Abstract This thesis focuses on two important challenges in wireless downlink transmission: multi-user (MU) precoder design and scheduling of users over time, frequency, and spatial resources at any given instant. Data streams intended for different users are transmitted by a multiple-input multiple-output (MIMO) multi-antenna orthogonal frequency division multiplexing (OFDM) system. The transmit precoders are designed jointly across space-frequency resources to minimize the number of backlogged packets waiting at the coordinating base stations (BSs), thereby implicitly performing user scheduling. Then the problem of multicast beamformer design is considered wherein a subset of users belonging to a multicasting group are served by a common group-specific data. The design objective is to either minimize the transmit power for a guaranteed quality-of-service, or to maximize the minimum achievable rate among users for a given transmit power. Unlike existing techniques, the proposed design utilizes both the spatial and frequency resources jointly while designing multi-group beamformers. As an extension to coordinated precoding, the problem of beamformer design for cloud radio access network is considered wherein beamformers are designed centrally, quantized and sent along with data to the respective BSs via backhaul. Since the users can be served by multiple BSs, beamformer design becomes a nonconvex combinatorial problem. Unlike existing solutions, beamformer overhead is also included in the backhaul utilization along with the associated data. As the number of antennas increases, backhaul utilization is dominated by the beamformers. Thus, to reduce the overhead, two techniques are proposed: varying the quantization precision, and reducing the number of active antennas used for transmission. Finally, to reduce the complexity involved in the design of joint space- frequency approach, a two-step procedure is proposed, where a MU-MIMO scheduling algorithm is employed to find a subset of users for each scheduling block. The precoders are then designed only for the chosen users, thus reducing the complexity without compromising much on the throughput. In contrast to the null-space-based existing techniques, a low-complexity scheduling algorithm is proposed based on vector projections. The real-time performance of all the schedulers are evaluated by implementing them on both Xilinx ZYNQ-ZC702 system-on-chip (SoC) and TI TCI6636K2H multi-core SoC. / Tiivistelmä Tässä väitöskirjassa keskitytään kahteen tärkeään langattoman tiedonsiirron haasteeseen alalinkkilähetyksissä: usean käyttäjän (MU) esikooderisuunnitteluun ja käyttäjien skedulointiin aika-, taajuus- ja tilaresurssien yli. Eri käyttäjille tarkoitettuja datavirtoja lähetetään käyttämällä monitulo-monilähtötekniikkaa (MIMO) yhdistettynä monikantoaaltomodulointiin (OFDM). Lähettimien esikooderit suunnitellaan yhteisesti tila- ja taajuusresurssien yli, jotta keskenään yhteistoiminnallisten tukiasemien jonossa olevien pakettien määrää voitaisiin minimoida samalla kun tehdään epäsuorasti käyttäjien skedulointia. Tämän jälkeen työssä paneudutaan monilähetysten (multicast) keilanmuodostussuunnitteluun, jossa monilähetysryhmään kuuluvien käyttäjien alijoukolle lähetetään yhteistä ryhmäspesifistä dataa. Suunnittelun päämääränä on joko minimoida kokonaislähetysteho tietyllä palvelunlaatuvaatimuksella tai maksimoida pienin saavutettavissa oleva siirtonopeus käyttäjien joukossa tietyllä lähetysteholla. Toisin kuin olemassa olevat menetelmät, ehdotetussa mallissa käytetään yhteisesti sekä aika- että taajuusresursseja usean ryhmän keilanmuodostusta suunniteltaessa. Laajennuksena yhteistoiminnalliselle esikoodaukselle, väitöskirjassa käsitellään myös keilanmuodostusta pilvipohjaisessa radioliityntäverkkoarkkitehtuurissa. Keilanmuodostajat suunnitellaan keskitetysti, kvantisoidaan ja lähetetään datan mukana tukiasemille käyttäen runkoverkkoyhteyttä. Koska käyttäjiä voidaan palvella usealta tukiasemalta, keilanmuodostussuunnittelu muuttuu ei-konveksiksi kombinatoriseksi ongelmaksi. Toisin kuin olemassa olevissa ratkaisuissa, ehdotettu malli sisällyttää käyttäjien datan lisäksi keilanmuodostajien resursoinnin tarpeen runkoverkkoon. Tukiaseman antennien määrän lisääntyessä, keilanmuodostajien osuus runkoverkon käyttöasteesta kasvaa suureksi. Jotta keilanmuodostajien aiheuttamaa ylimääräistä tiedonsiirtotarvetta voitaisiin minimoida, esitellään kaksi tekniikkaa: kvantisointitarkkuuden muunteleminen sekä lähetykseen käytettävien aktiivisten antennien määrän vähentäminen. Lopuksi, jotta yhdistetyn tila-taajuussuunnittelun aiheuttamaa kompleksisuutta saataisiin vähennettyä, ehdotetaan kaksivaiheista menetelmää. MU-MIMO skedulointialgoritmin avulla etsitään ensin alijoukko käyttäjiä jokaiselle skedulointilohkolle. Esikooderit suunnitellaan vain valituille käyttäjille, mikä vähentää kompleksisuutta, heikentämättä suorituskykyä kuitenkaan olennaisesti. Poiketen nolla-avaruuteen perustuvista tekniikoista, esitetään yksinkertainen vektoriprojektioihin perustuva skeduleri. Kaikkien skedulerien reaaliaikasuorituskykyä on arvioitu toteuttamalla ne ohjelmoitavilla Xilinx ZYNQ-ZC702 system-on-chip (SoC) ja TI TCI6636K2H moniydinalustoilla.

