[kand] Sähkötekniikan korkeakoulu / ELEC

Permanent URI for this collectionhttps://aaltodoc.aalto.fi/handle/123456789/13

Browse

Recent Submissions

Now showing 1 - 20 of 3865
  • Radiaattoriverkoston tasapainotus sähköisillä toimilaitteilla asuinkiinteistössä
    (2026-04-02) Sereda, Petja
    School of Electrical Engineering | Bachelor's thesis
    Radiaattoriverkostojen tasapainotus perustuu perinteisesti mitoitusulkolämpötilassa laskettuihin huonekohtaisiin lämmitystehoihin, virtaamiin ja esisäätöarvoihin. Laskennalliset arvot eivät varsinaisesti huomioi auringon, ihmisten tai sähkölaitteiden tiloihin tuomaa lämpökuormaa. Ylimääräisiä lämpökuormia varten radiaattorit varustetaan patteritermostaateilla, joiden tarkoituksena on rajoittaa lämmitystä osakuormatilanteissa. Toimilaitepohjaisessa tasapainotuksessa pyritään säilyttämään huonelämpötilat jatkuvasti asetusarvossaan. Esisäätöarvojen laskemisen sijaan toimilaitteella haetaan venttiilin asento, jossa virtaus ja radiaattorin lämmönluovutus säilyttävät halutun huonelämpötilan. Verkostosta voidaan huoneantureiden ja toimilaitteiden lähettämän datan avulla löytää huoneet ja säätöpiirit, jotka eivät saavuta asetusarvojaan ja ovat täten epätasapainossa. Työssä perehdytään radiaattoriverkoston toimintaan, verkoston tasapainoon vaikuttaviin tekijöihin sekä varsinaiseen tasapainotustyöhön. Lisäksi vertaillaan perinteisen ja toimilaitepohjaisen tasapainotuksen eroavaisuuksia sekä huonekohtaisen lämmityksen säätöpiirejä. Toimilaitepohjaista tasapainotusta voidaan pitää ratkaisuna, jossa aiemmin perinteisesti tasapainotetusta verkostosta voidaan löytää ongelmalliset säätöpiirit ja keskittyä vain niiden tasapainon löytämiseen. Näin patteriventtiilien säätötyötä erillisillä huonelämpötilamittauksilla ja oikeiden esisäätöarvojen hakemisella ei tarvitse toteuttaa koko kiinteistöön. Nämä kaksi tasapainotustapaa eivät poissulje toisiaan, vaan niitä voidaan hyödyntää myös yhdessä. Sähköisten toimilaitteiden yhdessä huonelämpötila-anturien kanssa tuottama data auttaa myös huoltohenkilöstöä ja kiinteistönomistajaa jatkossa yksittäisten vikatilanteiden selvittämisessä. Sähköisillä toimilaitteilla patteriventtiileitä voidaan jatkossa ohjata erilaisten ohjausalgoritmien avulla tarkemmin ennustettujen lämpökuormien tai tilojen käyttöprofiilien mukaan.
  • Role of Ca2+ and calcium channels in age-related macular degeneration
    (2026-03-23) Fierens, Elin
    School of Electrical Engineering | Bachelor's thesis
    Age-related macular degeneration (AMD) is one of the leading causes of irreversible vision loss among the elderly, affecting visual acuity and detail perception. Although the disease process remains unclear, impaired function of the retinal pigment epithelium (RPE) is considered an early and critical factor in the development of AMD. Recently, disturbances in calcium homeostasis have been linked to impaired RPE-function, thereby potentially contributing to AMD. Through a literature review, this thesis discusses how calcium signalling and calcium channels regulate key physiological processes in RPE cells and how disruption in these mechanisms may be involved in AMD. The concentration of calcium is regulated either by release form intracellular calcium stores or by transportation through different types of calcium channels in the plasma membrane of the RPE, including voltage-gated calcium channels, transient receptor potential channels and ligand-activated ion channel. Disruptions in calcium signalling involve abnormal calcium influx, impaired intracellular storage, and disrupted channel activity. Disruptions in calcium signalling can affect the response to oxidative stress, the molecular signalling pathways linked to autophagy and inflammasomes, as well as the secretion of vascular endothelial growth factor. In summary, calcium signalling plays a crucial role in maintaining RPE function, and disruptions in this signalling appear to contribute to several pathological mechanisms involved in AMD. A better understanding of calcium-dependent signalling pathways may provide new insights into the pathogenesis of AMD and contribute to the identification of new treatment approaches.
