Premise plumbing, the pipes and fixtures at the building level, present a unique challenge for maintaining drinking water quality. Of particular concern are opportunistic pathogens, including Legionella pneumophila which can regrow in premise plumbing and cause disease in immunocompromised populations. The goal of this work was to explore engineering methods for control of L. pneumophila and total regrowth. The first line of study involved a series of experiments with simulated glass water heaters (SGWHs) to investigate interactions between specific water chemistry factors and L. pneumophila regrowth, and the second used laboratory grade purified water to investigate the limits of a nutrient control approach for biological stability. Several water chemistry factors including assimilable organic carbon (AOC) content, granular activated carbon (GAC) biofiltration, plumbing materials, copper concentrations and temperature were investigated using SGWHs. AOC is the carbon available for bacteria growth in drinking water. Results indicated that AOC reduction may be a promising method for controlling L. pneumophila and total bacteria regrowth, but there may be a point at which AOC reduction is no longer effective. Prior GAC biofiltration removed organic carbon and was effective in controlling total bacterial regrowth in SGWHs, but actually encouraged L. pneumophila regrowth. A wide variety of materials typically encountered in premise plumbing was investigated and only had limited effect on proliferation of L. pneumophila and total bacteria. The effects were dynamic, even with long-term studies. Copper pipes held promise for control of L. pneumophila, as did copper concentration across a range of pHs. Aqueous copper concentration released from pipes was dependent on temperature, however, and thus this control method may not be applicable in all hot water lines. The peak temperatures for L. pneumophila proliferation fell between 41 and 45 °C, temperatures which could be encountered in a hot water distribution system when the water heater is set to 48 °C, as is often recommended with scalding and energy concerns. A constant temperature of 53 °C seemed to provide control of L. pneumophila, but recolonization is possible even at these high temperatures. Work with laboratory grade water indicated that extreme control of nutrients was not enough to completely control regrowth in premise plumbing. With stagnation in the cleanest conditions, a 2-log increase of a diverse group of bacteria was observed within 10 days. As drinking water can never achieve such nutrient removal, this study presents the limits of nutrient removal as a strategy for regrowth control. This work explored both the potential and the limitations of several mechanisms for controlling regrowth in premise plumbing. Understanding how these water chemistry factors affect L. pneumophila and total bacterial regrowth is critical to identifying the most effective engineering controls. / Master of Science

Nitrification is increasingly of concern in US potable water systems, due to changes from chlorine to chloramine as a secondary disinfectant in order to comply with new regulations for disinfectant by-products. The ammonia that is released from the chloramine decay supports nitrification. A comprehensive literature review systematically examined the complex inter-relationships between nitrification, materials corrosion and metals release. That analysis suggested that nitrification could accelerate decay of chloramine, enhance corrosion of water distribution system materials, and increase leaching of lead and copper to potable water under at least some circumstances. Moreover, that certain plumbing materials would inhibit nitrification, but that in other situations the plumbing materials would enhance nitrification. Experiments verified that nitrification could affect the relative efficacy of chlorine versus chloramine in controlling heterotrophic bacteria in premise plumbing. Without nitrification, chloramine was always more persistent and effective than chlorine in controlling biofilms. But with nitrification and in pipe materials that are relatively non-reactive with chlorine, chloramine was much less persistent and less effective than chlorine. In materials that are reactive with chlorine such as iron pipes, the relative efficacy of chloramine versus chlorine depends on the relative rate of corrosion and rate of nitrification. High rates of corrosion and low rates of nitrification favor the use of chloramine versus free chlorine in controlling bacteria. Plumbing materials had profound impacts on the incidence of nitrification in homes. Effects were due to toxicity (i.e., release of Cu⁺²), recycling of nitrate back to ammonia substrate by reaction (zero-valent iron, lead or zinc materials), or release of nutrients that are essential to nitrification by leaching from concrete or other materials. As a general rule it was determined that concrete and iron materials encouraged growth of nitrifiers in certain oligotrophic waters, materials such as lead, PVC/plastic pipe, glass and surfaces of other materials were readily colonized by nitrifiers, and materials such as copper and brass were very toxic and relatively resistant to nitrifier colonization. Dependent on circumstance, nitrification had no effect, increased or decreased aspects of materials corrosion. Nitrification markedly increased lead contamination of low alkalinity potable water by reducing the pH. In some cases nitrification dramatically decreased leaching of zinc to potable water from galvanized iron, because of lowered dissolved oxygen and reduced pH. Nitrification did not affect copper solubility in low alkalinity water, but is expected to increase copper solubility in higher alkalinity waters. Finally, nitrification in homes plumbed with PVC or plastics can drop the pH and increase leaching of lead from downstream brass materials in faucets. This can explain why some modern homes plumbed with PVC can have more lead in water when compared to homes plumbed with copper pipe. Phosphate had profound impacts on the incidence of nitrification and resulting effects on water quality. While phosphate levels below about 5 ppb could strongly inhibit nitrification due to a nutrient limitation, nitrifiers can obtain sufficient phosphate from plastic, concrete, copper and iron pipe materials to meet nutritional needs. High levels of phosphate inhibitor can reduce the concentration of Cu⁺² ions and make nitrification more likely, but phosphate can also sometimes lower the corrosion rate and increase the stability of disinfectant and its efficacy in controlling nitrifiers. Phosphate plays a key role in determining where, when and if problems with nitrification will occur in a given water distribution system. This work provides some new fundamental and practical insights to nitrification issues through a comprehensive literature review, lab experiments, solubility modeling and field studies. The results and practical tools developed can be used by utilities and consumers to predict nitrification events and resulting water quality problems, and to make rational decisions about practices such as inhibitor dosing, plumbing material selection and use of whole house filters. / Ph. D.

