This study explored how clients with distorted data affected the Federated Learning process using the FedAvg and FedProx algorithms. Different amounts of the three distortions, Translation, Rotation, and Blur, were tested using three different Machine Learning models. The models were a Dense network, the well-known convolutional network LeNet-5, and a smaller version of the ResNet architecture. The results of the study successfully showcases how different distortions affect the three models. Therefore, they also show that the risk of local data distortion is important to factor in when picking a Machine Learning model for Federated Learning.

In Sweden there is a shortage of housing whilst old, abandoned industrial buildingstake up space on desirable land. These buildings generally do not live up to today’s standards regarding comfort and energy efficiency and are therefore not used today. If these buildings were to be renovated into housing, they could help remedy the housing shortage. However, many of these buildings are protected from distortion by law which makes renovation and remodeling difficult. The purpose of this work has been to provide solutions for how an old industrialbuilding with cultural-historical values can be remodeled into housing. The solutions consist of floor plans and drawings modeled in Autodesk Revit. The possibilities and difficulties with this kind of project as well as solutions to expected problems are also presented in the report. This work was executed as a case study on one of the old locomotive depots in Bollnäs. Since the building for this project is an old locomotive depot, it is positioned close to the railroad. Noise and ground pollution are therefore important to consider and demand certain measures, for example retaining walls, noise barriers and excavation of polluted soil. An energy simulation of the building was made using BV2 to determine the energy usage of the building. The walls, roof and slab were additionally insulated for the purpose of meeting the demands of energy efficiency and comfort defined by BBR 29. The insulation had to be internally constructed in order to avoid distortion of the facade. Additionally, all windows and doors had to be made energy efficient while remaining aesthetically preserved. Internal insulation is generally problematic due to problems with moisture. The solution to this problem was to insulate the walls with polyurethane foam and the slab with cellular plastic. In order to make the building energy efficient, the building should be heated using district heating in combination with photovoltaic panels. This building is well suited for solar panels due to the large, black metal sheet roof and shadow free placement. The study shows the complexity of refurbishing this type of building. The renovation is extensive, and the floor plans are characterized by compromises. It is possible to renovate the building without distorting it. A better alternative would be to use the building for other functions such as a store, restaurant or bar. / I Sverige råder det bostadsbrist samtidigt som gamla industribyggnader stårövergivna på attraktiv mark. En anledning till att dessa byggnader i nuläget inteanvänds är att de av byggnadstekniska skäl inte uppfyller dagens krav på komfort och energieffektivitet. Om dessa byggnader skulle kunna göras om till bostäder vore deten lösning för att minska bostadsbristen. Många av dessa byggnader omfattas dock av förvanskningsförbud och varsamhetskrav vilket försvårar renovering och ombyggnation. Syftet med arbetet har varit att ta fram ett ändringsförslag på hur en gammal industribyggnad med ett kulturhistoriskt värde kan göras om till bostäder. Arbetet har också gått ut på att undersöka vilka möjligheter och svårigheter som finns vid ett sådant ändringsarbete och ta fram lösningar till eventuella problem. Arbetet utfördes i form av en fallstudie på ett av lokstallen i Bollnäs. Då byggnaden i projektet är ett gammalt lokstall ligger denna nära tågspåren. Buller och markföroreningar är därför viktigt att ta hänsyn till och kräver åtgärder i formav uppförandet av bullerplank och schaktning. En energisimulering gjordes av byggnaden i BV2 för att få ett värde på byggnadensenergianvändning. För att byggnaden ska kunna hålla en god komfort och klarakraven i BBR 29 tilläggsisoleras klimatskalet. Varsamhetskravet gör att all tilläggsisolering måste ske på byggnadens insida. Alla fönster och dörrar byts ut till energismarta sådana under förutsättning att de är estetiskt likvärdiga mot de ursprungliga för att motverka förvanskning. Att tilläggsisolera en gammal byggnad invändigt är ur fuktsynpunkt negativt. Lösningen på detta var att isolera väggarna med polyuretanskum och betongplattan underifrån med cellplast. För energieffektiviseringen valdes också att byggnaden skulle värmas med fjärrvärme då den ligger centralt och i nära anslutning tillfjärrvärmenätet. Denna typ av byggnad är väl lämpad för användning av solenergi tack vare dess stora svarta plåttak och skuggfria placering. Därmed valdes fjärrvärmen att kombineras med solcellspaneler. Studien visar att ändringsarbeten på denna typ av byggnader är komplexa. Energirenoveringen är omfattande och arbetet med planlösningarna präglas av kompromisser. Det är fullt möjligt att energirenovera byggnaden utan att förvanska den. Byggnaden i det här projektet vore dock bättre lämpad för exempelvis butik, restaurang eller krog än för lägenheter.

