liked the connection between the issue tracking system and the test results. 75\% also gave a 5 when asked about if they liked the way the test results were presented. The answers they gave about the implementation made it possible to conclude that it is possible to achieve a solution that can solve some of the presented problems. A better way to connect various tools to present and analyze the test results coming from multiple different sources.

As the vision of the Internet of Things (IoT) becomes a reality, thousands of devices will beconnected to IoT platforms in smart cities and regions. These devices will actively send dataupdates to cloud-based platforms, as part of smart applications in domains like healthcare, trafficand pollution monitoring. Therefore, it is important to study the ability of modern IoT systemsto handle high rates of data updates coming from devices. In this work we evaluated the per-formance of components of the Internet of Things Services Enablement Architecture of theEuropean initiative FIWARE. We developed a testbed that is able to inject data updates usingMQTT and the CoAP-based Lightweight M2M protocols, simulating large scale IoT deploy-ments. Our extensive tests considered the vertical and horizontal scalability of the componentsof the platform. Our results found the limits of the components when handling the load, and thescaling strategies that should be targeted by implementers. We found that vertical scaling is notan effective strategy in comparison to the gains achieved by horizontally scaling the databaselayer. We reflect about the load testing methodology for IoT systems, the scalability needs ofdifferent layers and conclude with future challenges in this topic.

An L2 cache is a device that buffers data in fast memory closer to the Central Processing Unit(CPU) in order to deliver its contents with much lower latency than can otherwise be achieved bymain memory. This provides a substantial performance increase in many systems as the memoryinterface is often a bottleneck. The goal of this thesis is to develop a simple L2 cache usingVHDL for Cobham Gaisler’s open source hardware library GRLIB which currently lacks such acore. The outcome of the thesis is the IP core L2C-Lite which will be released in Febuary of 2022 as an addition to GRLIB. L2C-Lite has been integrated into multiple systems and has providedmajor performance gains in applications running under linux as well as other benchmarks. Inaddition, some potential improvements have been identified to further increase the performanceof the cache, as well as improve its usability in systems.

This book is a relevant reference for any readers interested in the security aspects of Cyber-Physical Systems and particularly useful for those looking to keep informed on the latest advances in this dynamic area.

Cyber-Physical Systems (CPSs) are characterized by the intrinsic combination of software and physical components. Inherent elements often include wired or wireless data communication, sensor devices, real-time operation and automated control of physical elements. Typical examples of associated application areas include industrial control systems, smart grids, autonomous vehicles and avionics, medial monitoring and robotics. The incarnation of the CPSs can therefore range from considering individual Internet-of-Things devices through to large-scale infrastructures.  

Presented across ten chapters authored by international researchers in the field from both academia and industry, this book offers a series of high-quality contributions that collectively address and analyze the state of the art in the security of Cyber-Physical Systems and related technologies. The chapters themselves include an effective mix of theory and applied content, supporting an understanding of the underlying security issues in the CPSs domain, alongside related coverage of the technological advances and solutions proposed to address them. The chapters comprising the later portion of the book are specifically focused upon a series of case examples, evidencing how the protection concepts can translate into practical application. 

The contribution of this paper is a new strategy of integrating multiple recognition outputs of diverse recognizers. Such an integration can give higher performance and more accurate outputs than a single recognition system. The problem of aligning various Optical Character Recognition (OCR) results lies in the difficulties to find the correspondence on character, word, line, and page level. These difficulties arise from segmentation and recognition errors which are produced by the OCRs. Therefore, alignment techniques are required for synchronizing the outputs in order to compare them. Most existing approaches fail when the same error occurs in the multiple OCRs. If the corrections do not appear in one of the OCR approaches are unable to improve the results.We design a Line-to-Page alignment with edit rules using Weighted Finite-State Transducers (WFST). These edit rules are based on edit operations: insertion, deletion, and substitution. Therefore, an approach is designed using Recurrent Neural Networks with Long Short-Term Memory (LSTM) to predict these types of errors. A Character-Epsilon alignment is designed to normalize the size of the strings for the LSTM alignment. The LSTM returns best voting, especially when the heuristic approaches are unable to vote among various OCR engines. LSTM predicts the correct characters, even if the OCR could not produce the characters in the outputs. The approaches are evaluated on OCR’s output from the UWIII and historical German Fraktur dataset which are obtained from state-of-the-art OCR systems. The experiments shows that the error rate of the LSTM approach has the best performance with around 0.40%, while other approaches are between 1,26% and 2,31%.