Legionella pneumophila (Lp) and nontuberculous mycobacteria (NTM) are opportunistic pathogens that can be transmitted via drinking water, when tiny droplets containing the bacteria are aerosolized and inhaled during activities such as showering. The resulting respiratory illnesses, Legionnaires' Disease and NTM lung disease, are among the leading sources of drinking water associated disease in the United States and other parts of the world. Lp and NTM are both difficult to control, because they establish as part of natural biofilms that form within the interiors of pipes and fixtures that deliver drinking water to the point of use. These pathogens are especially problematic within premise (i.e., building) plumbing, where intermittent use throughout the day leads to long periods of stagnation, increased water age, warmer temperatures, and depleted disinfectant residuals that exacerbate bacterial growth. The recent advent of high throughput DNA sequencing has led to the discovery that drinking water microbiomes are diverse, complex, and largely comprised of non-pathogenic microbes. This has further led researchers to hypothesize that the microbial ecology of this diverse microbiome could be harnessed as a natural means to control Lp and NTM, i.e., a "probiotic" approach, but such an approach has not yet been demonstrated. The objective of this study was to assess this hypothesis by utilizing biologically active granular activated carbon (GAC) filters, which are already a widely used drinking water treatment both at the municipal and household scale, as a means to naturally shape the microbial ecology of downstream premise plumbing and inhibit Lp and NTM proliferation. GAC has an extremely high surface area that aids removal of organic carbon via adsorption but also provides an ideal habitat for establishment of biofilms, which removes organic carbon from the water via biodegradation. Convectively-mixed pipe reactors (CMPRs) were used for replicable simulation of premise plumbing distal taps. The CMPRs consisted of four-foot-long closed polyvinyl chloride (PVC) pipe segments with the sealed bottom portion resting in a ~48 °C water bath and with the top portion plugged and exposed to the cooler, ambient atmosphere (25 °C in this study), inducing convective mixing and resulting in an internal water temperature of 37 °C. PVC was chosen because it is common in premise plumbing and generally leaches the least organic carbon among the different types of plastic pipe. Four different influent water conditions were implemented in the experimental design: 1) Untreated, dechlorinated municipal tap water with high organic carbon and low biomass; 2) GAC-treated tap water with low organic carbon and elevated, viable biomass; 3) GAC-treated + 0.22-m pore size membrane-filtered tap water to remove both nutrients and biomass; 4) GAC-treated tap water pasteurized at 70 °C with low nutrients and elevated, killed biomass. The 0.22-m pore size membrane filter simulated the use of a building scale particle filter, while pasteurization simulated water passing through a hot water heater at an elevated temperature recommended for pathogen thermal disinfection. To understand the influence of these experimental conditions on older pipes containing mature biofilms versus new pipes that leach more organics and are being freshly colonized, a set of older pipes colonized with mature ~4-year-old biofilms were compared to newly purchased pipes. Each set of pipes was tested in triplicate for the four different experimental conditions with the full volume replaced three times a week for eight months, simulating infrequently used taps containing warm, continuously mixing water thought to create conditions at a very high risk for opportunistic pathogen growth. In the aged CMPR bulk water effluents, droplet-digital-polymerase-chain-reaction measurements showed a one-log reduction of Lp and NTM when receiving GAC-treated or GAC-treated + particle-filtered influent water versus receiving dechlorinated municipal tap water or GAC-treated + pasteurized water. These findings suggest that decreased biodegradable dissolved organic carbon achieved by GAC filtration acted to suppress Lp and NTM growth, while the additional step of biomass removal by particle filtration provided a more modest benefit. In the CMPRs consisting of new pipes, concentrations of Lp and NTMs in the effluent bulk water were similar among the experimental conditions, except that the CMPRs receiving the GAC-treated + particle-filtered influent water experienced a two-log reduction in NTMs. These results demonstrate that the colonization and proliferation of NTM within premise plumbing can be significantly controlled by limiting nutrients and biomass in the influent water. This work demonstrates the potential of harnessing GAC-treatment as a means to Control Lp and NTM in premise plumbing via nutrient removal. In scenarios where chemical disinfectants have been depleted, off-the-shelf GAC-treatment used as point-of-entry treatment to large buildings with recirculating plumbing could provide benefits that have previously been unrecognized. Alternatively, pasteurization in very hot water heaters could provide a short-term disinfection benefit, but eventually the nutrients embodied in the dead biomass undermine the positive influence of the nutrient removal provided by the GAC-treatment. Improved