  • Design and evaluation of ultra-low-power 433 MHz transmitter for batteryless IoT sensor nodes
    (2026-03-20) Koponen, Aarno
    School of Electrical Engineering | Bachelor's thesis
    The increase in the number of Internet of Things (IoT) devices has brought challenges to the recycling and production of batteries. In addition, typical IoT sensors face congestion in the 2.4 GHz frequency band. These challenges have created a need for more sustainable batteryless sensor technologies based on energy harvesting. This Bachelor’s thesis examines the suitability of ultra-low-power Sub-GHz radio circuits for IoT sensors that use energy harvesting as a power source. The study consists of a literature review and experimental measurements. The literature review compares frequency bands, modulation schemes, and radio protocol energy efficiency. Additionally, the thesis examines typical IoT sensor architecture and compares existing off-the-shelf Sub-GHz transmitters. In the experimental section, the power consumption and reliability of S2-LP radio is measured for varying voltage and transmission power levels. Transmitter efficiency and energy per bit was calculated from the measurement data. According to the results, commercially available Sub-GHz radios, such as the S2-LP, are suitable for battery-free IoT sensors, provided that the startup energy required and the transmission current are low enough.
  • Investigation of array signal processing for resilient satellite-based navigation
    (2026-02-24) Mikkonen, Matias
    School of Electrical Engineering | Bachelor's thesis
    Intentional jamming and spoofing, in addition to unintentional interference, challenge the use of global navigation satellite system (GNSS) signals. Although array processing can improve the interference resiliency of radio systems compared to single-antenna receivers, export regulations and implementation complexity have slowed its wider adoption. A recent regulatory change in the United States removed interference- resilient civilian arrays from the munitions list, easing deployment of the technology in a broader range of applications. Furthermore, next-generation GNSS constellations will employ higher carrier frequencies, enabling more elements in arrays of comparable physical size. This thesis surveys known array signal processing methods for GNSS receivers and evaluates a subset of them experimentally. From the literature, spatially constrained power-minimization methods were found suitable when satellite signal directions of arrival are known in the array frame, as they preserve the desired signals without distortion. Techniques based on the known satellite transmission waveforms may risk locking onto spoofed signals while blind techniques make the array spatial phase response unstable. The experimental part uses Jammertest 2024 data collected with a four-element planar array to assess two blind mitigation techniques. The results show that even a small array can significantly improve interference-resilience and that selected techniques can be integrated with existing GNSS receiver architectures.
  • Liike- ja lämpöenergian kerääminen ja hyötykäyttö henkilöautossa
    (2026-02-16) Huhtala, Santeri
    School of Electrical Engineering | Bachelor's thesis
    Kandidaatintyön tavoitteena on selvittää, kuinka moottoriurheilussa, erityisesti Formula 1 -sarjassa, kehitettyä hybriditeknologiaa voidaan soveltaa henkilöautojen voimansiirrossa. Työssä tarkastellaan polttomoottorin hyötysuhteen parantamista hyödyntämällä kineettisen energian (MGU-K) ja lämpöenergian (MGU-H) talteenottojärjestelmiä. Työssä vertaillaan kilpa-autojen ja siviiliautojen teknisiä ratkaisuja ja niiden rajoitteita. Keskeisiä haasteita henkilöautoissa ovat akkujen lataus- ja purkusyklien hyötysuhde sekä polttomoottorin lämmönhallinta erityisesti kylmäkäynnistystilanteissa. Työssä esitellään ratkaisumalleina muun muassa transmoottoripohjainen voimansiirto, integroidut lämmöntalteenottojärjestelmät sekä lämpösähköiset generaattorit. Vaikka Formula 1 -tason komponentit ovat usein liian kalliita suoraan hyödynnettäviksi, niiden taustalla olevat innovaatiot energian kierrätyksessä ovat kriittisiä henkilöautojen polttoaineenkulutuksen ja päästöjen vähentämiseksi. Tulevaisuuden näkymissä korostuvat uudet materiaalit ja uusiutuvan energian, kuten aurinkopaneelien, integrointi osaksi hybridijärjestelmää.
