There exists an increasing demand for computing power and performance in real-time embedded systems, as new, more complex customer requirements and function-alities are appearing every day. In order to support these requirements and func-tionalities without breaking the power consumption wall, many embedded systems areswitching from traditional single-core hardware architectures to multi-core architec-tures. Multi-core architectures allow for parallel execution of tasks on the multiplecores. This introduces many benets from the perspective of achievable performance,but in turn introduces major issues when it comes to the timing predictability ofthe real-time embedded system applications deployed on them. The problem arisesfrom unpredictable and potentially unbounded inter-core interferences, which occuras a result of contention for the shared resources, such as the shared system busor shared system memory. This thesis studies the possible application of constraintprogramming as a resource optimization technique for the purpose of creating oineschedules for tasks in real-time embedded system applications executing on a dual-core architecture. The main focus is placed on tightening inter-core data-propagationinterferences, which can result in lower over-all data-propagation delays. A proto-type of an optimization engine, employing constraint programming techniques on ap-plications comprised of tasks structured according to the Phased Execution Model isdeveloped. The prototype is evaluated through several experiments on a large numberof industry inspired intellectual-property free benchmarks. Alongside the experimentsa case study is conducted on an example engine-control application and the resultingschedule is compared to a schedule generated by the Rubus-ICE industrial tool suite.The obtained results show that the proposed method is applicable to a potentially widerange of abstract systems with dierent requirements. The limitations of the methodare also discussed and potential future work is debated based on these results. / <p>Presentation was held over Zoom, due to the COVID-19 situation.</p>

This thesis studies the options of using PCB embedding bare die power MOSFET and new packaging of MOSFET to increase the power density in a PCB. This is to decrease the winding losses in an isolated DC/DC converter which, according to "Flex Power Modules", can be done by improving the interleaving between the layers of the transformer and/or decreasing the AC loop. To test the MOSFET packaging two layout are made from a reference PCB, one using embedded MOSFET and the other using the new packaging. The leakage induction and winding losses are simulated and if they are lower compared to the reference PCB prototypes are manufactured. The simulated result is that PCB embedded MOSFET decrease the leakage induction but the winding loss is higher. With the new packaging the leakage induction is higher and the winding loss has linear characteristics. Only the PCB with the new MOSFET packaging is made because the MOSFET die gate pad is too small for the PCB manufacturer to make a via connection to it. The PCB is tested that it operates as a DC/DC converter with a 40-60 V input and a 12 V output. The PCB is put on a test board in a wind-tunnel to test its characteristics under different wind speeds, input voltage and loads. The result is that the PCB has a higher efficiency than the reference PCB but it has worse thermal resistance. Further development of the design needs to be made to improve the thermal resistance. Using new packaging is a way to continue the development of power converter with lower efficiency but embedding MOSFET needs a less complicated manufacturing process before there is any widespread usage.

Global logistics market is expected to reach US$15.5 trillion by 2023 but many companies loss billions of dollars because of improper handling, theft and tracking problems. Especially when high value goods are transported there is a need to monitor their condition because they can be affected by the environmental changes that might be harmful or due to any dangerous mechanical conditions during transportation. Real time condition monitoring applications serve the purpose of providing alerts, analytics and reports of these harmful or dangerous conditions to ensure that goods are transported without any damage or with minimal damage. This thesis considers how to combine software and hardware solution to monitor different conditions and report during transportation. This report focusses on designing and testing the device for real time condition monitoring of consumer goods package during transportation. The proposed device, called Asset Condition Monitoring System, uses FTP protocol to transfer the GPS information along with monitored information through different sensors as an event over a GSM network to a server. The design can be adapted to any type of product. The prototype developed is equipped with different sensors that assess the environmental conditions along with some of mechanical aspects of the package throughout the transit.  The device is tested in a real time environment and provides information on the possible events that might compromise the integrity of the package. The system enables the customer to assess the quality of the package. The design is based on agile method in a user driven framework. The thesis design, model and tests the condition monitoring system according to the requirements and constraints, which can be different to different assets while considering packaging and mode of transportation. For this project, Seagate Hard Drives are considered as an asset to monitor.