mechanistic understanding of probiotic strategies to opportunistic pathogen control would be needed to overcome inherent limitations to the approaches examined herein, if more effective control is desired in the absence of thermal or chemical disinfection. / Master of Science / Legionella pneumophila (Lp) and nontuberculous mycobacteria (NTM) are bacterial pathogens that are the leading source of drinking water-associated disease in the US. Unfortunately, they are not effectively controlled by protections put in place by the US Safe Drinking Water Act (SDWA). Firstly, they cause respiratory infections, which are spread when tiny droplets of water are inhaled during activities such as showering, whereas the SDWA is specifically designed to protect against ingested pathogens. Secondly, unlike fecal-derived organisms (e.g. E. coli) that are the focus of the SDWA, Lp and NTM grow naturally in drinking water distribution systems, especially in premise (i.e., building) plumbing, where water is warmer and more stagnant. Therefore, even if water leaving the treatment plant is devoid of Lp or NTM, this does not guarantee that the consumer's tap water will be Lp- or NTM-free. Also, even though chlorine or other chemical disinfectant is required by the SDWA to be added to the water leaving the treatment plant to control downstream microbial growth, the disinfectant can be depleted or absent within the premise plumbing itself. Additionally, both Lp and NTM tend to more naturally resist chemical disinfectants than fecal-derived organisms. This research is aimed at overcoming these challenges, opening the door to new approaches to controlling Lp and NTM in premise plumbing. Historically, any microbial growth occurring in drinking water has been viewed as problematic, as it usually indicates the chemical disinfectant is inadequately protecting consumers. However, this work explores whether having an abundant community of beneficial bacteria could improve microbial water quality by competing against pathogens for limited space for attachment and nutrients. Such an approach would be analogous to the use of probiotics in humans, to establish a beneficial gut flora that is less susceptible to pathogen invasion. Granular activated carbon (GAC) filters are often used at drinking water treatment plants and by consumers as a point-of-use (e.g., installed on the kitchen tap or in a refrigerator) or whole-house treatment to remove any contaminants of concern and improve the taste and odor of tap water. The granules within GAC filters have a high surface area that helps remove contaminants, but also provides an environment where microbes can live and thrive. As water enters the filter, beneficial microbes can break down any remaining nutrients in the water (e.g., organic carbon and nitrogen). Additionally, the water leaving the filter carries high levels of microbes that grow on the GAC filter that are shed as water passes through. The resulting water with reduced nutrients and higher concentrations of potentially beneficial microbes could create a competitive environment that alters growth of harmful bacteria, like Lp and NTM, in downstream portions of plumbing. The incoming cold water is also warmed by the building envelope, which increases bacterial growth rates. Thus, the underlying hypothesis of this research is that GAC treatment could provide a combination of reduced nutrients and competitive microbes as water enters downstream premise plumbing and reduce the growth of Lp and NTM. However, GAC-treated water within a building can be further altered by other treatments, like a very hot water heater, which would heat and kill the microbes flowing through it, or a particle filter, which could remove the microbes in the water. This work also seeks to understand how these additional treatments might improve or interfere the nutrient reduction and addition of competitive microbes provided by GAC treatment. This research explores how all these different scenarios affect the growth of Lp and NTM using a lab-scale simulated premise plumbing system constructed out of polyvinyl chloride (PVC) pipe that is a common plumbing material used in homes. Water that was added to the pipes was prepared in four different ways to test the probiotic control hypothesis across distinct experimental conditions that replicate the different influent water scenarios. The four conditions were implemented over the course of eight months with regular chemical and biological analyses conducted to understand the effects of the different influent waters on Lp and NTM. It was discovered that premise plumbing with mature biofilms receiving GAC-treated water or GAC-treated + particle-filtered water contained ~90% less Lp and NTM than premise plumbing receiving non-filtered municipal tap water. However, if the GAC-treated water passes through a water heater, the capacity to limit Lp or NTM growth was lost. While GAC filters are currently thought of as an instantaneous treatment that removes contaminants from water, this work demonstrates how GAC treatment might provide prolonged benefits to water, after it has passed through the filter on its journey to a shower head or faucet. Increased understanding of the exact mechanisms of limited pathogen growth gained by this research can lead to new and effective approaches to protect people from contracting diseases caused by Lp and NTM.