Against Adaptation: Toward Transdisciplinarity and Minor Cinema Linda Petříková Abstract Over the past decades, the field adaptation studies has been trying to break new grounds and escape the confines of the predominant fidelity discourse. This thesis wants to propose new perspectives that have been widely underrepresented, at least in the Anglo-American context, drawing attention to the great relevance to adaptation of the writings of French critical thinkers, most notably Gilles Deleuze and his two-volume publication on cinema and Jacques Rancière and his continuation/reevaluation of Deleuze's film-related concepts. Without directly addressing questions of adaptation, the way both philosophers think about cinema is inseparable from their thinking about literature and indeed about other arts and media, exemplifying new transdisciplinary approaches to adaptation this thesis hopes to encourage. Even though it might seem counterintuitive, considering the efforts of adaptation studies to cut the roots it has grown within literary departments, I chose three Shakespearean adaptations for the case studies as I believe that such focus will enable us to see more clearly the significance of interstices as much as of the links the films form with the text. Jean-Luc-Godard's King Lear, Orson Welles's Chimes at...

The diploma thesis entitled Specific needs of patients during negative pressure wound therapy, consists of two main parts, a theoretical and a practical part. The theoretical part includes findings from research and publications dealing with the issues of wound healing through local application of vacuum therapy, treatment with other methods of wet healing, the impact of long-term non-healing wounds on human needs and the role of the nurse in this scenario. The practical part of the diploma thesis refers to the quantitative research carried out by 46 respondents who were treated with a negative vacuum method, another method of wet healing or a combined form.

This diploma thesis is focused on the study of bioactive wound dressings. During the thesis, hydrogel, lyophilized and nanofiber wound dressings were prepared. Hydrogel and lyophilized wound dressings were prepared on basis of two polysaccharides – alginate and chitosan. Nanofiber wound dressings were prepared by spinning polyhydroxybutyrate. All prepared wound dressings were enriched with bioactive substances, which represented analgesics (ibuprofen), antibiotics (ampicillin) and enzymes (collagenase). Into hydrogel and lyophilized wound dressings were all the mentioned active substances incorporated, whereas nanofiber wound dressings were only with ibuprofen and ampicillin prepared. The theoretical part deals with the anatomy and function of human skin. There was explained the process of wound healing and also there were introduced available modern wound dressings. The next chapter of the theoretical part deals with materials for preparing wound dressings (alginate, chitosan, polyhydroxybutyrate) and with active substances, which were used during the experimental part of this thesis. In the theoretical part, the methods of preparation of nanofiber wound dressings and also the methods of cytotoxicity testing used in this work were presented. The first part of the experimental part of this thesis was focused on preparing already mentioned wound dressings. Then, their morphological changes over time and also the gradual release of incorporated active substances into the model environment were monitored. The gradual release of ampicillin was monitored not only spectrophotometrically, but also by ultra-high-performance chromatography. In wound dressings, in which collagenase was incorporated, was also the final proteolytic activity of this enzyme monitored. The effect of the active substances was observed on three selected microorganisms: Escherichia coli, Staphylococcus epidermidis and Candida glabrata. The cytotoxic effect of the active substances on the human keratinocyte cell line was monitored by MTT test and LDH test. A test for monitoring the rate of wound healing – a scratch test – was also performed.

The aim of the Master thesis is the investigation of the influence of microwave discharge for skin wound healing. Microwave discharge used for this work was argon microwave plasma generated by the surface wave and direct vortex torch. The theoretical part is focused on basic information about plasma and processes that occur in plasma discharge under specific conditions. Plasma generates various active particles such as hydroxyl radicals, nitric oxide radicals, excited nitrogen molecules, atomic nitrogen, argon and oxygen. All of these particles together with plasma generated photons are usable in biomedical applications and summary of them is shown in the theoretical part. The experimental part is focused on the comparison of torch discharge and microwave plasma generated surface wave in skin wound healing. The model wounds on laboratory mousses were treated by plasma and wound healing was examined during 3 weeks after the plasma treatment. Both plasma systems showed healing acceleration. Application of torch discharge was proved to be the most effective method in the healing of skin defects. Additionally, determination of active particles was taken by optical emission spectroscopy. Based on these measurements, plasma parameters were determined: electron temparutare, rotational and vibrational temperatures. To determine role of different plasma active species, the treatment of indigo coloured artificial skin model was treated under various conditions by both plasma systems. Results show that the direct interaction between plasma particles is the main effect, role of radiation, only, is more or less negligible. Finally, the plasma vortex system was visualized using fast camera at selected powers and gas flows.

