21684 research outputs found
Sort by
Leveraging machine learning and computer vision for advanced UAV communications
Get PDFIn the rapidly developing field of wireless communication, there is a growing demand for technologies that can provide flexible deployment, extended coverage, and enhanced performance in next-generation networks. Traditional networks often struggle with high mobility and environmental blockages, highlighting the need for innovative solutions like Unmanned Aerial Vehicles based (UAV-based) dynamic base stations. UAVs offer a promising solution by functioning as dynamic base stations in 5G and 6G networks, with the potential to address these challenges and improve communication reliability and efficiency.
However, the integration of UAVs into wireless communication presents significant challenges. Ensuring reliable communication in high-mobility environments, optimizing beam management techniques, predicting blockages in real time, and managing the latency inherent in UAV-assisted networks all require innovative solutions. These challenges are combined by the need to balance power consumption and processing capacity, especially when performing complex tasks such as on-device machine learning and computer vision-based beamforming.
The first study of this dissertation focuses on the challenge of beam management in milimeter wave (mmWave) 5G and beyond networks, where speedy environmental changes in highmobility scenarios degrade signal quality. Previous studies have highlighted the limitations of traditional beamforming approaches, especially in their ability to adjust to dynamic environments. To enhance this, a novel technique is proposed that integrates computer vision (CV) with ensemble learning, employing the "you look only once" (YOLO-v5) for precise UAV detection and positioning. By stacking two neural networks to refine a meta-learner, this method achieves approximately 90% top-1 accuracy in K-beam predictions, significantly enhancing the signal-to-noise ratio and improving network performance in high-mobility scenarios.
The second study focuses on the problem of proactive blockage prediction and management in mmWave communications, where maintaining line-of-sight connectivity is necessary. Previous studies have stated that traditional reactive handover methods often result in service disruptions due to unexpected blockages. Computer vision techniques used previously resulted to a 40% improvement in user connectivity by predicting and managing blockages. Extending this concept, the study addresses proactive blockage prediction and management in mmWave communications, employing UAVs not only as base stations but also as proactive agents in handover processes. By leveraging CV to detect potential blockages and monitor user movement, the system facilitates proactive handovers to maintain line-of-sight connectivity. This approach, evaluated using a publicly available dataset and incorporating advanced antenna modeling techniques, has demonstrated a 20% enhancement in network performance.
The third study reveals a new approach that utilizes vision-aided machine learning for efficiently and precisely predicting the optimum beam orientations for UAVs using mmWave and terahertz (THz) technologies. Previous research has shown that, while utilizing visual data from UAVs can increase beam prediction accuracy, there are still issues in reducing beam training overhead and managing real-time mobility. Employing data from UAV cameras, the proposed method achieves approximately 90% accuracy in predicting the best beam direction for the top-1, and nearly 100% for the top-3. Performing these computations directly on the UAV (on-device inference) reduces communication delays by 15% and lowers the cost of communication by 50% in terms of power consumption in comparison with ground-based processing, greatly increasing the efficiency of real-time UAV communication. Collectively, these studies underline the potential of using UAVs to improve wireless communication providing innovative solutions for network expansion, precise beam management, and proactive blockage prediction.