Background: Cloud- and on-premise environments have been compared in terms of security several times. Many of these comparisons based their assessments on qualitative data rather than quantitative metrics. Some recent articles have considered comparing environments by using quantitative data. These methodologies are often complicated and based on incident simulations that might not be relevant in a real-life scenario. Therefore it could be troublesome for a company to evaluate and compare two environments before deciding which environment they would prefer in terms of security. Before an environment migration, it is decisive to know if that environment has been a target for recent cyberattacks. Unfortunately, this data is not available to the public. Objectives: This study aims to provide the reader with an overview of the environmental aspects of the victims of recent cyberattacks. It will reveal what environment cybercriminals have targeted the most. The study will also propose a methodology to compare two environments to each other based on quantitative measurements. The measurements were based on cybersecurity metrics that quantified the threats in each environment. Methods: A structured literature- and dataset review was conducted to find how much each environment had been exposed to cybersecurity incidents. Several expert interviews were held to help explain the findings made in the reviews. A threat analysis was used as the foundation for the proposed comparison methodology. A case study of a recent environment migration was used to test the proposed comparison methodology. Results: The results show that on-premise environments have been more exposed to cybersecurity incidents during recent years than cloud environments. The proposed methodology showed that the cloud environment was the preferred choice in the conducted case study. Conclusions: In recent years, cloud environments have been the preferred choice in terms of security as long as the cloud consumer takes heed to best practices. There is a knowledge gap when it comes to cloud environments. It has been the same for both cloud consumers and cybercriminals. However, according to recent threat reports, cybercriminals have started to improve. Therefore there will likely be more cloud-related incidents in the future. It was determined that the proposed methodology could represent the security posture of each environment. However, a decision should not be based entirely on this methodology because it has not been tested on a large scale. / Bakgrund: Moln- och on-premise-miljöer har jämförts vad gäller säkerhet flera gånger. De flesta jämförelser baserade sina bedömningar på kvalitativ data snarare än kvantitativa mått. Några nya artiklar har jämfört miljöer med hjälp av kvantitativ data. Dessa metoder är ofta komplicerade och baserade på incidentsimuleringar som kanske inte är relevanta i ett verkligt scenario. Därför kan det vara besvärligt för ett företag att utvärdera och jämföra två miljöer innan de bestämmer sig för vilken miljö de skulle föredra vad gäller säkerhet. Innan en miljömigrering är det avgörande att veta om den miljön har varit ett mål för de senaste cyberattackerna. Tyvärr är denna information inte tillgänglig för allmänheten. Syfte: Denna studie syftar till att ge läsaren en översikt av miljöaspekterna hos offren för de senaste cyberattackerna. Det kommer att avslöja vilken miljö cyberkriminella har riktat sig mest mot. Studien kommer också att föreslå en metodik för att jämföra två miljöer med varandra baserat på kvantitativa mått. Mätningarna baserades på cybersäkerhetsmått som kvantifierade hoten i varje miljö. Metod: En strukturerad litteratur- och datasetgranskning genomfördes för att ta reda på hur mycket varje miljö har varit utsatt för cybersäkerhetsincidenter. Flera expertintervjuer hölls för att förklara resultaten som gjorts i granskningarna. En hotanalys genomfördes för att ge underlag för den föreslagna jämförelsemetodiken. Jämförelsemetoden testades i en fallstudie av en nyligen genomförd miljömigrering. Resultat: Resultaten visar att on-premise miljöer har varit mer utsatta för cybersäkerhetsincidenter under de senaste åren än molnmiljöer. Den föreslagna metoden visade att molnmiljön var det föredragna valet i den genomförda fallstudien. Slutsatser: Under de senaste åren har molnmiljöer varit det föredragna valet när det gäller säkerhet så länge som molnkonsumenten tar hänsyn till bästa praxis. Det finns en kunskapslucka när det kommer till molnmiljöer. Det har varit samma sak för både molnkonsumenter och cyberkriminella. Men enligt de senaste hotrapporterna har cyberkriminella börjat kommit ikapp. Därför kommer det troligen att finnas fler molnrelaterade incidenter i framtiden. Det fastställdes att den föreslagna metoden kunde representera säkerheten för varje miljö väl. Ett beslut bör dock inte baseras helt på denna metodik eftersom den inte har testats i stor skala.