In an era where networks have become such complex and large structures, establishments with bigger networks may find it very daunting to make changes directly to their networking devices. This is because any changes done to the physical network equipment directly without any prior testing carries a high risk of fatal consequences due to misconfiguration. This includes commonly occurring issues like security vulnerabilities, suboptimal performance, downtime, and service disruptions. At the same time, the technologies of simulation and emulation reign supreme and are at their height with regard to today's technology. Organizations like Repay Invest AB, a client of Atea, face the issues discussed above. Navigating, testing, and implementing changes within Repay's expansive network, spanning numerous branch offices across Sweden, often feels like a high-stakes gamble. With each adjustment or configuration task, the potential for unforeseen consequences becomes bigger. This thesis embarks on a journey to analyze Repay's physical network and to construct a virtual counterpart using the renowned network emulation tool, GNS3. The reason for simulating Repay’s physical network into GNS3 is that they expressed the need for a safer, more agile approach to network experimentation and configuration testing; their request serves as the impetus for this thesis.

The primary objective of this thesis is to conduct an analysis of Repay’s physical network and simulate it using GNS3 with a high level of accuracy. Additionally, the thesis also aims to compare today's modern networks to legacy/traditional network architectures. This comparison serves as background information for discussing the decisive and crucial changes that traditional networks underwent to adapt to the continuously expanding and advancing network structures. Furthermore, the solutions that traditional networks had to utilize during the changes will be brought up, highlighting the importance of software-defined networking and cloud computing in shaping today’s modern networks. Beyond the confines of the network analysis that will be conducted, this thesis also aims to delve deep into exploring GNS3 as a tool, highlighting limitations, features, and the reliability of GNS3 when used to simulate real-world network environments. In essence, this thesis is not merely a technical exercise but a voyage of discovery, one that seeks to further the understanding of what network simulation and emulation is and how they could be used to benefit the network of an average IT-company. Through meticulous analysis, informed comparison, and relentless exploration, the goal is to pave the way towards a more secure, efficient, and resilient network future for Repay Invest AB.

The most commonly used programming languages for modern software development usually belong to either the Imperative Paradigm (IP) or Declarative Paradigm (DP). These paradigms often come with drawbacks like code coupling and structural and/or temporal redundancy. The Notification Oriented Paradigm (NOP) comes as a new approach for software development that works based on small, smart, and reactive notifiable entities. That is how NOP facilitates software development to achieve several features like responsiveness by avoiding code redundancy and distributiveness by allowing code decoupling. This research focused on examining some of NOP green potentials in a simulated sensor correlation application, in smart city-like, by comparing the performance of a NOP Implementation with other common and popular object-oriented programming languages. The NOP implementation is the so-called NOP C++ Framework 4.0, which is the current state of the technics in this domain. In order to explore the NOP C++ Framework 4.0, an air quality monitoring system prototype was developed considering the presence of air quality sensors in three different locations of a supposed smart city. Beyond the prototype implemented in NOP C++ framework 4.0, it was as well implemented in C++ and Java programming languages in order to compare them. The aim is to evaluate the performance of the NOP state of technics, which will help to identify the green potentials of NOP and also its applicability in a smart city context. Two air quality datasets collected from real-time sensors located in two different cities of different countries were used to evaluate the performance of the applications. The performance analysis shows that the NOP application outperformed the other two for both datasets in terms of execution time, memory usage, and energy consumption. Future works should consider the prototypical NOP programming language, the so-called NOP state of the art, that has better performance than the NOP C++ Framework 4.0 because its compiler generates low-level-like code.

Bookkeeping data free of fraud and errors is a cornerstone of legitimate business operations. Highly complex and laborious financial auditors’ work calls for finding new solutions and algorithms to ensure the correctness of financial statements. Both supervised and unsupervised machine learning (ML) techniques, nowadays, are being successfully applied to detect fraud and anomalies in data. In accounting, it is a long-established problem to detect financial misstatements deemed anomalies in General Ledger (GL) data. Currently, widely used techniques such as random sampling and manual assessment of bookkeeping rules become challenging and unreliable due to increasing data volumes and unknown fraudulent patterns. To address the sampling risk and financial audit inefficiency, we applied seven supervised ML techniques inclusive of Deep Learning and two unsupervised ML techniques such as Isolation Forest and Autoencoders. We trained and evaluated our models on a real-life GL dataset and used data vectorization to resolve journal entry size variability. The evaluation results showed that the best trained supervised and unsupervised models have high potential in detecting predefined anomaly types as well as in efficiently sampling data to discern higher-risk journal entries. Based on our findings, we discussed possible practical implications of the resulting solutions in the accounting context.