  • Vyöryfotodiodien karakterisointi fotoluminesenssikuvantamisen avulla
    (2026-01-14) Goumaras, Stefanos
    School of Electrical Engineering | Bachelor's thesis
    Työ on kirjallisuustutkimus siitä, miten fotoluminesenssikuvantamista (PL-kuvantamista) voidaan käyttää vyöryfotodiodien (APD) karakterisoinnissa ja erityisesti laadunvalvonnassa ennen hitaita sähköisiä testejä. Tässä kandidaatintyössä käydään läpi puolijohdefysiikan perusta energiavyö rakenteista, seostuksesta, suorasta- ja epäsuorasta kaistaväli, varauksenkuljettajan generaatiosta ja rekombinaatiosta, PN-liitoksesta ja vyöryläpilyönnistä. Kandidaatintyö esittelee PL-kuvantamisen periaatteen ja spektrisen tulkinnan: mitä emissiotyyppejä spektristä etsitään (vyöltä vyölle, defektit, kenttävaikutteiset komponentit) ja mitä ne kertovat materiaalista ja prosessista. Työ käsittelee menetelmän rajoitteet, painottaen piipohjaisten rakenteiden ongelmaa: heikko radiatiivinen rekombinaatio, ei-radiatiivisten prosessien (SRH/Auger) signaalia vaimentava ja tulkintaa sekoittava vaikutus sekä se, ettei syvällä syntyvä emissio välttämättä pääse pinnalle. Johtopäätös on, että PL-kuvantaminen on perusteltu, kontaktiton ja ei-tuhoava esivalidointimenetelmä APD-rakenteille: sen avulla voidaan paikantaa ja luokitella poikkeamia (defektikeskittymät, dopanttivariaatiot, jännityskeskittymät) ja lisäksi tunnistaa ennenaikaisia vyörylaukaisupisteitä hyödyntämällä vyöryprosessiin liittyvää heikkoa luminesenssia (piissä tyypillisesti vaatii herkän NIR-detektorin).
  • Methods of single-cell RNA sequencing data analysis
    (2025-12-10) Lehtonen, Anssi
    School of Electrical Engineering | Bachelor's thesis
    A description of the abstracted analysis pipeline, reviewing some of the most used open access tools in the field currently. Includes theoretical observations about clustering and data integration.
  • User experience with AI-based chatbots
    (2025-12-28) Sameer Saral, Ishaan
    School of Electrical Engineering | Bachelor's thesis
    Artificial intelligence (AI) and large language models (LLMs) have transformed digital interaction and modern websites, with AI-based chatbots becoming widely used across e-commerce, education, healthcare, and customer service. This thesis conducts a state-of-the-art literature review to examine how these chatbots influence user experience (UX). The findings show that AI chatbots enhance efficiency, provide instant responses, improve accessibility, and deliver personalized interactions through context awareness, and domain-specific adaptation. Additionally, psychological factors such as trust, perceived usefulness, perceived ease of use, emotional engagement, and anthropomorphic design play a major role in influencing user satisfaction and long-term usage. However, there are also limitations with these chatbots such as hallucination defined as fictional or incorrect responses. Technical issues like latency and crashes also happen. Chatbots can also lack the most up-to-date information. Distrust and concerns around data privacy and ethical transparency are also a problem for many users that require security. The thesis concludes that effective chatbot design requires balancing technical performance with relational quality. It also highlights key areas for future research, such as multimodal interfaces, hallucination reduction, privacy-preserving personalization, and improved dialogue evaluation metrics. These improvements can ameliorate trust and reliability for users.