This diploma thesis was focused on the possibility of using a plasma nozzle to accelerate the wound healing process. The benefits of using low-temperature plasma in medicine or biomedical applications are known from many studies, and low-temperature plasma is already used to sterilize medical devices, materials or surgical instruments. Some studies also report a high potential of usinh plasma nozzle in the treatment of skin wounds. In the experimental part of this work, an in vitro wound healing test was performed using two different low-temperature plasma sources. Source No. 1 was a surface wave microwave discharge and source No. 2 was a torch microwave discharge. An in vitro scratch healing test was performed on a monolayer of HaCaT keratinocytes and testing was performed using various parameters. The influence of the plasma treatment time was monitored, as well as the influence of the plasma discharge power and also the influence of the argon working gas flow. Especially when using a torch microwave discharge, faster wound healing was recorded at most of the parameters used compared to the control. Thus, it can be said that this source appears to be potentially suitable for faster wound healing. Furthermore, in the work using the MTT cytotoxicity test, the viability of skin cells after their plasmination was also monitored using the same conditions as in the in vitro wound healing test. When performed in the standard MTT assay, none of the settings or sources used showed any cytotoxic effects on keratinocytes. LDH cytotoxicity tests were also performed concurrently to verify the accuracy of the MTT assays. The results of both tests agreed and the use of low-temperature plasma in skin treatment can be considered as safe. Overall, the results show that the plasma nozzle can find use in medicine in the healing of skin wounds and chronic defects as a potentially fast, inexpensive and effective method.

Many organizations have realized the importance of managing what they know in a proper way, with the sharing of knowledge as one of the most central aspects. However, the practices of knowledge sharing are seldom fully understood, and thus implementations of technical systems intended to improve knowledge sharing have less effect than expected. This master's thesis project identifies that the case company – the well-known, Swedish telecom giant Ericsson – suffers some potential knowledge gaps regarding its knowledge sharing practices, and the project thus serves the purpose to identify how and why employees indulge in sharing, with emphasis on the informal and employee-driven knowledge sharing.  To understand and find ways to improve the knowledge sharing practices, the thesis project performs a case study in the rapidly expanding Ericsson Cloud RAN project. The qualitative approach of Multi-Grounded Theory is used, to focus on the perceived situation as described by the employees, with previous research as a second grounding-point. The qualitative data is collected through literature analysis, semi-structured interviews, and exploration of the digital platforms and tools internally used to share and document knowledge. The research identifies that the Cloud RAN project needs improvements of the knowledge sharing culture, and create a norm to reuse the documented knowledge. In spite of this, many employees appreciate shared knowledge and contribute to the common good of knowledge. A joy in helping others, personal benefits, and contributions to a greater good drive many employees to share. On the flipside are barriers such as a high bar for contributions from a technical point of view, and a risk of limited reuse of knowledge. As reuse is key to make use of knowledge sharing, the master's thesis report contains identified mechanisms realized in mock-up versions of tools and websites. There, emphasis is put on the importance of having proper tools and access to a contact network to navigate the documented knowledge. Summarized, the findings suggest to utilize the enthusiasm for knowledge sharing among employees for a supportive role, to champion a refined culture and lower the bar to contribute. Frequent reminders of available tools and the impact of sharing what is known in a good way, alongside creating awareness of the direction of the organization, can create company-wide awareness and motivations for improving knowledge sharing.

No description available.

The maturing capabilities of Artificial Intelligence (AI) and Machine Learning (ML) have resulted in increased attention in research and development on adopting AI and ML in 5G and future networks. With the increased maturity, the usage of AI/ML models in production is becoming more widespread, and maintaining these systems is more complex and likely to incur technical debt when compared to standard software. This is due to inheriting all the complexities of traditional software in addition to ML-specific ones. To handle these complexities the field of ML Operations (MLOps) has emerged. The goal of MLOps is to extend DevOps to AI/ML and therefore speed up development and ease maintenance of AI/ML-based software, by, for example, supporting automatic deployment, monitoring, and continuous re-training of models. This thesis investigates how to construct an MLOps workflow by selecting a number of tools and using these to implement a workflow. Additionally, different approaches for triggering re-training are implemented and evaluated, resulting in a comparison of the triggers with regards to execution time, memory and CPU consumption, and the average performance of the Machine learning model.