Användandet av molnbaserade språkmodeller blir allt vanligare i takt med att de blir mer tillgängliga, och deras kapacitet att bearbeta och generera text förbättras ständigt. Men för användare och organisationer som arbetar med känslig information uppstår en utmaning, eftersom all data som behandlas av dessa modeller överförs till serveroperatören. I och med detta växer behovet av så kallade On-Premisemodeller, som kan köras lokalt på en användares eller organisations egna servrar. Därför ämnar detta projekt till att identifiera, testa och jämföra flera on-premise-språkmodeller med inriktning på textbehandling. Detta inkluderar uppgifter som att upptäcka fel i stavning, grammatik och bruk av skiljetecken, samt att analysera språkanvändning. Projektet innebar att söka efter modeller på hemsidan HuggingFace.co som hanterar både svenska och engelska och sedan implementera dem på olika AWS EC2-instanser. Efter en inledande granskning av flera modeller, identifierades två särskilt lovande alternativ, Viking och GPT-SW3, som var tillräckligt kapabla att genomföra de mer avancerade sorters tester. De utvalda modellerna genomgick en omfattande process av prompt-engineering för att optimera deras prestanda och maximera deras förmåga att demonstrera sin kompetens. Efter detta skapades standardiserade tester för grammatik, stavning och bruk av skiljetecken, samt översättning. Dessutom skapades mer avancerade tester som inkluderade omstrukturering av text, samt förenkling och anpassning av texter till en specifik persona. Resultaten av dessa tester visade att båda prövade modeller hade god förmåga att korrigera fel i stavning, grammatik, och bruk av skiljetecken i texter, samt översättning av texter. GPT-SW3 visade även god förmåga i personatesterna, medan modellen Viking inte var kapabel till att genomföra denna sorts behandling av texter. Parallellt med testandet av språkmodellerna utvecklades ett skript för att analysera hårdvarans prestanda och kostnad under exekveringen av språkmodellerna. Detta skript kördes på flera olika AWS EC2-instanser, var och en utrustad med olika typer av hårdvara. Prestationerna från de olika instanserna jämfördes, och resultaten visade att körning av modellerna på Inferentia2chip gav den bästa prestandan och den lägsta kostnaden per genererat ord, fastän denna hårdvara hade den högsta timkostnaden. Alla CPUinstanser som testades presterade betydligt sämre än GPU-instansen, men var markant billigare att hyra per timme. Detta kan vara en fördel för on-demand-applikationer där kostnadseffektivitet prioriteras. / The use of cloud-based language models is becoming increasingly common as they become more accessible, and their capabilities to process and generate text are continually improving. However, a challenge arises for users and organizations handling sensitive data because all processed data is transmitted to the server operator. As a result, there’s a growing need for so-called On-Premise models, which can be operated locally on the servers of a user or organization. This project aims to identify, test, and compare several on-premise language models, "Focusing on tasks such as detecting errors in spelling, grammar, and punctuation, as well as analyzing language usage. The project involved searching for models that support both Swedish and English on the website Huggingface.co and implementing them on various AWS EC2 instances. After an initial assessment, two particularly promising models, Viking and GPT-SW3, were identified as capable enough for more detailed testing. The selected models underwent an extensive prompt-engineering process to optimize their performance and maximize their ability to demonstrate their capabilities. Following this, standardized tests for grammar, spelling, punctuation, and translation were developed. Additionally,more advanced tests were designed, including restructuring, simplifying, and adapting texts to specific personas. The results from the tests showed that both tested models had high proficiency in correcting errors regarding spelling, grammar, and punctuation, and in translating texts between Swedish and English. GPT-SW3 had high proficiency in rewriting texts to use other styles, such as using formal or simplistic language, while the Viking model had no proficiency in achieving these types of tasks. Concurrently with the model testing, a script was developed to analyze the performance and costs of the hardware during the model executions. This script was implemented on several AWS EC2 instances, each equipped with different hardware types. The performance of these instances was compared, and the results showed that running the models on the Inferentia2 chip offered the best performance and the lowest cost per generated word, despite having the highest hourly rental cost. All CPU instances tested performed significantly worse than the GPU instance but were a lot cheaper to rent per hour. This could be advantageous for on-demand applications where cost efficiency is a priority.

No / Initial ranging is the primary and important process in wireless networks for the customer premise equipments (CPEs) to access the network and establish their connections with the base station. Contention may occur during the initial ranging process. To avoid contention, the mandatory solution defined in the standards is based on a truncated binary exponential random backoff (TBERB) algorithm with a fixed initial contention window size. However, the TBERB algorithm does not take into account the possibility that the number of contended CPEs may change dynamically over time, leading to a dynamically changing collision probability. To the best of our knowledge, this is the first attempt to address this issue. There are three major contributions presented in this paper. First, a comprehensive analysis of initial ranging mechanisms in wireless networks is provided and initial ranging request success probability is derived based on number of contending CPEs and the initial contention window size. Second, the average ranging success delay is derived for the maximum backoff stages. It is found that the collision probability is highly dependent on the size of the initial contention window and the number of contending CPEs. To achieve the higher success probability or to reduce the collision probability among CPEs, the BS needs to adjust the initial contention window size. To keep the collision probability at a specific value for the particular number of contending CPEs, it is necessary for the BS to schedule the required size of the initial contention window to facilitate the maximum number of CPEs to establish their connections with reasonable delay. In our third contribution, the initial window size is optimized to provide the least upper bound that meets the collision probability constraint for a particular number of contending CPEs. The numerical results validate our analysis.