  • Benchmarking cell-type and gene expression deconvolution methods in spatial transcriptomics
    (2025-12-12) von Konow, Elsa
    School of Electrical Engineering | Bachelor's thesis
    The importance of cellular composition is apparent in cancer, where the disease progression is influenced by intercellular communication and spatial properties of the cells, such as their location. Cell-type deconvolution is a computational alternative to single-cell RNA sequencing that is both cheaper and less demanding in terms of data preprocessing. It can be used to estimate the cell-type composition of a biological sample or the cell-type-specific gene expression when an estimate is sufficiently accurate for the research purpose. In this bachelor's thesis, six deconvolution methods (BayesPris, Bisque, CIBERSORTx, PRISM, SpatialDecon, xCell) are benchmarked using both bulk RNA sequencing data, which represents the average gene expression of a sample, and spatial transcriptomic data, which preserves the spatial organization of cells. The best performing methods for estimating cell-type proportions from spatial data were PRISM, BayesPrism and SpatialDecon. PRISM and BayesPrism performed best in estimating cell-type proportions and cell-type-specific gene expression from bulk RNA sequencing data. These methods excelled in estimating the proportion of malignant epithelial cells, but differences emerged in the estimation of rarer immune cell-types and in modeling immune-derived gene expression. In this respect, BayesPrism produced more accurate results than PRISM.
  • Exploring user similarity metrics: A review across interaction, text, and user profile domains
    (2025-12-08) Määttä, Thuy
    School of Electrical Engineering | Bachelor's thesis
    The rapid growth of online communities attracts variety of research especially user similarity metrics, a key component for understanding the relationships between users in online environments. A number of metrics exist to determine user similarity, from traditional metrics, such as Jaccard coefficient or cosine similarity, to more complex metrics. However, in the modern online platforms, the relationships between users have become increasingly complex, which reveals the limitations in the traditional metrics. This thesis studies on different existing user similarity metrics through an overall literature review. The metrics are grouped into three categories based on interaction, text, and profile information of the users. Each category is examined in terms of its definitions, strengths, and limitations. The study concludes that each of these categories is suitable for analyzing the similarity relationships between users on certain platforms. Furthermore, it finds that combining metrics from different categories can benefit the user similarity results. Finally, the thesis highlights the needs of an application of the metrics on an unified platform, such as Reddit, and an evaluation of their computational efficiencies in order to conclude a better comparison. Additionally, it calls for more research and a proposal of a new hybrid metric which can yield a more accurate similarity score and perform well in a sparse dataset.
  • Implementation of federated learning on microcontrollers
    (2025-12-12) Halttunen, Kauri
    School of Electrical Engineering | Bachelor's thesis
    Federated learning is a decentralised method of machine learning. It enables ML where data movement and aggregation are infeasible due to sensitive content or data transmission limitations. The advantages of distributed learning vary based on the specifics of the implementation, hardware, design goals, and other details. Thus, the inherent decentralisation of the IoT creates a suitable opportunity for FL. This thesis aims combine the disciplines of microcontroller programming and machine learning to produce a working example of FL and evaluate the associated challenges. The study utilised a freely available human activity recognition dataset gathered from eight subjects to train multiple independent neural networks. To resemble a practical deployment of microcontrollers, every worker received and trained on data from one subject. The neural networks were merged to create a global model using a simplified variant of FedAvg. The training loss and accuracies of the independent models are compared with the global model, which is also tested with new subject data to simulate new entrants to the FL system.