Målet med detta examensarbete är att undersoka om det är möjligt att reducera energiförbrukningen i ett inbyggt system m.h.a. en Field Programmable Gate Array (FPGA) med låg effektförbrukning. Genom att flytta en del funktioner från systemets Micro Controller Unit (MCU) till en FPGA, hoppas uppdragsgivaren att systemets MCU kan ges mojligheten att gå över i ett mer energisnålt sömnlage under tillräckligt långa perioder. Rapporten beskriver utvecklingsarbetet från förstudie till implementeriung och test av framtagen design i en FPGA, AGLN250 fran Microsemi. Examensarbetet har visat att det ar fullt mojligt att reducera ett inbyggt systems effektförbrukning m.h.a. en FPGA. Dock måste man, p.g.a. en FPGA:s arkitektur, vara extra aktsam pa hur designen implementeras för att effektförbrukningen inte skall bli högre än förvantat. / The purpose of this thesis is to examine the possibility of reducing an embedded system's power consumption through the use of a low-power Field Programmable Gate Array (FPGA). The customer's hope was that by relocating some of the functionality from the system's Micro Controller Unit (MCU) to an FPGA, the system's MCU could remain in its most efficient power saving mode long enough to reduce the average power consumption to an acceptable level. This paper documents the development work, from initial background material studies up to the implementation and test of suggested designs in an actual FPGA, an AGLN250 from Microsemi. The thesis work has demonstrated that it is possible to reduce the power consumption of the customer's system by relocating some of the MCU functionality to an FPGA. However, due to an FPGA's architecture, care must be taken to ensure that the design is implemented in such a way that the signal activity is reduced as far as possible. Otherwise the power consumption might end up higher than expected.

This report is about implementing an Internet of Things (IoT) system that is meant to measure the water quality of the Nissan river located in Halmstad. The system measures the temperature, pH value, turbidity, dissolved oxygen, total dissolved solids and electrical conductivity of the water. The project is named Catfish, and the authors of the report are in the Data and Sensors team. The purpose of the Data and Sensors team is to program the prototype to collect the specified environmental values, combining all functionalities of the prototype in the main code, and sending the data wirelessly to a database in real-time, including creating the database. To collect the data, two Arduino microcontrollers and sensors from DFRobot are used. For the central computing, a Raspberry Pi is used. To send the data in real-time, a 4G dongle is connected to the Raspberry Pi and to transmit the data, the protocol TCP/IP is used. In order to create a local database, SQLite is used, and for the database server, MariaDB is used. The project has succeeded to collect data from the specified sensors, send the data wirelessly over the Internet with the help of a 4G modem, and store the data in a database on a remote temporary server. However, the sensors are not calibrated due to lack of time, which means the data received from the sensors can not be trusted. The server that stores the data is a temporary server implemented by the Data and Sensors team since there was no other place to store the data. / Denna rapporten handlar om en implementation av ett Internet of Things (IoT) system som är menad för att mäta vattenkvalitéen i Nissan i Halmstad. Systemet mäter temperaturen, pH värde, grumlighet, upplöst syre, totalmängden upplösta ämnen och elektrisk konduktivitet i vattnet. Detta projektet heter Catfish och författarna av denna rapport är del av Data och Sensor gruppen. Syftet av Data och Sensor gruppen är att programmera prototypen till att samla de specifierade miljövärdena, kombinera prototypens funktioner i huvud-koden och att skicka datan till en databas trådlöst och i real-tid, vilket inkulderar att skapa databasen.  För att samla datan används två Arduino mikrokontrollers och sensorer från DFRobot. För den centrala datoranvändningen används en Raspberry Pi. För att skicka data i realtid används ett 4G modem som kopplas till Raspberry Pi:n och skickar datan till servern genom TCP/IP. Skapandet av en lokal databas är gjord i SQLite och för databasen i servern används MariaDB.  Projektet har lyckats med att samla data från de specifierade sensorerna, skicka datan trådlöst över Internetet med hjälp av ett 4G modem och lagra datan i en databas i en temporär fjärrserver. Sensorerna är dock inte kalibrerade på grund av tidsbrist, vilket betyder att man inte kan lita på datan från sensorerna. Servern där datan lagras är temporär och implementerad av Data och Sensor gruppen eftersom det inte fanns någon annanstans att lagra datan.

Aus der Vision, Computertechnik für den schmalen Geldbeutel technisch interessierten Jugendlichen verfügbar zu machen entstand ein kreditkartengroßer Einplatinencomputer, der Raspberry Pi. Wir möchten den Raspi im Vortrag etwas näher vorstellen und in einer kleinen Demonstration Anregungen für eigene Projekte geben. Gespannt sind wir auch auf Projekte, welche die Zuhörer bereits mit dem Raspberry Pi umgesetzt haben.