Augmented Reality (AR) and Mixed Reality (MR) have increased in attention recently and there are several implementations in video games and entertainment but also work-related applications. The technology can be used to guide workers in order to do the work faster and reduce human error while performing their tasks.  The potential of this kind of technology is evaluated in this thesis through a proof-of-concept prototype which guides a novice in the kitchen in following a recipe and completing a dish. The thesis shows a comparison between five different object detection algorithms, selecting the best in terms of time performance, energy performance and detection accuracy. Then the selected object detection algorithm is implemented in the prototype application.

I många tusen år har vad som anses vara etiskt försvarbart diskuterats. Allt fler perspektiv beaktas

bland annat beroende på den tekniska utvecklingen. På senare tid har utvecklingen av de

digitala medierna inneburit att personlig information värderas och hanteras på nya sätt. Tekniken

innebär att nya möjligheter skapas samtidigt som det finns en risk att personlig information

missbrukas. Detta kan påverka vad som är etiskt försvarbart.

Detta arbete handlar om hur personlig information hanteras med hjälp av digital teknik. Det

undersöks om användare av IT är medvetna om att deras personliga information samlas in,

sprids och används för att skapa en mängd nya individuella tjänster. Det utforskas om denna

hantering, samt de nya möjligheterna att individualisera tjänster, är något som skapar dilemma.

Till sist sammanställs rekommendationer som framkommit under studien gällande

vad enskilda individer kan göra om de vill minska riskerna för att personlig information missbrukas.

Studien inleds med en litteraturgenomgång vilken belyser hur IT påverkat vårt samhälle samt

hur personlig information hanteras med den nutida teknikens möjligheter.

Det beskrivs bland annat:

 Hur företag samlar in användardata i bakgrunden utan att tydligt informera om det

 Att det finns profiler om oss alla

 Att den information vi får är anpassad till oss som individer

 Att personlig information finns lättillgänglig och sökbart för alla bland annat för arbetsgivare

Etik och moral bör ligga till grund för alla resonemang som avser avgöra vad som är rätt och

fel. Företagen har riktlinjer för hur de bör handla enligt etiska och moraliska principer. Det

finns lagstiftning och regelverk som styr men informationshantering utvecklas snabbt och är

global vilket gör situationen komplex. Tekniken utvecklas ofta snabbare än regelverket.

Metoden för insamlande av empiri bestod av två delar. Primärdata samlades in via intervjuer

och sekundärdata söktes via nätet och medier för att få aktuell data som underlag till intervjuerna.

Intervjuerna genomfördes som semikonstruerade intervjuer där datorer användes som

hjälpmedel för att exemplifiera data som fanns tillgänglig om respondenten. Den empiri som

framkom från intervjuerna innehåller en bred beskrivning av ämnesområdet även om urvalsgruppen

var liten. Insamlad data analyserades genom att notera och sammanställa svaren samt

söka mönster och samband mellan olika teman i intervjun.

Slutsatser sammanställdes vilka visar på att det finns oroväckande låg medvetenhet inom ämnesområdet.

Slutsaterana innehåller också tre dilemman kopplade till hur den tekniska utvecklingen

hanterar personlig information. Dessa är faktaresistens, kränkning av den personliga

integriteten och utformning av lagar, avtal och normer. Sist sammanställs rekommendationer

avsedda för de som vill minska riskerna med hantering av personlig information.

I slutet diskuteras att det är först när allmänheten börjar reagera som en diskussion kan uppstå.

Först då kan en eventuell förändring ske där en reglering som behandlar hur individer får beröras

och hur tekniken får användas växer fram.

Due to the ongoing climate crisis, reducing waste and carbon emissions has become hot topic in many fields of study. Cloud data centers contribute a large portion to the world’s energy consumption. In this work, methodologies are developed using machine learning algorithms to improve prediction of the energy consumption of a container in a data center. The goal is to share this information with the user ahead of time, so that the same can make educated decisions about their environmental footprint.This work differentiates itself in its sole focus on optimizing prediction, as opposed to other approaches in the field where energy modeling and prediction has been studied as a means to building advanced scheduling policies in data centers. In this thesis, a qualitative comparison between various machine learning approaches to energy modeling and prediction is put forward. These approaches include Linear, Polynomial Linear and Polynomial Random Forest Regression as well as a Genetic Algorithm, LSTM Neural Networks and Reinforcement Learning. The best results were obtained using the Polynomial Random Forest Regression, which produced a Mean Absolute Error of of 26.48% when run against data center metrics gathered after the model was built. This prediction engine was then integrated into a Proof of Concept application as an educative tool to estimate what metrics of a cloud job have what impact on the container power consumption.