  • LLM-based multi-agent architecture for transport network automation using MCP and A2A protocols
    (2025-12-11) Dinh, Tue
    School of Electrical Engineering | Bachelor's thesis
    Large language models (LLMs) have recently been used as the brains behind many AI agents, which can reason and act in natural language. While single-agent systems have demonstrated their potential in automating tasks, their robustness is bounded in fields with complicated behaviors. In this case, multi-agent systems (MASs) provide a natural extension. They enable agents to collaborate in groups to achieve the goal. Transport network management is a suitable domain for application. It includes coordinated workflows such as network monitoring, fault localizing, and trouble ticketing. However, current MAS implementations often rely on custom frameworks and manual connections. This limits the interoperability and modularity. To address this challenge, this thesis proposes an MAS architecture that leverages two recent open standards: the Model Context Protocol (MCP) and the Agent-to-Agent (A2A) protocol. The former protocol standardizes agent-to-tool interactions and the latter one supports agent-to-agent communication. Together, these protocols enable a framework-neutral plug-and-play approach to integrate agents, tools, and services. A prototype of this architecture was designed and implemented on a representative transport network automation controller. Its effectiveness was evaluated on key performance indicators such as task success rate, error isolation, modularity, and LLM token cost. The results demonstrate that an open standard-based MAS architecture can improve the interoperability and scalability of autonomous network operations. Beyond the prototype, the thesis aims to contribute a reference design pattern to use such LLM-based MAS in high-stakes automation domains.
  • LLM agent for autonomous CoAP-based IoT systems
    (2025-12-12) Nguyen, Duc Tung
    School of Electrical Engineering | Bachelor's thesis
    IoT is one of the core technologies that enables "Industry 4.0" and facilitates people's work in various domains. Due to their software implementation with statically defined functions, traditional IoT systems can solve tasks that developers have previously anticipated, implemented, and deployed. However, IoT systems require complex processes to extend their capabilities, and they struggle with context-aware decision-making. Recent advances in large language models and agentic AI frameworks have created opportunities for developing autonomous AI agents that emerge as a possible solution to the problem. Therefore, this thesis investigates the architecture, implementation, and performance of an LLM agent in a CoAP-based IoT system. The architecture consists of an LLM agent developed with the SmolAgents library, with GPT-4o as the brain of the agent, CoAP-specific tools to enable interaction with the IoT environment, and memory components to enhance the performance of the agent. Experiments with a simulated smart home CoAP-based IoT system are conducted to evaluate the performance of the agent in response to user requests. The results demonstrate the potential of the LLM agent architecture in consistently and accurately solving both explicit and implicit user requests. Although there are several limitations remaining, such as the range of LLMs used, the lack of experiments with real-world IoT systems, and the absence of advanced AI-related methods, this research lays a foundation for applying an LLM agent in CoAP-based IoT systems.
  • Vision-language-action models for humanoid robots in elderly care applications
    (2025-12-05) Saarinen, Jaakko
    School of Electrical Engineering | Bachelor's thesis
    This thesis explores Vision-Language-Action models and their potential applications in humanoid robots designed for elderly care. The study reviews how multimodal learning, large language models, and embodied AI contribute to natural and adaptive human-robot interaction. The research is based on a comprehensive literature review focusing on integration methods, training approaches such as imitation and reinforcement learning, and ethical considerations. The findings highlight that VLA-based humanoids can enable more personalized and intuitive interaction, potentially alleviating future labor shortages in elderly care. However, challenges remain regarding data requirements, computational costs, and societal acceptance.