Surface acoustic wave (SAW) technology has been in use for well over one century. In the last few decades, due to its low cost and high performance, this technology has been widely adopted in modern wireless communication systems, to build filtering devices at radio frequency (RF). SAW filters and duplexers can be virtually found inside every mobile handset. SAW devices are traditionally recognized as passive devices with high linear signal processing behavior. However, recent deployments of third generation (3G) and fourth generation (4G) mobile networks require the handsets to handle an increasing number of frequency bands with more complex modulation /demodulation schemes and higher data rate for more subscribers. These requirements directly demand more stringent linearity specifications on the front end devices, including the SAW duplexers. In the past, SAW duplexer design was based on empirically obtained design rules to meet the linearity specifications. Lack of predictability and an understanding of the root cause of the nonlinearity have limited the potential applications of SAW duplexers. Therefore, research on the nonlinearity characterization and an accurate modeling of SAW nonlinearity for mobile device applications are very much needed. The Ph.D. work presented here primarily focuses on developing a general nonlinear model for SAW resonators/duplexers. Their nonlinear characteristics were investigated by measuring the harmonic and intermodulation distortions of resonators. A nonlinear Mason model is developed and the characterization results are integrated into SAW duplexer design flows to help to simulate the nonlinear effects accurately and improve the linearity performance of the products. iv In this dissertation, first, a novel nonlinear Mason equivalent circuit model including a third order nonlinear coefficient in the wave propagation is presented. Next, the nonlinear distortions of SAW resonators are analyzed by measuring large-signal harmonic and intermodulation spurious emission on resonators using a wafer probe station. The influence of the setups on the measurement reliability and reproducibility is discussed. Further, the nonlinear Mason model is validated by comparing its simulation results with harmonic and intermodulation measurements on SAW resonators and a WCDMA Band 5 duplexer. The Mason model developed and presented here is the first and only nonlinear physical model for SAW devices based on the equivalent circuit approach. By using this new model, good simulation measurement agreements are obtained on both harmonic and intermodulation distortions for SAW resonators and duplexers. These outcomes demonstrate the validity of the research on both the characterization and modeling of SAW devices. The result obtained confirms that the assumption of the representation of the 3 rd order nonlinearity in the propagation by a single coefficient is valid

Application Specific Integrated Circuits (ASICs) are critical to delivering on 5G’s promises of high speed, low latency, and expanded capacity. Digital Front-End (DFE) ASICs are particularly important components because they enhance crucial signal processing activities. It handles duties including carrier mixing, up-sampling, and modulation-demodulation, allowing for efficient data transmission and reception inthe complicated 5G environment. The main aim of this work is to develop a power estimation tool for DFE radio ASICs and to understand the different use cases. It also studies the spread of power consumption, taking into account process and metal variations. The thesis provides a detailed case study of the DFE ASIC, including its Intellectual Property (IP) blocks, configurations, and protocols. It investigates the power consumption of DFE ASICs under various conditions, including active processing, power-saving mode, and no clock. In this thesis we build a power model that describes how the factors affect the ASIC’s power consumption. It also performs spread analysis to evaluate the impact of all factors using MATLAB tool. Based on this we then generate three distributionmodels to study the combined likelihood of the variations. It also uses Monte Carlo simulation to understand total power usage. Through this work we can conclude that the power consumption of DFE ASICs is affected by a variety of factors. The power model and spread analysis can be usedto forecast and optimize power usage in 5G systems.

No description available.

Understanding distortion transfer and generation through fan and compressor blade rows is able to assist in blade design and performance prediction. Using full annulus unsteady RANS simulations, the effects of distortion as it passes through the rotor of a transonic fan at five radial locations (10%, 30%, 50%, 70%, and 90% span) are analyzed. The inlet distortion profile is a 90-degree sector with a 15% total pressure deficit. Fourier distortion descriptors are used in this study to quantitatively describe distortion transfer and generation. Results are presented and compared for three operating points (near-stall, design, and choke). These results are used to explain the relationship between inlet total pressure distortion, pressure-induced swirl, total pressure distortion transfer, total temperature distortion generation, and circumferential rotor work variation. It is shown that very large changes in pressure-induced swirl and distortion transfer and generation occur between near-stall and design, but only small changes are seen between design and choke. The greatest changes are shown to be near the tip. Local power variations are shown to correlate with total pressure distortion transfer and total temperature distortion generation.It can be difficult to predict the transfer of distortion through a fan or compressor because traditional experimental and computational methods are very expensive and time consuming. The Harmonic Balance approach is a promising alternative which uses Fourier techniques to represent fluid flow solutions and which can provide unsteady solutions much more quickly than traditional unsteady solvers. Relatively little work has been done to assess how much Fourier information is necessary to calculate a sufficiently accurate solution with the Harmonic Balance Solver. A study is performed to analyze the effects of varying the amount of modal content that is used in Harmonic Balance simulations. Inlet distortion profiles with varying magnitudes are used in order to analyze trends and provide insight into the distortion flow physics for various inlet conditions. The geometry is a single stage axial compressor that consists of an inlet guide vane followed by the NASA Stage 37 rotor. It is shown that simulations with greater magnitudes of distortion require more modal content in order to achieve sufficiently accurate results. Harmonic Balance simulations are shown to have significantly lower computational costs than simulations with a conventional unsteady solver.