Bakgrund och syfte. Ordflödestest undersöker en persons förmåga att generera så många ord som möjligt under en minut. Orden kan börja på en viss bokstav eller tillhöra en särskild kategori. FAS är ett vanligt ordflödestest där orden som ska genereras börjar på bokstäverna F, A och S. Exempel på kategoriska ordflödestest är djurflöde, som går ut på att säga så många djur som möjligt, samt verbflöde där instruktionen är att säga så många saker man kan göra (handlingar). Forskare har undersökt vilka mentala färdigheter som ligger till grund för ordflödesförmågan, framför allt planerande och reglerande (exekutiva) funktioner samt språklig förmåga. I forskning används ofta FAS och djurflöde för att undersöka olika delar av den språkliga förmågan. Även verbflöde har undersökts i dessa sammanhang, men inte i lika stor utsträckning. Däremot har verbflöde en större roll i forskning på planerande och reglerande funktioner hos äldre personer som har exempelvis Alzheimers eller Parkinson. Forskningen har lett till att ordflödestest används i kliniska sammanhang som en del i utredningen av dessa sjukdomar. I praktiken används även FAS och djurflöde vid utredningar av dyslexi, språkstörning och koncentrations-/hyperaktivitetssvårigheter (ADHD). Forskning har visat motstridiga resultat om vilka typer av ordflödestest som är nedsatta vid dyslexi och språkstörning, eller vilka mentala förmågor som är viktiga vid genomförande av ordflödestest. I denna studie undersöks FAS, djurflöde och verbflöde hos studenter inom högre utbildning med och utan dyslexi. Syftet är att utreda om ordflödesförmågan är nedsatt hos studenter med dyslexi. Studien undersöker om ordflöde kan bidra till att förklara spridningen i läsförmåga. Metod. I undersökningen deltog 42 studenter, varav 16 hade dyslexidiagnos och 26 kontroller utan dyslexidiagnos. Deltagarna genomförde test som undersöker läsförmåga, fonologisk förmåga, snabb benämning samt ordflöde av FAS, djur och verb. Resultat. Prestationen på ordflödesförmåga var signifikant nedsatt hos studenter med dyslexi jämfört med kontrollgruppen. En multipel regression med bakåteliminering genomfördes för att undersöka om FAS, djurflöde och verbflöde kunde förutsäga spridningen i läsförmåga när fonologisk medvetenhet och snabb benämning kontrollerades för. Regressionsanalysen visade att verbflöde, tillsammans med fonologisk medvetenhet, kunde förutsäga läsförmåga hos studenter med och utan dyslexi. Diskussion. Den nedsatta ordflödesförmågan hos studenter med dyslexi diskuteras utifrån faktorer som utbildning och andra mentala förmågor. Resultatet pekar på ett unikt samband mellan verbflöde och läsförmåga hos studenter med och utan dyslexi. De strukturer i hjärnan som aktiveras vid verbflöde är även strukturer som ligger till grund för andra mentala förmågor. Dessa mentala förmågor har även visats vara nedsatta hos personer med dyslexi. Generellt indikerar dessa nya fynd att verbflöde har en betydelse i förhållande till läsförmåga och dyslexi som behöver undersökas vidare. Resultatet diskuteras även utifrån ett kliniskt perspektiv. / Verbal fluency is commonly measured in cognitive assessments and has been shown to measure aspects of verbal ability and executive function, as well as to involve specific cortical areas during performance. Verbal fluency tasks, in which participants generate words during a given time limit, have been used in research and assessments of neurobiological disorders and impairments. Dyslexia is a neurobiologically based reading disorder that is characterized by difficulties in word decoding and spelling. Research on verbal fluency in individuals with dyslexia shows that semantic and letter fluency is impaired. However, studies show inconsistent results. This study examines performance on semantic fluency (animals), action fluency (verbs), and letter fluency (FAS) in 42 students with developmental dyslexia (DD, n = 16) and a control group with typical reading development (TD, n = 26). Participants also perform a test battery that measures reading and phonological abilities, amongst others. Additionally, it is examined if verbal fluency performance can contribute to predicting reading ability, when phonological awareness and rapid automatized naming (RAN) are taken into account. Results show that verbal fluency performance was impaired in the DD group, and that action and letter fluency were relatively more impaired than semantic fluency. A backward elimination regression showed that action fluency and phonological awareness were significant predictors of reading ability, together explaining 48 % of the variance. The impaired verbal fluency ability is discussed in relation to factors such as education and cognitive abilities. Further, the findings point to a possible unique connection between action fluency and reading ability in students, in addition to phonological awareness. The possibility that the relationship between action fluency and reading may be partly explained by common neurocognitive underpinnings is discussed. These novel findings indicate that action fluency has a pertinent role in reading ability and dyslexia, which should be further examined.