Opportunistic premise (i.e., building) plumbing pathogens (OPPPs) and antibiotic resistant bacteria are emerging microbial concerns in drinking water. OPPPs, such as Legionella pneumophila, are the leading cause of drinking water disease in many developed countries. Contributing factors include the relative success in controlling fecal pathogens, the presence of complex building plumbing systems that create habitats for OPPPs, and the relative resistance of OPPPs to disinfectants, and aging populations that are susceptible to infection. Concurrently, drinking water is increasingly being scrutinized as a potential environment that is conducive to horizontal gene transfer of antibiotic resistance genes (ARGs), selection pressure for enhanced survival of resistant bacteria, and a route of transmission of antibiotic resistant pathogens. While maintaining a disinfectant residual is an established approach to controlling OPPPs in premise plumbing, some studies have indicated that co-resistance and cross-resistance to disinfectants can increase the relative abundances of resistant bacteria and ARGs. Thus, there may be trade-offs to controlling both OPPPs and antibiotic resistance in premise plumbing that call for controlled study aimed at optimizing residual disinfection application for this purpose. A critical review of the scientific literature in Chapter 2 revealed that premise plumbing is a biologically and chemically complex environment, in which the choice of pipe material has cascading effects on water chemistry and the corresponding premise plumbing microbiome. This, in turn, has broad implications for the control of OPPPs, which need to be elucidated through controlled experiments in which worst case premise plumbing conditions are held constant (e.g., warm temperature), while other variables are manipulated. Chapter 3 introduces the convectively-mixed pipe reactors (CMPRs) as a novel low-cost, small footprint approach to replicably conduct such experiments. The CMPRs were demonstrated to effectively simulate key chemical and biological phenomena that occur in distal reaches of premise plumbing. In Chapter 4, the CMPRs were leveraged to study the interactive effects of four disinfectants (chlorine, monochloramine, chlorine dioxide, and copper-silver ionization) and three pipe materials (PVC copper, and iron). The CMPRs were inoculated with two antibiotic-resistant OPPPs: Pseudomonas aeruginosa and Acinetobacter baumannii. It was found that pipe-material (PVC or PVC combined with iron or copper) profoundly impacted the water chemistry in a manner that dictated disinfection efficacy. In Chapter 5, we applied shotgun metagenomic shotgun sequencing to evaluate effects of the combination of pipe material and disinfectant type on the wider microbial community, especially their ability to select for or reduce ARGs. In Chapter 6, we used CMPRs and metagenomic sequencing in a study comparing Dutch drinking water practices to our prior testing in an American system. Dutch drinking water is of interest because of lack of historical use of disinfectants was hypothesized to result in a microbial community that is relatively depleted of ARGs or mobile genetic elements, which can enhance spread of ARGs as disinfectants are applied. Generally, it was found that OPPPs required higher doses of disinfectants for inactivation than the general microbial community, sometimes concentrations approaching the regulatory limits in the US (e.g., 4 mg/L of total chlorine). Even successful reductions were modest, typically ~1-log, and failed to eliminate either P. aeruginosa or A. baumannii. Moreover P. aeruginosa, A. baumannii, and non-tuberculous mycobacteria varied substantially in their preference for pipe material and susceptibility to disinfectants. We found that disinfectants tended to increase the relative abundance of OPPPs, ARGs, and mobile genetic elements. Disinfectants were sometimes associated with net increases in levels of these pathogens and genes when applied at low levels (e.g., 0.1 mg/L of monochloramine), which effectively acted to reduce competition from less resistant and non-pathogenic taxa. When a low dose of monochloramine was applied to PVC CMPRs in the US, we estimated from metagenomic sequencing data that this water contained roughly 100,000 cells per milliliter of taxa known to contain pathogenic members. The Dutch drinking water exhibited more diverse microbial communities and lower relative abundances of taxa containing pathogens. ARGs were two times proportionally more abundant in CMPRs operated in the US without disinfectant than in the corresponding CMPRs operated in the Netherlands. The findings of this dissertation can help to optimize the application of in-building disinfectant addition for addressing concerns related both to OPPPs and antibiotic resistance. The studies herein highlight the necessity of developing comprehensive OPPP and antibiotic resistance control strategies that emphasize not just disinfectant dose, but other key control parameters such as contact time, hydraulics, and temperature. The functional