This thesis emphasizes UAVs as a cornerstone technology for advancing future wireless communication, setting the stage for more reliable, efficient, and comprehensive communication systems
Parameterised complexity of spreading problems on temporal graphs
Get PDFSpreading problems are a class of decision problems on graphs that can be used to model the behaviour of a spread of a contagion over a network. Real world networks are often dynamic in nature, with the connections between the members of the network changing over time. Take for example the contact network of a population of people: individuals may come into contact for a period of time, and then move apart and contact other individuals. Graphs do not model this dynamic behaviour, and as such in this thesis we consider spreading problems on temporal graphs, which overcome this shortfall by augmenting a graph with temporal information. In particular we consider the problems of Temporal Firefighter and Temporal Graph Burning, extensions of the Firefighter and Graph Burning problems to temporal graphs. Temporal Firefighter asks how to best prevent the vertices of a temporal graph from “burning” when a fire is spreading over the graph, and Temporal Graph Burning asks how best to burn the vertices of a temporal graph to spread a fire as fast as possible. Both of these problems are NPcomplete, and unlikely to yield efficient algorithms in general. Parameterised complexity theory provides tools for obtaining efficient algorithms for NP-complete problems, and in this thesis we consider various parameters for temporal graphs. We find that both Temporal Firefighter and Temporal Graph Burning are in FPT when parameterised by vertex-interval-membership-width and by temporal neighbourhood diversity. Furthermore, we prove a meta-theorem for showing that a temporal graph problem is in FPT when parameterised by vertex-interval-membership-width. Overall we demonstrate the usefulness of temporal graph parameters, suggesting future work in applying these to more problems on temporal graphs
A study on identity (re)construction of Chinese students studying in higher education in the UK
No full textAbstract not currently available
Descriptors and predictors of symptom experience, information needs and caregiver burden in Thai patients with advanced lung cancer and their family caregivers during palliative radiotherapy
Get PDFIntroduction: Patients with advanced lung cancer undergoing palliative radiotherapy often face significant both symptoms from lung cancer and side effects from radiotherapy, which affect their quality of life and require effective self-management strategies. Family caregivers play a crucial role in supporting patients throughout their cancer journey, necessitating a comprehensive assessment of their information needs and understanding of symptom management. Research on symptom experiences, information needs, and caregiver burden among patients receiving palliative radiotherapy is limited, particularly within the context of Thai populations. This thesis aimed to explore and quantify the extent of the symptoms experienced by Thai patients with advanced lung cancer, the information needs of these patients and their family caregivers during palliative radiotherapy, and the caregivers’ burden, and also to identify predictors that influence symptom self-management, meeting information needs and affecting caregivers’ burden.
Methods: The thesis employed an observational, correlational, repeated measures design, which used the Memorial Symptom Assessment Scale (MSAS); the information needs subscale from the Supportive Care Need Survey (SCNS) for patients and caregivers, and the Zarit burden interview to gather information about symptom prevalence, frequency, and severity and the distress caused by these symptoms, information needs, and the level of caregivers’ burden. Participants included Thai patients diagnosed with advanced lung cancer and their family caregivers at four time points: before palliative radiotherapy (visit 1), weekly during palliative radiotherapy (visits 2 and 3), and a month post-palliative radiotherapy (visit 4). Quantitative data collection utilises standardised symptom assessment tools to quantify symptom burden, and demographic and clinical factors are examined as potential predictors.
Findings: Patients with advanced lung cancer (n=56) undergoing palliative radiotherapy and (n=56) of their caregivers were included. The most prevalent symptoms that patients reported were lack of energy, pain, cough, weight loss, and “I don’t look like myself”. Symptom prevalence peaked during the last week of treatment (Visit 3). The most frequent symptom across all four visits was fatigue. Symptom scores exhibited a consistent reduction from baseline at all visits. Most caregivers reported either no burden or little burden. The most frequent concern of caregivers was related to the apprehension about their relative’s future. Burden scores indicated a trend of a decrease from baseline. Both patients and caregivers sought information primarily from healthcare providers, with discussions being the preferred information format. For both patients and caregivers, wanting information about managing fatigue was the most frequently identified information needed across all visits. Patients’ information needs regarding self-help methods for recovery decreased from baseline across the visits. Caregivers' information needs concerning complementary and alternative therapies decreased from baseline in all categories. Symptom experience scores were associated with smoking history, gender, and radiotherapy type. Psychological symptom scores were higher in married patients with a smoking history. Symptom experience scores were associated with smoking history (p=0.027), gender (p=0.011), and radiotherapy type (p=0.037). Moreover, a smoking history was indicative of higher Global Distress Index scores. Gender and relationship status influenced increased caregivers’ burden. Smoking history, age, treatment area, education level, and radiotherapy dose were predictors of heightened information needs. Gender and relationship status influenced caregivers’ information needs. Male caregivers had a bigger reduction in information needs than females (p=0.041), and spouses/partners had a bigger information need than those with a different relationship to the patient (p=0.021). These findings provide valuable insights into the symptom experiences and information needs of patients during palliative radiotherapy for lung cancer and caregivers’ burden and caregivers’ information needs during taking care of these patients offering potential directions for tailored interventions and care strategies. Caregivers also experience their own challenges while supporting patients, highlighting the need for tailored information and support strategies for caregivers.