In todays society we become more and more dependent on the technology we develop, so it is of greater importance that we always have access to the information we need to do our daily tasks. It is more pressing than ever that the developed systems should work well and that you should have quick response times for everything to flow as smooth as possible and if there would be a critical system crash, the information should not be completely lost. Therefore, it has become more popular to go from, having personal servers to use more and more applications and information in the so-called "cloud". That´s why it is becoming increasingly important to keep track of some of the major cloud service providers and how to use them to archive a better business. This report will be about how a work was done at TeamNorrs request to make a comparison between Microsoft Azure, Google Cloud Platform and Amazon Web Services. It will also review the response times to some of the worlds data centers with a starting point from TeamNorrs headquarters in Umeå. It also describe how you can put up your own environment to test Microsoft Azure, Google Cloud Platform and Amazon Web Services, and how to do a so-called "Site to Cloud" solution with a Virtual Private Network tunnel in Microsoft Azure, where it will also show you how you can create an MSSQL backup in the cloud. Then discusses the pros and cons of the work done and what improvements could be relevant for the future.

Mobile cloud gaming (MCG) is an emerging concept which aims to deliver video games on-demand to users with the use of cloud technologies. Cloud technology allows the offloading of computation from a less powerful user device or thin client to more robust cloud servers to minimize power consumption and provide additional cloud services such as storage. MCG is therefore very helpful that can reduce the costs of expensive hardware, but the challenge is that it requires a high Quality of Service (QoS) in order to stream and play the games where the users have a high Quality of Experience (QoE). The goal of the study is to investigate how users' QoE is affected by network conditions while playing MCG and compare the results from a previous study.

A testbed was made in order to conduct subjective tests where users are going to play Counter Strike: Global Offensive (CS: GO) on a smartphone using Steam Remote Play. The testbed consists of a router, tablet, smartphone, headset, Xbox controller, USB-C multi-port adapter and four different PC's. Participants on campus, both students and non-students, were invited to participate in the experiment. A total of 24 participants completed the tests; however, results from two participants were excluded due to software issues. There were 23 network conditions that was tested for each user and included factors such as round-trip time (RTT), packet losses, bursty jitter, random jitter or combinations of different factors. A multi-platform tool, ALTRUIST, was used to control the applications and facilitate the data collection from the devices and NetEm changed the network conditions.

The results showed that the network condition [bj(rtt200i15)] had the highest mean opinion score (MOS) of the QoE of 4.5 for the users with 200 milliseconds of bursty jitter every 15 seconds. The worst network condition tested with the lowest QoE rating of 1.4 was network condition [rtt25pl12] that had 25 milliseconds of RTT and 12% packet losses. There were differences between the male and female participants where the MOS of the QoE results was significantly higher with up to 1.5 MOS QoE rating differences for the females compared to the males in network conditions with RTT with packet losses. However, the sample size was low with only 5 female participants compared to 18 male participants. The MOS of the QoE results separating play time under 10 hours per week and 10 or more hours per week showed no significant changes, where the largest QoE rating difference was 0.5 points. Network condition [rtt25pl12] and [rtt2pl35] had the largest differences in the MOS QoE ratings compared to the previous study, while both was not compared to the same corresponding network condition. The largest difference comparing the same network condition to the previous study was network condition [bj(rtt200i15)] with a difference of 1.1 points higher in the MOS QoE rating.

This thesis addresses the design of a productivity dashboard for small and medium-sized enterprises (SMEs) to track key performance indicators (KPIs) and highlight the requirements for SME’s undergoing the initial implementation of business intelligence (BI). The objective is to develop an easy-to-use web application prototype that incorporates a dashboard with data source selection, while considering research framework requirements and limitations. Further, the prototype aims to fulfil the conceptual requirements of the tool used for decision support systems, including remote access, scalability, customization, and intuitive data presentation.