diversity of OPPPs, antibiotic resistant bacteria, and the background premise plumbing microbiome further necessitates broad, holistic programs for monitoring and control. / Doctor of Philosophy / Efforts to provide safe drinking water face two emerging threats: the rise of pathogens that thrive in the plumbing environment that delivers water to the tap and the rise of antibiotic resistance. In the US and many other parts of the world, opportunistic pathogens are the predominant agents responsible for disease spread by tap water. Opportunistic pathogens tend to infect aged or immunocompromised individuals (hence, 'opportunistic') and grow well in in-building plumbing. Globally, antibiotic resistance is on the rise and becoming a fundamental threat to modern medicine. Pathogenic bacteria become resistant to antibiotics used to treat infections when they acquire antibiotic resistance genes (ARGs), which can happen either by mutation or from other resistant bacteria sharing ARGs. Overuse or misuse of antibiotics can impose selection pressure that stimulates horizontal gene transfer and enhance survival of bacteria that are resistant. Prior studies have suggested that under some circumstances, disinfectants used to control pathogens in drinking water can also select for antibiotic resistant bacteria. Thus, the overarching goal of this research was to optimize the type and dose of disinfectant used, depending on building-level factors such as pipe material, for effectively controlling proliferation of both opportunistic pathogens and antibiotic resistance. This dissertation largely focuses on in-building plumbing systems, which are home to potentially tens of thousands of bacterial cells per milliliter of water or per square centimeter of internal pipe surfaces. These bacteria interact not only with each other and other microbes, but also with features of the plumbing environment, such as the water chemistry or the pipe materials. Building plumbing systems are highly intricate ecosystems that can undermine the effectiveness of disinfectants provided by utilities. One major contribution of this research is the development of the convectively-mixed pipe reactors (CMPRs) as a simple and easy-to-use test system that recreates combinations of features of interest encountered in in-building plumbing. We applied the CMPRs to study two common residual disinfectants (chlorine and monochloramine) supplied by water utilities, and two other disinfectants (chlorine dioxide and copper-silver ionization) which are commonly dosed by building operators, especially in hospitals and other buildings housing individuals susceptible to infection. These four disinfectants were applied to CMPRs consisting of PVC, copper, and iron pipe. Chemical, culture, and DNA methods were used to understand how these disinfectants affected the microbes and their ecology. We then took the opportunity to set up CMPRs in the Netherlands, where there has been no historical exposure to chlorine because their water quality regulations emphasize limiting nutrients in the water and elevating the hot water line temperatures as means to control microbial growth. The CMPRs effectively produced worst-case plumbing scenarios, where opportunistic pathogens were especially difficult to control through residual disinfection. Dosed disinfectants tended to be no longer measurable in the water after five hours. The CMPRs also showed that the disinfectant most effective for one pathogen could be the least effective for another. If doses were applied near regulatory limits, the concentrations of pathogens and antibiotic resistance genes decreased. However, opportunistic pathogens tended to survive better than background populations of bacteria. Bacteria carrying ARGs also survived some disinfectant conditions better as well. Thus, if doses were applied at levels that could inactivate some microbes, but not the opportunistic pathogens, pathogen abundances sometimes increased. These results were largely confirmed in the experiment with Dutch drinking water. Here, chlorine appeared to be more problematic than monochloramine in terms of enriching pathogens and antibiotic resistance. We also noted that Dutch waters garnered more diverse microbial communities, with fewer DNA markers for pathogens and antibiotic resistance. In general, this research takes a key step towards optimizing application of residual disinfectants for control of both opportunistic pathogens and antibiotic resistance. Because disinfectants can have negative impacts on drinking water microbial communities when supplied insufficiently, it is important that the other features of in-building plumbing, such as the selection of pipe material or the hydraulics, facilitate disinfectants reaching all portions of plumbing and at the necessary concentrations. It is recommended that the selection process for disinfectant type and dose considers the plumbing materials and other conditions such that disinfection can be aimed towards controlling multiple opportunistic pathogens, which can vary in their susceptibility, and antibiotic resistance.