  • Avoimen lähdekoodin ohjelmistoprojektien käyttö tieteellisessä tutkimuksessa
    (2025-12-19) Jääskeläinen, Veikko
    School of Electrical Engineering | Bachelor's thesis
    Avoimen lähdekoodin ohjelmistojen käyttö ja kehitys tutkimuskäytössä on yleistynyt. Yli 70 prosenttia tutkijoista sanoo, ettei heidän tutkimuksensa tekeminen olisi mahdollista ilman heidän käyttämiään ohjelmistoja. Ohjelmistoprojekteissa viitataan yhä useammin julkaistuun tutkimukseen, ja vastaavasti tutkimuksissa viitataan entistä useammin ohjelmistoprojekteihin. Tässä kandidaatintyössä selvitetään, miten avoimen lähdekoodin ohjelmistoprojektit sopivat tutkimuskäyttöön. Työ on toteutettu kirjallisuuskatsauksena. Työssä on esitelty, mitä ominaisuuksia avoimen lähdekoodin ohjelmistoprojekteilla on, kuten ohjelmiston lisenssi sekä projektin versionhallintaohjelma ja dokumentaatio. Työhön on koottu, mitä lisäpiirteitä avoimen lähdekoodin ohjelmistoprojekteilta vaaditaan tutkimuskäytössä. Esimerkiksi arkistointi- ja viitetiedot ovat vaadittuja lisäpiirteitä. Näiden lisäksi työssä on tarkasteltu tutkimuskäytön erityisvaatimuksia. Tutkimuskäytössä on tärkeää, että ohjelmiston lisenssi sallii käytön tutkimuksessa ja että projektin dokumentoinnissa on huomioitu projektin taustalla oleva tieteellinen tutkimus.
  • From pilot to profit: A conceptual framework for AI-driven design automation
    (2025-12-12) Peltonen, Rudolf
    School of Electrical Engineering | Bachelor's thesis
    The rapid advancement of artificial intelligence (AI) technology has compelled enterprises and organizations to reconfigure their operational models. Despite substantial resource allocation toward AI implementation, a significant proportion of projects lack the anticipated results. Recent reports indicate that AI solutions fail at a rate nearly double that of IT projects from a decade ago. However, different factors have been identified between failed pilot initiatives and solutions that successfully scale to production. The reasons for failure are mainly strategic, including ambiguous objectives, poor data quality, and friction in workflow integration. While general-purpose solutions for AI implementation exist, frameworks in the context of design automation are scarce. The objective of this thesis is to address this research gap by developing a conceptual framework for AI development within the context of design automation. The study is a literature review that synthesizes causes of failure, critical success factors, and theoretical aspects key to success, such as requirements engineering and data management. The literature review revealed that in addition to technical capabilities, success depends on the ability to integrate solutions into existing workflows. Based on the analysis, the resulting Pilot-to-Profit framework is structured around four primary concepts: Goal Granularity, Data Maturity, Task-Technology Fit, and Human-AI Collaboration. The framework provides a systematic model for bridging the so-called GenAI Divide and developing productive solutions. This model can be utilized to evaluate the automation potential of design work and the viability of implementing AI solutions.
  • Analys av LED-drivdon med effektfaktorkorrigering
    (2025-12-08) von Knorring, Casimir
    School of Electrical Engineering | Bachelor's thesis
    I takt med att LED-belysning ersätter konventionella ljuskällor såsom glödlampor och fluorescerande lysrör, ökar även behovet av effektiva och tillförlitliga drivdon. Dessa är anpassade för att garantera en jämn strömtillförsel till LED-armaturer och på så sätt förbättra livslängden, effektiviteten samt styrbarheten hos lysdioder. Detta arbete analyserar funktionen och strukturen hos LED-drivdon som är baserade på switch-mode-omvandlare, med särskilt fokus på hur de upprätthåller en stabil och effektiv strömtillförsel till lysdioder. Inledningsvis presenteras den teori som anknyter till drivdonets funktion, vilket utgör grunden för den efterföljande, mer fördjupade analysen. Genom att analysera olika switch-mode-omvandlare och deras roll vid strömreglering belyses hur drivdonen formar och anpassar den inkommande spänningen efter lysdiodens behov. Arbetet redogör även för effektfaktorkorrigeringens betydelse, särskilt med avseende på hur ett integrerat effektfaktorkorrigeringssteg kan minska harmonisk distorsion och därigenom förbättra effektfaktorn. Vidare behandlas drivdonets övergripande struktur samt olika omvandlartopologiers inverkan på drivdonets prestanda, kostnad och komplexitet. Arbetet innefattar en produktanalys av en utvald produkt som innehåller ett LED-drivdon. Analysen ger en konkret och tillämpbar förståelse för hur de behandlade principerna realiseras i en faktisk utformning av ett LED-drivdon. Genom denna helhetsöversikt klargörs de centrala principerna bakom moderna LED-drivdon samt de avvägningar som krävs för att uppnå hög energieffektivitet och lång livslängd.