The thesis successfully developed a prototype web application that allows remote access through a browser. The interface of the application received positive feedback from respondents and demonstrates high usability. As a result, the prototype's scalability is confirmed through the gradual integration of new functionality, depicted in a hierarchical diagram to guide future enhancements. However, limitations arise from the technology choices, making the integration of new data sources more challenging due to specific data structure and attribute requirements. The study provides clear guidelines and inspiration for SMEs and web developers in integrating BI tools during the initial phase of adopting decision support systems. The research offers a well-documented development and evaluation process, empowering SMEs to make informed decisions when implementing BI solutions.  

Industriella arbetsmiljöer utgör trots omfattande säkerhetsarbete fortfarande en risk för hälsa och välbefinnande för arbetstagarna. Moderna sensorer och tekniker möjliggör att upptäcka risker och olyckor i tid och därmed öka säkerheten inom industrier. Industriella miljöer utrustas idag ofta med vältäckande trådlösa kommunikationsnät som möjliggör positionering och kommunikation med sensorer som bärs av personal. Den här rapporten beskriver aktuella tillämpningar och tekniklösningar. Förstudien har inte identifierat någon särskild tillämpning som kraftfullt driver utveckling av bärbara sensorer för industriella miljöer. Däremot har ett flertal lovande tillämpningar hittats som för närvarande provas av industrier eller finns kommersiellt tillgängliga som tidiga produkter. Några initiativ kan stödja flera tillämpningar och/eller funktioner med samma arkitektur och hårdvara. Flera lösningar bygger på positionering och i viss mån kontextanpassning.

För fortsatt arbete föreslås utvärdera tillämpningar såsom (1) insamling av information för bättre uppföljning och analys av tillbud och olyckor, (2) stöd för genomförande av säkerhetsförbättrande åtgärder, baserat på analys av tillbud/olycka eller av annan anledning, samt (3) automatisk larmning vid ensamarbete och/eller särskilt riskfyllt arbete. Som ansats för fortsatt arbete föreslås att (A) definiera en flexibel arkitektur som möjliggör tester med olika typer av sensorer för olika tillämpningar, och etablera ett sådant testsystem, (B) identifiera existerande system till vilka integration behövs, samt (C) definiera återanvändbara funktioner för att säkert skydda den personliga integriteten efter behov som styrs av aktuell tillämpning och överenskommelse med företrädare för personal (dvs. fackföreningar), samt (D) hitta tydliga och väl avgränsade tillämpningar som kan provas praktiskt i målmiljöer.

Digital twin (DT) emerges as a key concept of the Industry 4.0 paradigm and beyond. However, the current literature lacks focus on humans and human activities as a part of complex system DTs. Acknowledging human aspects in DTs can enhance work performance, well-being, motivation, and personal development of professionals. This study examines emerging requirements for human digital twins (HDTs) in three use cases of industry–academia collaboration on complex systems. The results draw together the overall design problem and four design objectives for HDTs. We propose to combine the machine and human-related aspects of DTs and highlight the need for virtual-to-virtual interoperability between HDTs and machines alike. Furthermore, we outline differences between humans and machines regarding digital twinning by addressing human activities and knowledge-based behavior on systems. Design of HDTs requires understanding of individual professional characteristics, such as skills and information preferences, together with twinning between the physical and digital machine entities and interactions between the human and machine DTs. As the field moves toward including humans as a part of the DT concept, incorporating HDTs in complex systems emerges as an increasingly significant issue.

Ett problem som Ventiro stött på i sin verksamhet är en brist på underlag vid felsökning och fakturering. I dagsläget finns inga möjligheter att övervaka datatrafikmängden hos en specifik kund och därför har de varken vetskap om eller underlag som visar hur mycket datatrafik en kund står för. Detsamma gäller felsökningen om något går fel i deras serverhall.

Innehållet i denna rapport behandlar vägen mot en lösning på deras problem, där datainsamling från loggfiler, behandling av informationen som denna genererar samt en början på utvecklingen av en prototyp av en applikation ingår. Denna skapas med uppdraget att visa upp denna information på ett lättillgängligt sätt, och för att förse Ventiro med ett statistiskt underlag vid de redan nämnda situationerna.

För insamlingen av loggfiler användes Logstash, som är en del av Elastic Stack, som hämtade loggfiler från brandväggarna i deras serverhall. Dessa filtrerades sedan med hjälp av Elasticsearch, för att sedan visas upp i en applikation. Denna applikation var tänkt att skrivas i PHP med hjälp av ramverket CodeIgniter, men eftersom tiden var begränsad och inte räckte till så avbröts utvecklingsarbetet. Arbetets fokus hamnade istället på att ta reda på vilka delar av de redan insamlade loggfiler som skulle kunna användas, vilket presenteras tillsammans med förslag på justeringar som hade kunnat förbättra applikationen och skräddarsy den ytterligare för att bättre passa Ventiros verksamhet.