Consumer exposure to opportunistic pathogens in potable water systems poses a significant challenge to public health as manifested by numerous cases of pneumonia, non-tuberculosis lung disease, and keratitis eye infections. Water utilities have extensive understanding in control of heterotrophic and coliform bacteria re-growth in water distribution systems via disinfection, control of assimilable organic carbon (AOC), and biologically degradable organic carbon (BDOC). However, much little is known about the effect of AOC on the proliferation of heterotrophic bacteria and pathogens within premise plumbing. This thesis is the first systematic examination of opportunistic pathogen persistence and amplification in simulated glass water heaters (SGWH) as a function of influent organic matter concentration. The role of plumbing conditions that may internally generate AOC is critically examined as part of this evaluation. Strong correlations were often observed between influent organic matter and heterotrophic bacteria in effluent of SGWH as indicated by 16S rRNA gene abundance (average R2 value of 0.889 and 0.971 for heterotrophic organisms and 16S rRNA respectively). The correlation was strongest if water turnover was more frequent (every 48-72 hours) and decreased markedly when water changes were less frequent (stagnation up to 7 days). No simple correlations were identified between the concentration of pathogenic bacteria (L. pneumophila, M. avium, A. polyphaga, and H. vermiformis) and AOC, although correlations were observed between M. avium and TOC over a limited range (and only for a subset of experiments). Indeed, there was little evidence that Legionella and Acanthamoeba proliferated under any of the conditions tested in this work. Parallel experiments were conducted to examine the extent to which factors present in premise plumbing (e.g. sacrificial magnesium anode rods, cross-linked polyethylene, nitrifying bacteria, and iron) could influence water chemistry and influence growth of bacteria or specified pathogens. Although these factors could strongly influence pH, dissolved oxygen concentrations, and levels of organic matter (e.g. iron, magnesium, nitrifying), there was no major impact on effluent concentrations of either heterotrophic bacteria or premise plumbing pathogens under the conditions investigated. While additional research is needed to confirm these findings, at present, there is no evidence of correlations between organic matter and pathogen concentrations from SGWH under conditions tested. Substantial effort was also invested in attempting to identify SGWH and oligotrophic nutrient conditions that would consistently support L. pneumophila and A. polyphaga amplification. A review of the literature indicates no prior examples of large scale amplification of these microorganisms at nutrient levels commonly found in synthesized potable water. It is likely that a complex combination of abiotic and biotic factors (i.e. micronutrients, necrotrophic growth, ambient water temperature, disinfectant type and dose, plumbing materials, water usage patterns), which are not yet fully understood, control the amplification and viability of these pathogenic organisms in premise plumbing systems. / Master of Science

The increased usage of cloud computing has transformed modern information technology by providing organisations with a scalable, flexible, and cost-effective alternative to the traditional on-premise service model. Both service models have their own set of advantages and disadvantages. One key aspect both service models have in common is the importance of keeping private data secure. There is an ongoing debate on whether cloud computing is safe enough to store private data. This thesis will help organisations understand the security considerations of the different service models. This will be accomplished through a case study researching the different authentication models of both service models and an experiment to gain further insights. The case study and experiment will conclude with a heuristic that organisations can use when picking an authentication model. The main conclusion of this thesis is that we consider the cloud computing service model less secure than the on-premise Windows service model. We also concluded that we consider an LDAP on-premise Windows authentication model and the Azure authentication model to have a higher chance of being less secure than the other authentication models researched in this thesis.

With the rise of the usage of AI, the need for computing power has grown exponentially. This has made cloud computing a popular option with its cost- effective and highly scalable capabilities. However, due to its popularity there exists thousands of possible services to choose from, making it hard to find the right tool for the job. The purpose of this thesis is to provide a methodological approach for evaluating which alternative is the best for machine learning applications deployed in the cloud. Nine different instances were evaluated on a major cloud provider and compared for their performance relative to their cost. This was accomplished by developing a cost evaluation model together with a test environment for image recognition models. The environment can be used on any type of cloud instance to aid in the decision-making. The results derived from the specific premises used in this study indicate that the higher the hourly cost an instance had, the less cost-effective it was. However, when making the same comparison within an instance family of similar machines the same conclusion can not be made. Regardless of the conclusions made in this thesis, the problem addressed remains, as the domain is too large to cover in one report. But the methodology used holds great value as it can act as guidance for similar evaluation with a different set of premises. / Användingen av Artificiell Intelligens har aldrig varit så stor som den är idag och behovet av att kunna göra tyngre och mer komplexa beräkningar har växt exponentiellt. Detta har gjort att molnet, cloud, ett mycket populärt alternativt för sin kostadseffektiva och skalbara förmåga. Däremot så finns det tusentals alternativ att välja emellan vilket gör det svårt att hitta rätt verktyg för jobbet. Syftet med denna uppsats är att förse läsaren med en användbar metodik för att evaluera vilket instans som passar bäst för maskininlärnings applikationer som distribueras i molnet. Nio stycken olika instanser evaluerades på en molnleverantör genom att jämföra deras prestanda kontra deras kostnad. Detta gjordes genom att utveckla en kostnadsmodell tillsammans med en testmiljö för bildigenkänningsmodeller. Testmiljön som användes kan appliceras på flertal instanser som inte ingick i denna rapport för att tillåta andra att använda den för egna tester. Resultaten för studien var att de instanserna med högre timkostnad tenderar till att vara mindre kostnadseffektiva. Gör man samma jämförelse med endast instanser av samma typ som är anpassade för maskininlärning så är samma slutsats inte lika självklar. Oavsett slutsatser som ges i denna rapport så består problemet. Detta beror på att molnet berör så många olika faktorer som bör värderas i evalueringen, till exempel utvecklingstid och modellens förmåga att förutspå en bild vilket alla kräver sin egna tes. Men metodiken som används kan definitivt vara till stor nytta om man vill göra en liknande utvärdering med andra premisser.