41530 research outputs found
Sort by
Detecting Software Anomalies in Microservices Using Spectrograms, Scalograms, and Convolutional Neural Networks
No full textThe growing complexity of cloud-native microservices has intensified the need for robust anomaly detection mechanisms, particularly for misconfiguration-induced failures that often manifest subtly and propagate across service boundaries. This thesis proposes a comparative framework for runtime anomaly detection by transforming system metrics into frequency-domain representations—spectrograms using Short-Time Fourier Transform and scalograms using Continuous Wavelet Transform—and classifying them with Convolutional Neural Networks (CNNs). Evaluated on DeathStarBench, a production-grade microservices benchmark with injected anomalies in the Reservation and Recommendation services, the framework also includes baseline Long Short-Term Memory (LSTM) models trained on raw time-series features. Stratified 5-fold cross-validation shows that CNNs trained on scalograms outperform all other models, achieving 98.54% accuracy, 98.53% F1-score, and a Matthews Correlation Coefficient (MCC) of 0.9628, compared to 84.76% accuracy and 85.10% F1-score for the best LSTM model (128×8). Scalograms also deliver superior class-wise F1-scores for both anomalies—0.9451 for A1 and 0.9276 for A2—highlighting their strength in capturing bursty and slowly varying patterns often missed by fixed-window spectrograms and temporal baselines. These findings support the hypothesis that wavelet-based representations offer enhanced sensitivity and robustness for detecting misconfigurations in dynamic, resource-constrained microservice environments, paving the way for future research in multi-resolution anomaly detection and real-time telemetry analysis
Peripheral Vagus Nerve Stimulation (PVNS) Device for Epilepsy
Get PDFOur research team developed a PVNS for Epilepsy device which is intended to act as a preventative measure against seizures for patients with epilepsy. The device is a less invasive alternative to the VNS device on the market using electroacupuncture at the Shenmen and Xin points along the posterior pinna to stimulate the vagus nerve
Degradation Characterization of Nopal-based Biopolymer made from Opuntia sp. Juice, Beeswax, Animal protein, and Polyvinyl Alcohol
Get PDFSince the introduction of polymers in the 1960s, there has been an exponential increase in use. Driven by rapidly growing demand, the volume of polymers in the production, application, and end-of-life stages has surged. The widespread use of these materials has led to escalating environmental pollution, which disrupts ecosystems, and contributes to global ecological degradation. While advancements in recycling technologies and infrastructure have shown promise, they have not maintained pace with polymer production and disposal. This imbalance warrants research into sustainable alternatives. Environmentally benign polymers derived from renewable sources and exhibiting minimal ecological impact at the end of their life cycle represent a critical area of research in addressing this global challenge. Dra. Sandra Pascoe Ortiz has developed a promising natural biopolymer derived from the juice of the Opuntia cactus. This biopolymer is non-toxic, fully biodegradable in natural environments, and can be produced with minimal ecological footprint. This study investigates the effects of incorporating a water-soluble synthetic polymer, polyvinyl alcohol (PVOH), into Opuntia-based biopolymer. The research focuses on the mechanical properties and degradation behavior of the resulting composite material in a marine environment
FLAIR: A Fog-Native Workload Placement Framework for QoS-Aware Scheduling in Smart Manufacturing
Get PDFSmart manufacturing environments utilize a variety of heterogeneous IoT sensors that continuously stream large volumes of data. To detect and mitigate critical events such as fires or accidents in real time, this data must be promptly analyzed by machine learning (ML) image classifiers. These classifiers are typically deployed on fog computing nodes located within the manufacturing warehouse facilities to ensure low-latency responses and to protect data privacy. Although fog nodes offer a better alternative to remote cloud data centers for such latency-sensitive tasks, they are often resource-constrained and highly sensitive to the composition of workloads, particularly in heterogeneous environments where multiple types of classification tasks run concurrently on the same fog node. To address this challenge, this thesis proposes FLAIR (Fog Layer Architecture with Intelligent Routing), a fog-native workload placement framework that ensures latency-based Quality of Service (QoS) requirements are met for real-time classification tasks. FLAIR integrates a data-driven predictive model for estimating the response time of new tasks, a Digital Twin mechanism that simulates co-location scenarios to assess resource utilization, and a lightweight, threshold-based algorithm that automatically places new classification tasks on suitable fog nodes to ensure that they meet QoS requirements. Experimental evaluations on an AWS-based testbed show that FLAIR consistently outperforms traditional Queuing Network Models (QNM), particularly in heterogeneous deployments, reducing the mean absolute percentage error (MAPE) from 39.56% to 6.43% and avoiding task placements that would otherwise violate QoS constraints. These results underscore FLAIR’s effectiveness in supporting latency-critical ML tasks within the dynamic and resource-limited fog infrastructure of smart manufacturing environments
Optimization of Downy Mildew Management and Baby Kale Production on the Central Coast of California
No full textBaby kale (Brassica oleracea and other species) is a vital commodity as a salad green in spring mix and is widely produced on the Central Coast of California. Downy mildew, caused by Hyaloperonospora brassicae, is a devastating disease infecting economically important crops like cabbage, broccoli, cauliflower, horseradish, rapeseed, kohlrabi, Brussels sprouts, and kale. This foliar disease causes plant tissue to exhibit chlorotic and necrotic flecking and gray-white sporulation, affecting both the yield and quality of the crop and rendering the affected leaves unmarketable. The Central Coast’s cool climate and nighttime moisture, combined with dense plantings and sprinkler irrigation, make the region highly conducive for downy mildew infections. Downy mildew management primarily relies on chemical applications for disease prevention. However, the crop’s quick harvest time and the pathogen’s rapid polycyclic cycle and spore dissemination, paired with limited fungicide modes of action, present challenges for disease control.