Conclusion: The results relate to Thailand but cannot be generalised to all patients and caregivers in Thailand due to the specific characteristics of the study sample and healthcare settings. This study highlights the importance of systematic symptom-monitoring to identify which symptoms may persist that may require more intense supportive care intervention. The caregivers report a low burden level, but fostering open communication channels between caregivers and healthcare providers can help address any concerns to ensure that they feel supported throughout the patient’s treatment journey. The findings have implications for the development of patient-centred interventions that aim to enhance symptom self-management e.g. x and provide targeted information to both patients and their family caregivers. By addressing the identified predictors, healthcare professionals can better support the unique needs of patients and caregivers, ultimately improving the quality of care and patients’ quality of life. Tailored interventions, if implemented, have the potential to address these complex needs and enhance the well-being of patients with advanced lung cancer who are receiving palliative radiotherapy
The regulatory effect of glucocorticoids on perineuronal nets occurs through different mechanisms: potential relationship to schizophrenia
Get PDFPerineuronal nets (PNNs) are extracellular matrix structures surrounding mainly parvalbumin (Pv)-expressing γ-aminobutyric acid (GABAergic) interneurons, providing several cellular or neural functions during the brain developmental period, such as maintaining cellular or synaptic connections, regulating neural plasticity, controlling the closure of critical period and protecting neurons from being damaged by external substrates. Disrupted expression of PNNs and PNN components could result in brain dysfunction, and could be observed in various brain disorders, including schizophrenia. Schizophrenia is a psychiatric disorder, affecting approximately 1% of the population worldwide. Patients diagnosed with schizophrenia exhibit positive, negative and cognitive symptoms. With regard to the aetiology of schizophrenia, genetic and environmental risk factors are associated with schizophrenia, and prenatal maternal stress is robustly detected as an environmental risk factor, roughly doubling the likelihood of the condition in offspring. How risk factors, particularly environmental stress, are associated with schizophrenia is still unclarified.
PNN and Pv-expressing neurons are one of the neuronal systems involved in the pathology of schizophrenia, with several lines of evidence that abnormalities of PNN structure and Pv expression are observed in schizophrenia. Hence a potential pathway through which stress could increase the risk of schizophrenia is by altering PNN and Pv formation or expression. There is some evidence that stress could alter the general expression and formation of PNNs and PNN components, including chondroitin sulfate proteoglycans (CSPGs) (aggrecan, brevican, neurocan, versican and phosphacan), hyaluronan synthase (Has) and link proteins (Haplns), and tenascin R (TnR). Apart from PNNs, evidence demonstrated disrupted Pv expression after stress exposure, and abnormal PNN and Pv expression, with disturbed density and intensity of staining, is observed in schizophrenia patients. Additionally, glutamate decarboxylase (Gad), a critical factor contributing to GABA synthesis, is consistently demonstrated to be disturbed in schizophrenia subjects in previous studies, supporting disrupted GABA neurotransmission in schizophrenia.
In this case, disrupted Pv, Gad and PNN components influenced by external stress might be associated with the increased risk of schizophrenia, which could be a potential pathway underlying the aetiology of schizophrenia. However, the alterations of Pv and Gad expression by stress remained controversial, and whether the expression of PNN components is affected by stress is not fully investigated. Thus, the current study aimed to investigate the effect of glucocorticoids (GCs) – likely mediators of the effects of prenatal maternal stress on the foetus -on PNN component genes, Pv and Gad expression, and further to investigate whether the altered expression was associated with schizophrenia-like changes.