The low-latency requirements of 5G are expected to increase the demand for distributeddata storage and computing capabilities in the form of small-scale data centers (DC)located at the edge, near the interface between mobile and wired networks. These edgeDC will likely be of modular and standardized designs, although configurations, localresource constraints, environments and load profiles will vary and thereby increase theDC infrastructure diversity. Autonomy and energy efficiency are key objectives for thedesign, configuration and control of such data centers. Edge DCs are (by definition)decentralized and should continue operating without human intervention in the presenceof disturbances, such as intermittent power failures, failing components and overheating.Automatic control is also required for efficient use of renewable energy, batteries and theavailable communication, computing and data storage capacity.

These objectives demand data-driven models of the internal thermal and electricprocesses of an autonomous edge DC, since the resources required to manually defineand optimize the models for each DC would be prohibitive. In this thesis machinelearning methods that are implemented in a modular design are evaluated for thermalcontrol of such modular DCs. Experiments with small server clusters are presented, whichwere performed in order to investigate what parameters that are important in the designof advanced control strategies for autonomous edge DC. Furthermore, recent transferlearning results are discussed to understand how to develop data driven models thatcan be deployed to modular DC in varying configurations and environmental contextswithout training from scratch.

The first study demonstrates how a data driven thermal model for a small clusterof servers can be calibrated to sensor data and used for constructing a model predictivecontroller for the server cooling fan. The experimental investigations of cooling fancontrol continues in the next study which explores operational sweet-spots and energyefficient holistic control strategies. The machine learning based controller from the firststudy is then re-purposed to maintain environmental conditions in an exhaust chamberfavourable for drying apples, as part of a practical study how excess heat produced bycomputation can be used in the food processing industry. A fourth study describes theRISE EDGE lab - a test bed for small data centers - built with the intention to exploreand evaluate related technologies for micro-grids with renewable energy and batteries,5G connectivity and coolant storage. Finally the last work presented develops the modelfrom the first study towards an application for thermal based load balancing.

The National Space Data Lab is a collaboration project between Swedish National Space Agency, RISE Research Institutes of Sweden, Luleå University of Technology and AI Sweden. It will be a national knowledge and data hub for Swedish authorities’ work on earth observation data and for the development of AI-based analysis of data, generated in space systems. The platform is deployed on Kubernetes.

Purpose

• Increase the availability of space data for the benefit of developing society and industry

• Provide platform for accessing space data and analytical tools

In some fields of industrial automation, such as nuclear power plant (NPP) industry in Finland, thorough verification of systems and demonstration of their safety are mandatory. Model checking is one of the techniques to achieve a high level of reliability. The goal of this paper is practical: we explore which type of model checking - either explicit-state or symbolic - is more suitable to verify instrumentation and control (I&C) applications, represented as function block networks. Unlike previous studies, in addition to the common open-loop approach, which views the controller model alone, we consider closed-loop verification, where the plant is also modeled. In addition, we present a procedure to translate block networks to the language of the SPIN explicit-state model checker.

To facilitate correctness and safety of mission-critical automation systems, formal methods should be applied in addition to simulation and testing. One of such formal methods is model checking, which is capable of verifying complex requirements for the system's model. If both the controller and the controlled plant are formally modeled, then the variant of this technique called closed-loop model checking can be applied. Recently, a technique of automatic plant model generation has been proposed which is applicable in this scenario. This paper continues the work in this direction by presenting two plant model construction approaches which are much more scalable with respect to the previous one, and puts this work into a more practical context. The approaches are evaluated on a case study from the nuclear automation domain

To reduce uploading bandwidth and address privacy concerns, deep learning at the network edge has been an emerging topic. Typically, edge devices collaboratively train a shared model using real-time generated data through the Parameter Server framework. Although all the edge devices can share the computing workloads, the distributed training processes over edge networks are still time-consuming due to the parameters and gradients transmission procedures between parameter servers and edge devices. Focusing on accelerating distributed Convolutional Neural Networks (CNNs) training at the network edge, we present DynaComm, a novel scheduler that dynamically decomposes each transmission procedure into several segments to achieve optimal layer-wise communications and computations overlapping during run-time. Through experiments, we verify that DynaComm manages to achieve optimal layer-wise scheduling for all cases compared to competing strategies while the model accuracy remains untouched.