In this thesis, novel progressive scalable region-of-interest (ROI) image coding schemes with rate-distortion-complexity trade-off based on residual vector quantization (RVQ) and matching pursuit (MP) are developed. RVQ and MP provide the encoder with multi-resolution signal analysis tools, which are useful for rate-distortion trade-off and can be used to render a selected region of an image with a specific quality. An image quality refinement strategy is presented in this thesis, which improves the quality of the ROI in a progressive manner. The reconstructed image can mimic foveated images in perceptual image coding context. The systems are unbalanced in the sense that the decoders have less computational requirements than the encoders. The methods also provide interactive way of information refinement for regions of image with receiver 's higher priority. The receiver is free to select multiple regions of interest and change his/her mind and choose alternative regions in the middle of signal transmission. The proposed RVQ and MP based image coding methods in this thesis raise a couple of issues and reveal some capabilities in image coding and communication. In RVQ based image coding, the effects of dictionary size, number of RVQ stages and the size of image blocks on the reconstructed image quality, the resulting bit rate, and the computational complexity are investigated. The progressive nature of the resulting bit-stream makes RVQ and MP based image coding methods suitable platforms for unequal error protection. Researchers have paid lots of attention to joint source-channel ( JSC) coding in recent years. In this popular framework, JSC decoding based on residual redundancy exploitation of a source coder output bit-stream is an interesting bandwidth efficient approach for signal reconstruction. In this thesis, we also addressed JSC decoding and error concealment problem for matching pursuit based coded images transmitted over a noisy memoryless channel. The problem is solved on minimum mean squared error (MMSE) estimation foundation and a suboptimal solution is devised, which yields high quality error concealment with different levels of computational complexity. The proposed decoding and error concealment solution takes advantage of the residual redundancy, which exists in neighboring image blocks as well as neighboring MP analysis stages, to improve the quality of the images with no increase in the required bandwidth. The effects of different parameters such as MP dictionary size and number of analysis stages on the performance of the proposed soft decoding method have also been investigated. / Thesis / Doctor of Philosophy (PhD)

In the early days of gear manufacturing, engineers aimed to produce teeth as close to the theoretical shape as possible. As machinery advanced and demand for high-speed, high-torque, and quieter transmissions grew, optimizing gear teeth became increasingly important [1]. Techniques such as tip and root relief reduced engagement shocks, while lead modifications like lead slope and crowning increased load-carrying capacity. Gas turbines are complex, high-load machines that require the use of suitable gearboxes with highly reliable components. The importance of this reliability cannot be overstated, as any downtime can result in significant financial repercussions. By implementing the right modifications, it is possible to reduce the causes of failure and improve the overall performance of the gearboxes, leading to more efficient and cost-effective operations. The primary focus of this thesis is to explore the use of various modification methods to achieve the most optimal load distribution across the face width of a double helical gear. This is accomplished through the use of FVA workbench and analytical calculations, allowing for a thorough and comprehensive analysis of these modifications. / Ingenjörer när växeltillverkning var i sin linda, hade som avsikt att producera kuggar så nära dess teoretiska form som möjligt. I takt med att maskiner utvecklades ökade proportionell efterfrågan på transmissioner med hög hastighet och vridmoment med tystare drift, sedan dess har optimeringen av kuggarna fått grepp. Profilmodifieringar som topp- och fotavlättning minskade ingreppsstötar, på engelska ”lead modification”, ”lead slope” och ”crowning” ökade lastkapaciteten. Gasturbiner är höglastade maskiner som kräver behov av lämplig växellåda med behov av mycket pålitliga komponenter för att minska och eventuellt undvika stillestånd då antalet maskiner är många och det kommer att kosta mycket i sådana fall. Om det görs rätt kan modifieringar minska orsakerna till fel och förbättra växellådornas prestanda. Denna avhandling behandlar användningen av dessa modifieringsmetoder för att erhålla den mest optimala lastfördelningen över kuggflanken på en snedkuggsväxel, vilket görs med FVA-Workbench och analytiska beräkningar.

No description available.

No description available.