In primary cultured mouse cortical neurons, the data from the current study reported that GCs could alter the gene expression of specific PNN components, including, versican (Vcan), hyaluronan synthase 1 and hyaluronan synthase 3 (Has1, Has3), Hapln4 and TnR and Pv. Altered gene expression was detected primarily at the mRNA level, with corresponding protein changes proving harder to detect. However, altered structural properties of PNNs in the cultures were detected following GC exposure, using Wisteria floribunda agglutinin (WFA) staining. Reduced length of PNNs covering neuronal dendrites was observed, indicating that expression and formation of PNNs were affected by GCs. These results confirmed the hypothesis that over-exposure to stress could disrupt the expression of some of PNN components and suppress the PNN structure.
GCs proved to be able to modify the expression of a number of different PNN component genes, but this was evident primarily at 7 days in vitro (DIV), and also at 14 DIV, but not at 21 DIV, suggesting a major modulatory effect early in development as PNNs are forming. Interestingly, the effects on PNN structure were detected at 21 DIV as well as 14 DIV, a time when no effects on PNN gene expression were observed, suggesting a distinct mechanisms of action.
The precise mechanisms involved in these GC actions appeared diverse, but difficult to identify. The suppressive effects of GCs on Vcan, Hapln4 and TnR mRNA expression appeared to occur through a non-genomic pathway, as they were rapid (detectable within 4h) and not reproduced by the mineralocorticoid agonist aldosterone, or blocked by the GR antagonist mifepristone. The possibility of a post-transcriptional action to accelerate mRNA decay was tested, but no evidence was obtained in support of this idea. The suppressive effects on Has1, Has3 and Pv were also rapid, but were sensitive to mifepristone, suggesting mediation through GRs. Mifepristone alone affected the expression of some PNN component mRNAs, including aggrecan (Acan), brevican (Bcan), neurocan (Ncan), and Has1, implying that basal (non-stressed level) GCs in the culture medium were exerting effects on these genes.
The changes in protein expression that were detected, including GC-induced deceases in the levels of Has2 and Gad65, and increased levels of Has3, TnR and Gad1 induced by mifepristone alone, appeared to occur within 4h, and in the absence of any corresponding changes in mRNA levels. The possibility that GCs might be modulating the activity of the proteasome was considered. Decreased activity was observed after GC exposure that was rapid and specific for the chymotrypsin-like activity. While this novel action cannot explain the GC-induced reduction in Has2 protein, it could contribute to the ability of mifepristone to increase PNN component protein levels in the absence of any mRNA changes. Further work should assess the ability of mifepristone alone to affect proteasome activity in these cultures.
To examine the expression of PNNs and Pv in the brain from mouse models of aspects of schizophrenia, an overview of the distributions of PNNs and Pv in various brain regions is needed. In brain sections from mice expressing Td-tomato from the GABAergic interneuron specific Nkx2.1 promoter, PNNs and Pv-expressing interneurons were largely distributed in cortical layers2/3 and layers 4/5, in prefrontal cortex (PFC) and retrosplenial granular cortex. No sex differences in PNN expression were detected. PFC is an important cortical region regulating executive functions and associated with various abnormal behaviours in schizophrenia. Since duplications of chr.16p11.2 are one of the strongest genetic variants associated with schizophrenia, Pv, PNN component and Gad expression gene expression was assessed in wild-type mice and 16p11.2 duplication mice, both without and with an environmental risk factor manipulation, specifically, maternal immune activation (MIA). Similar changes in Pv and CSPG component expression were observed in MIA-exposed adult offspring in both conditions with dams or littermates as experimental units. The expression of CSPG genes was elevated by MIA, as opposed to the decreases observed with GC exposure in vitro. No overt effects of the genetic manipulation were observed, nor interactions with the MIA.