  • Generatiiviset tekoälymallit uusien syöpälääkkeiden kehittämisessä
    (2025-12-10) Tirri, Miranda
    School of Electrical Engineering | Bachelor's thesis
    This bachelor's thesis is a literature review that examines how artificial intelligence can be used to accelerate the development and implementation of cancer drugs. The study investigates what types of AI models can be used to evaluate potential cancer drug candidates, as well as how these models have already been applied in practice. In this work, we review several generative AI models that have been used in pharmaceutical research. We also examine which stages of the drug development process these generative models support. These stages include molecular design, target molecule identification, and optimization of drug candidates. Furthermore, this thesis compares the advantages of AI-assisted cancer drug research with traditional pharmaceutical research. These advantages include speed and cost-efficiency.
  • Puolijohdekvanttipisteitä käyttävät virusten tunnistusmenetelmät
    (2025-12-18) Huotari, Rebekka
    School of Electrical Engineering | Bachelor's thesis
    Virus detection, especially of pathogenic viruses, is important for individuals and for societies. For individuals, it is important to get precise treatment at the right time. Getting the right treatment also prevents the virus from spreading. The prevention of spreading and the right treatment cuts the indirect costs caused by viruses. Traditional virus detection usually requires a laboratory and trained personnel to do the detection. There is still room for improvements of the detection methods. One possibility would be to use semiconducting quantum dots as part of virus detection methods. Quantum dots are semiconducting nanoparticles with excellent optical properties. Electrical conductivity can also be used as an advantage. Several studies have shown that virus detection can be enhanced using quantum dots. In these studies, precision, specificity, cost efficiency and usage outside laboratories have been enhanced. This bachelor’s thesis makes an introduction to the structure of quantum dots and manufacturing them. Virus detection in general and with quantum dots are also addressed in this literature research.
  • Optical properties of nanostructured glass
    (2025-12-12) Hurnanen, Santeri
    School of Electrical Engineering | Bachelor's thesis
    This bachelor's thesis focuses on nanostructured glass solar concentrators. The technology is based on a nanoscale layer manufactured on the glass surface with a femtosecond laser, which aims to direct part of the incident light hitting the device to solar panels installed at the edges of the glass. In particular, this thesis examines the effect of laser parameters on these properties. In this study, spectrometry measurements were performed on eight samples. Based on the measurement results, the work presents the absorption, reflection, and transmission spectra of nanostructured glass at wavelengths of 300–1800 nm. Additionally, the work compares nanostructured glass to other key transparent solar concentrators. Based on the comparison, nanostructured glass has several promising features. Unlike luminescent solar concentrators (LSC), nanostructured glass has very stable properties, and its performance does not deteriorate over time. The manufacturing of devices is also inexpensive, as no materials other than glass are needed for production, and adding laser processing to current building glass manufacturing processes does not cause major difficulties. Furthermore, nanostructured glass is non-flammable, which is important for materials used in construction. However, the measurement results of this study show that the implementation of nanostructured glass also involves significant challenges. The device's absorption spectrum, on which the device's efficiency depends, decreases very rapidly as the laser power used to process the glass decreases. On the other hand, the transmittance of glass manufactured with higher laser power decreases, while the glass reflectance increases. In addition, nanostructures manufactured with higher laser power also make the glass hazy. It follows that the technology's biggest challenge is low efficiency, particularly compared to other similar technologies. Nanostructured glass has useful properties, but weak efficiency and haziness limit the technology's use in its current form.