Networks connected to the internet are under a constant threat of attacks. To protect against such threats, new techniques utilising already connected hardware have in this thesis been proven to be a viable solution. By equipping network switches with lightweight machine learning models, such as, Decision Tree and Random Forest, no additional devices are needed to be installed on the network.When an attack is detected, the device may notify or take direct actions on the network to protect vulnerable systems. By utilising container software on Westermo's devices, a model has been integrated, limiting its computational resources. Such a system, and its building blocks, are what this thesis has researched and implemented. The system has been validated using multiple different models using a range of parameters.These models have been trained offline on datasets with pre-recorded attacks. The recordings are converted into flows, decreasing dataset size and increasing information density. These flows contain features corresponding to information about the packets and statistics about the flows. During training, a subset of features was selected using a Genetic Algorithm, decreasing the time for processing each packet. After the models have been trained, they are converted to C code, which runs on a network switch. These models are verified online, using a simulated factory, launching different attacks on the network. Results show that the hardware is sufficient for smaller models and that the system is capable of detecting certain types of attacks.

There are scenarios in the modern world of today when several things are being cooked at the same time on the stove in the kitchen. You typically have a saucepan that is boiling a carbohydrate. This requires attention and can result in elevated levels of mental exertion. Would it not be useful then to aid the cooking process by removing the boiling process as a point of attention by automating the boiling process?   Food to be boiled can be identified through image recognition. There is thus a possibility to automate boiling by using machine learning. In this project machine learning is used to automate the boiling of carbohydrates. A prototype has been developed which consists of a camera and a Raspberry Pi in which a convolutional neural network (CNN) model has been implemented. The prototype can identify pasta, potato, rice, their corresponding boiling states, and give correct indication when any of them is ready. A dataset has been created from scratch, containing 5607 images that were taken and labeled, and then used to train the CNN model.   The CNN model has been evaluated through a confusion matrix applied to an image dataset which was captured by the prototype. It was also evaluated through tables of successful boiling trials. The evaluation results show that the performance of the CNN model can identify carbohydrates in limited stove scenarios. The confusion matrix shows that the precision scores are 0.846, 0.959, 0.870, 0.688 for pasta, potato, rice and "no boiling item", respectively. Recall scores are 0.967, 0.848, 0.844 and 0.681 for pasta, potato, rice and "no boiling item", respectively. But it is not sufficiently reliable to be able to work in a wide range of scenarios because of the limited dataset. It has also been shown that it is possible to use the CNN model to guide the boiling of carbohydrates. But still the dataset is not sufficiently large to quantify the error rate of the boiling system. There is potential for this type of application but further work is needed.

I detta examensarbete utreddes huruvida elektriska tester av styrenheter för tunga fordon kan automatiseras på befintliga testmiljöer. Arbetet utfördes på uppdrag av Scania CV AB för arbetsgruppen Electronic Hardware som utför tester vid utveckling och verifiering av krav på styrenheter. Genom en automatiserad testprocess kan testarbetet effektiviseras och kvaliteten höjas. Testarbetet sker på testriggar som innehåller fullskalig hårdvara från lastbilar för att kunna emulera styrenhetens autentiska arbetsmiljö. För att komma fram till ett testsystem som kunde leva upp till de krav och behov som formulerats inom kravprocessen i arbetet användes utvecklingsmodeller. Utvärderingsmatriser användes för att välja den mjuk- och hårdvara som skulle vara mest lämplig för det automatiserade testsystemet utifrån kraven. Resultatet visade att testsystemet skulle bestå av en testprogramvara som körs på en vanlig persondator och ett inbyggt system med elektroniska komponenter för att kunna utföra de elektriska testerna. För att demonstrera testsystemets möjligheter i ett helhetstest konstruerades en prototyp som har funktionalitet att utföra ett testfall som kan dra stor nytta av att automatiseras, nämligen att testa omslagsnivåer för en digital ingång. Prototypen baserades på ett mikrokontrollerkort från Arduino och ett kretskort konstruerades till detta för att kunna utföra elektriska tester med högre spänningsnivåer som krävs för styrenheter på Scaniafordon. För att utforma testfall och hantera testprocessen valdes LabVIEW, en programvara där tester designas med ett grafiskt programmeringsspråk. Testresultaten för prototyptestsystemet visade att verifieringen av kraven för omslagsnivåer förenklas radikalt vid ett automatiserat förfarande, eftersom testtiden kunde minskas drastiskt, i synnerhet vid ett repetitivt förfarande. / This thesis work intends to study the possibilities to automate electrical testing of electronic control units in an existing test environment. The work was executed on behalf of Scania CV AB for the department Electronic Hardware, who run tests in development and verification of requirements for electronic control units. By using an automated testing system, the testing process could be made more effective and raise the quality. The testing work is done on test rigs which are equipped with full-scale hardware from real trucks to emulate the working environment the electronic control unit operates into. Development models were used to be able to create a testing system which could meet the requirements that were defined during the requirement engineering process. Evaluation matrices were used to choose the soft- and hardware that could be considered the most appropriate for the automated testing system according to the requirements. The result shown that the testing system should consist of a testing software that runs on an ordinary computer and an embedded system equipped with electronic components to enable the testing system for electrical tests. To demonstrate the possibilities of the whole testing system a prototype was manufactured which was designed to execute a test case that could greatly benefit from the advantages that comes with automation. It was a test case set out to measure voltage levels when switching in a digital input. The prototype was based on a microcontroller card from Arduino and was joined by a printed circuit board to be able to run electrical tests with the higher voltage levels that were demanded by electronic control units. To design test cases and manage the test process, LabVIEW was chosen, a software in which tests are designed in a graphical programming language. Test results for the prototype test system showed that verification of the requirements for voltage levels when switching was radically simplified by an automated procedure, as the test time could be drastically reduced, particularly in a repetitive procedure.