Taken together, the data presented in the current study suggested that GCs could suppress expression of Pv, several PNN component genes, and PNN morphology, and the altered Pv and CSPG expression were also caused by MIA. A remarkable diversity of GC effects on PNN component gene expression were observed, acting on multiple targets, utilizing a number of different mechanisms, and with clear developmental regulation. The results indicated that external and prenatal stressors might be associated with the increased risk of schizophrenia by disrupting PNN and Pv expression. However, to fully investigate the underlying mechanisms and their relationship to the pathology of schizophrenia, these alterations in the function of PNNs and Pv need to be further investigated
Modelling and characterisations of millimetre-wave InP HEMT transistors
No full textAbstract not currently available
Advancing sensitivity and performance of Capacitive Micromachined Ultrasound Transducers (CMUTs) for medical imaging
Get PDFUltrasound transducer technology has significantly contributed to advancements in medical imaging, continuously improving resolution, efficiency, and integration. Capacitive Micromachined Ultrasound Transducers (CMUTs) use electrostatic transduction to generate and detect sound waves, promise broader bandwidth, smaller size, and better integration with electronics than piezoelectric transducers, which improves Transduction efficiency and frequency response for next-generation imaging such as X-radiation (X-rays), Magnetic Resonance Imaging (MRIs), or standard photography.
This thesis focuses on improving CMUT sensitivity and performance for medical imaging through innovative design strategies, analytical modelling and optimized fabrication processes. The literature review covers ultrasound imaging principles, CMUT operation, and their use in diagnostics, therapy, biosensing, and airborne systems, while a comparison with PMUTs emphasizes CMUTs’ improved electromechanical coupling and frequency response, supporting their role in advanced imaging. A fabrication process compatible with Complementary Metal Oxide Semiconductor (CMOS) is developed to support the integration of CMUTs with advanced electronic circuits, enabling the creation of compact and efficient imaging devices.
The research takes an experimental approach to improve a low-temperature sacrificial release method, enabling precise membrane formation with reduced stress. Various microfabrication techniques, including photolithography with photomask design using COMSOL, thin-film deposition, and etching, are refined to develop a scalable and reproducible fabrication process. To enhance CMUT sensitivity, the study uses analytical modelling with Hooke’s Law and EulerBernoulli Beam Theory to analyse axial and bending stiffness in straight beams, then compares them to meander beams to assess their effect on CMUT sensitivity.
Experimental validation through capacitance-frequency (C-F) and capacitance-voltage (C-V) measurements confirms reliability, showing the Silicon Oxide (SiO₂) sacrificial layer remained stable without causing capacitance loss. Additionally, the measured permittivity of the sputtered SiO₂ closely aligns with values reported in previous studies, demonstrating consistency in material properties and fabrication precision.
In summary, this research has established a CMOS-compatible fabrication process and refines the dry etching release method to reduce stiction, laying the groundwork for advancing CMUT technology with improved reliability and adaptability for medical imaging. Future work aims to improve CMUT fabrication by adjusting material choices and enhancing the sacrificial release process
Beyond boundaries: Unveiling human-drone proxemic dynamics using Virtual Reality
Get PDFSocial drones—autonomous unmanned aerial systems operating in inhabited environments—are a rapidly developing transformative technology. While they promise significant benefits, their successful integration into everyday life depends not only on technological advancements but also on their harmonious incorporation into people’s environments. Individuals intuitively navigate their surroundings, adjusting their distances from both people and non-living entities. This raises a critical question: How will the integration of social drones affect these subtle spatial dynamics? Inappropriate spacing can lead to discomfort, stress, and even defensive reactions, such as evasive maneuvers or attacks. Therefore, before drones are deployed in inhabited spaces, it is essential to understand their spatial relationships with people—an area of study known as human–drone proxemics (HDP). Despite the importance of this issue, significant gaps remain. First, there is a lack of theoretical frameworks for interpreting human–drone proxemics. Second, there is an absence of valid methodological