Additionally, growers on the Central Coast face strict regulations limiting water and nitrogen inputs to protect natural resources and ecosystems while sustaining agricultural production for years to come. The Sustainable Groundwater Management Act was implemented to manage groundwater and prevent aquifer depletion caused by rising temperatures and intensified drought events in California. Ag Order 4.0 was put into effect, limiting nitrogen application amounts to address water quality and run-off issues. These limitations make it difficult for agriculture production to meet yield demands. However, plant breeding approaches can combat these obstacles by identifying and exploiting favorable genotypes to produce superior baby kale varieties with downy mildew resistance or high water and nitrogen use efficiency (WNUE).
This study focused on a collection of 212 accessions of baby kale. First, three rounds of baby kale host resistance screenings were conducted against eight downy mildew isolates collected from eight distinct locations across the Central Coast of California to identify sources of downy mildew resistance and facilitate resistance breeding in baby kale. The first screening test all 212 baby kale accessions against one downy mildew isolate (BKG22). The second screening tested 50 accessions against seven isolates. Finally, 25 accessions were tested against 4 downy mildew isolates across three replications. Artificial inoculation was performed by spraying a sporangia suspension onto baby kale plants, which were incubated in a humidity chamber. Disease severity was assessed by examining both surfaces of each leaf for chlorotic and necrotic symptoms and sporulation and then quantified using an established rating scale. Screening of all 212 accessions revealed an average disease severity of 28%, with severities ranging from 0 to 100%. The initial subset screening showed average disease severities ranging from 2.2 to 9.4%, depending on the isolate. The final subset screening demonstrated a range of 0.003 to 0.072% average disease severity among the four isolates, with 13 accessions exhibiting 100% estimated resistance probability, 11 accessions between 99.0 and 99.9%, and one accession below 99.0%. These results suggested that downy mildew could be effectively managed in baby kale through resistant varieties.
Following the downy mildew resistance screening, a subset of 46 baby kale accessions showing reduced disease susceptibility were assessed for water and nitrogen use efficiency. The study was conducted under greenhouse conditions, and all baby kale plant accessions were treated with two water and nitrogen treatment levels. Plants were harvested after approximately six weeks, and fresh weight was recorded. Accessions were also individually dried and encapsulated for isotope analysis. Screening of the 46 accessions revealed broad genetic variation among the three species, with species significantly affecting all measured variables under both water and nitrogen treatments. Among species under 50% inputs, B. napus maintained high performance, ranking first for yield, WUE, and PFP under 50% nitrogen and first for yield, WUE, and NUP under 50% water. Nitrogen treatment significantly affected all variables, with greater variation observed between the treatments. Notably, B. napus × B. oleracea accessions consistently ranked second across all measured variables under both nitrogen levels. These findings highlight the genetic capacity within Brassica spp., demonstrate the potential of interspecific crossbreeding, and suggest that breeding strategies can help growers remain productive while adapting to evolving regulations and constraints.
While collecting downy mildew isolates for disease resistance screening, a constantly high disease incidence was observed in nurseries. It was hypothesized that such management failure was caused by fungicide resistance as a result of frequent fungicide exposure. Therefore, a study was conducted to evaluate the impact of prior fungicide exposure on fungicide resistance in H. brassicae by comparing disease severity between high-exposure nursery isolates against presumed low-exposure isolates collected from a school garden and breeding plot. H. brassicae isolates were evaluated for their sensitivity to five fungicides through artificial inoculation of baby kale cotyledons treated with low and high concentrations of each fungicide. Results revealed distinct patterns in fungicide efficacy. Fluopicolide completely inhibited infection caused by the low-exposure isolates, while infection caused by high-exposure isolates was only partially inhibited. All isolates displayed varying levels of insensitivity to mefenoxam. Infections caused by each of the four isolates were completely inhibited by ametoctradin + dimethomorph, mancozeb, and fluoxapiprolin. Patterns observed with fluopicolide indicated that repeated exposure likely contributed to resistance development. The insensitivity to mefenoxam across all isolates indicated that resistance was likely to increase with prolonged use of chemicals. However, the lack of resistance to mancozeb, one of the multi-site modes of action fungicides used in the nurseries, suggested that factors other than fungicide resistance may significantly contribute to the ineffective control of downy mildew in the nursery settings