Because of Machine Learning, machines have become extremely good at image classification in near real time. With using significant training data, powerful machines can be trained to recognize images as good as any human would. Till now the norm has been to have pictures sent to a server and have the server recognize them. With increasing number of sensors the trend is moving towards edge computing to curb the increasing rate of data transfer and communication bottlenecks. The idea is to do the processing locally or as close to the sensor as possible and then only transmit actionable data to the server. While, this does solve plethora of communication problems, specially in industrial settings, it creates a new problem. The sensors need to do this computationally intensive image classification which is a challenge for embedded/wearable devices, due to their resource constrained nature. This thesis analyzes Machine Learning algorithms and libraries from the motivation of porting image classifiers to embedded devices. This includes, comparing different supervised Machine Learning approaches to image classification and figuring out which are most suited for being ported to embedded devices. Taking a step forward in making the process of testing and implementing Machine Learning algorithms as easy as their desktop counterparts. The goal is to ease the process of porting new image recognition and classification algorithms on a host of different embedded devices and to provide motivations behind design decisions. The final proposal goes through all design considerations and implements a prototype that is hardware independent. Which can be used as a reference for designing and then later porting of Machine Learning classifiers to embedded devices. / Maskiner har blivit extremt bra på bildklassificering i nära realtid. På grund av maskininlärning med kraftig träningsdata, kan kraftfulla maskiner utbildas för att känna igen bilder så bra som alla människor skulle. Hittills har trenden varit att få bilderna skickade till en server och sedan få servern att känna igen bilderna. Men eftersom sensorerna ökar i antal, går trenden mot så kallad "edge computing" för att stryka den ökande graden av dataöverföring och kommunikationsflaskhalsar. Tanken är att göra bearbetningen lokalt eller så nära sensorn som möjligt och sedan bara överföra aktiv data till servern. Samtidigt som detta löser överflöd av kommunikationsproblem, speciellt i industriella inställningar, skapar det ett nytt problem. Sensorerna måste kunna göra denna beräkningsintensiva bildklassificering ombord vilket speciellt är en utmaning för inbyggda system och bärbara enheter, på grund av sin resursbegränsade natur. Denna avhandling analyserar maskininlärningsalgoritmer och biblioteken från motivationen att portera generiska bildklassificatorer till inbyggda system. Att jämföra olika övervakade maskininlärningsmetoder för bildklassificering, utreda vilka som är mest lämpade för att bli porterade till inbyggda system, för att göra processen att testa och implementera maskininlärningsalgoritmer lika enkelt som sina skrivbordsmodeller. Målet är att underlätta processen för att portera nya bildigenkännings och klassificeringsalgoritmer på en mängd olika inbyggda system och att ge motivation bakom designbeslut som tagits och för att beskriva det snabbaste sättet att skapa en prototyp med "embedded vision design". Det slutliga förslaget går igenom all hänsyn till konstruktion och implementerar en prototyp som är maskinvaruoberoende och kan användas för snabb framtagning av prototyper och sedan senare överföring av maskininlärningsklassificatorer till inbyggda system.