approaches, largely due to constraints in real-world studies. To address these gaps, this research aims to: 1) provide researchers with essential interpretive tools by establishing a solid theoretical foundation for human–drone proxemics, and 2) develop an effective approach for studying these behaviors by investigating Virtual Reality (VR) as a promising alternative to real-world studies. Following an extensive literature review on proxemics, Human–Drone Interaction (HDI), and VR, we devised a course of action. Through five user studies conducted in virtual environments specifically designed for studying proxemics, we evaluated the applicability of four frameworks that explain people’s spatial relationships with drones. Each study included theoretical grounding, empirical assessments, drone design considerations, and concrete guidelines for adopting these frameworks. Our findings reveal that distancing behaviors with external entities are shaped by various motivations—including goal oriented actions, protective instincts, social appropriateness, and arousal regulation. These motivations, influenced by how individuals perceive sensory information, often conflict, prompting physical, environmental, or cognitive adjustments to reconcile competing desires, such as the urge to approach the drone while maintaining distance. These insights culminated in a model of human proxemics with external entities, grounded in human–drone proxemics but extendable to other entities, integrating diverse motivations, highlighting their interactions, and offering new insights into the sensory processes underlying these behaviors. It equips researchers with a framework to better motivate, predict, and interpret findings in HDP studies. Additionally, our research developed a practical understanding of VR as a methodological tool for exploring HDP. VR proved to be a powerful approach, offering significant advantages over constrained real world studies. We formulated a methodological protocol and practical guidelines to equip researchers with the tools they were lacking, enabling more effective exploration of human–entity proxemics. These contributions not only fill critical gaps in the field but also provide a robust foundation for future research, fostering cumulative knowledge development in HDI
Developing a synthetic bone marrow niche for testing novel leukaemia therapies
Get PDFHaematopoietic stem cells (HSCs) are responsible for haematopoiesis, the continuous production of blood and immune cells within the bone marrow (BM) niche. The BM niche maintains the stem cell pool and haematopoietic activity with a variety of stimuli. Erroneous haematopoiesis can result in disorders and diseases such as acute myeloid leukaemia (AML). Chimeric antigen receptor (CAR) T-cell therapy has the potential to significantly improve AML patients’ outlook. However, the efficacy of novel therapies is difficult to assess prior to clinical trials. A model BM niche was developed to provide insight into therapies’ efficacies.
The model used mesenchymal stromal cells (MSCs), which support HSCs in the BM niche, as a feeder layer. The MSCs were cultured on a surface coated with poly(ethyl acrylate). This material caused fibronectin, which the surface was also coated with, to assemble in an open conformation. The fibronectin, which was loaded with the osteogenic growth factor bone morphogenic protein 2, directed the phenotype of MSCs cultured on top. A synthetic, peptide-based hydrogel was also layered on top of the MSCs, forming a barrier between MSCs under the gel and HSCs cultured on top of the gel, and acting as a culture interface between the different cell populations. This system promoted a niche-like phenotype in MSCs, which supported HSCs, mimicking the BM niche.
A CAR T-cell therapy which specifically eliminated AML cells was also developed. AML cells lack differentiable surface markers that can be used as CAR T-cell targets without harming healthy myeloid cells. To overcome this, CD33, a myeloid and AML cell surface marker, was disrupted in HSCs. This allowed CD33 to be targeted with CAR T-cells, theoretically eliminating all CD33 expressing cells, including AML cells, while a CD33del population of haematopoietic cells with the potential to repopulate the niche survived. This therapeutic approach was tested in the model BM niche and found to effectively eliminate AML cells, but also affect haematopoietic cells, warranting further research.
The successful modelling of a healthy, diseased, and treated BM niche demonstrated the usefulness of in vitro models for testing novel therapies for BM associated disorders and diseases
Data sources:
Access Repository Dashboard
Do you manage Open Research Online? Become a CORE Member to access insider analytics, issue reports and manage access to outputs from your repository in the CORE Repository Dashboard! 👇