The objective of this research is to characterize and manage lifetime reliability, microarchitectural performance, and power tradeoffs in multicore processors. This dissertation is comprised of three research themes; 1) modeling and simulation method of interacting multicore processor physics, 2) characterization and management of performance and lifetime reliability tradeoff, and 3) extending Amdahl’s Law for understanding lifetime reliability, performance, and energy efficiency of heterogeneous processors. With continued technology scaling, processor operations are increasingly dominated by multiple distinct physical phenomena and their coupled interactions. Understanding these behaviors requires the modeling of complex physical interactions. This dissertation first presents a novel simulation framework that orchestrates interactions between multiple physical models and microarchitecture simulators to enable research explorations at the intersection of application, microarchitecture, energy, power, thermal, and reliability. Using this framework, workload-induced variation of device degradation is characterized, and its impacts on processor lifetime and performance are analyzed. This research introduces a new metric to quantify performance-reliability tradeoff. Lastly, the theoretical models of heterogeneous multicore processors are proposed for understanding performance, energy efficiency, and lifetime reliability consequences. It is shown that these system metrics are governed by Amdahl’s Law and correlated as a function of processor composition, scheduling method, and Amdahl’s scaling factor. This dissertation highlights the importance of multidimensional analysis and extends the scope of microarchitectural studies by incorporating the physical aspects of processor operations and designs.

Calcium (Ca2+), a crucial effector for many biological systems, has been associated with diseases such as cardiovascular disease, Alzheimer’s, Parkinson’s, cancer, and osteoporosis. It is important to develop calcium sensors to measure intracellular Ca2+ dynamics at various biological and pathological states. Our lab has engineered such probes by designing a Ca2+ binding site into fluorescent proteins such as Enhanced Green Fluorescent Protein (EGFP) and mCherry. In this thesis, we aim to improve optical properties and metal binding properties of green EGFP-based sensor CatchER and mCherry based red sensors by site-directed mutagenesis and protein engineering, various spectroscopic methods and cell imaging. The green EGFP-based sensor CatchER, with a Ca2+ binding pocket charge of -5, displays the greatest affinity for Ca2+ and has the greatest fluorescence intensity change with Ca2+ when compared to its variants with a less negative binding pocket charge. In addition, we have also designed several SR/ER targeting CatchER variants using Ryanodine receptor and Calnexin transmembrane domains. These constructs were shown to display a strong presence in the SR/ER lumen and further designed for a new luminal orientation. Further, we have shown that the optical properties of two red calcium sensors can be significantly improved by modifying the local environment of the chromophore.

Microneedle (MN) enhanced drug delivery is a safe, effective and efficient enhancement method for delivery of drug molecules across the skin. The “poke (press) and patch” approach employs solid stainless steel MN to permeablize the skin prior to application of a regular drug patch over the treated area. It has been previously shown that MN can be used to deliver naltrexone (NTX) at a rate that provides plasma concentrations in the lower end of the therapeutic range in humans. The drug delivery potential of this technique is, however, limited by the re-sealing of the micropores in a 48-72h timeframe. The goal of the current research was to optimize the formulation for a 7 day MN enhanced delivery system for NTX either by adding a second active pharmacological moiety or by optimizing formulation characteristics alone. Three different formulation strategies were explored: formulation pH optimization with NTX; a codrug approach with NTX and a nonspecific cyclooxygenase inhibitor, diclofenac (DIC); and a topical/transdermal approach with NTX and an enzyme inhibitor of the cholesterol synthesis pathway, fluvastatin (FLU). The results indicated that formulation pH cannot be used to extend micropore lifetime, although formulation optimization leads to enhanced transport and thus drug delivery across MN treated skin. The codrug approach was successful in extending the micropore lifetime and further screening of codrug structures and formulation optimization helped in selection of a codrug candidate suitable for evaluation in animal pharmacokinetic studies. Local treatment with FLU helped to keep the micropores open and enabled delivery of NTX for an extended period. The pores re-sealed on removal of treatment within a 30-45 minute timeframe, indicating that infection/irritation should not be a major issue, as in the case of other topical chemical enhancers. Thus, overall it can be concluded that different formulation strategies can be utilized to extend micropore lifetime and enhance delivery of drug molecules across the skin.

The technique of Fluorescence Lifetime Imaging Microscopy (FLIM) has been employed to quantitatively and spatially map the fluid composition and temperature within microfluidic systems. A molecular probe with a solvent-sensitive fluorescence lifetime has been exploited to investigate and map the diffusional mixing of fluid streams under laminar flow conditions within a microfluidic device. Using FLIM, the fluid composition is mapped with high quantification and spatial resolution to assess the extent of mixing. This technique was extended to quantitatively evaluate the mixing efficiency of a range of commercial microfluidic mixers employing various mixing strategies, including the use of obstacles fabricated within the channels. A fluorescently labelled polymer has been investigated as a new probe for mapping temperature within microfluidic devices using FLIM. Time Correlated Single Photon Counting (TCSPC) measurements showed that the average fluorescence lifetime displayed by an aqueous solution of the polymer depended strongly on temperature, increasing from 3 ns to 13.5 ns between 23 and 38 oC. This effect was exploited using FLIM to provide high spatial resolution temperature mapping with sub-degree temperature resolution within microfluidic devices. A temperature-sensitive, water-soluble derivative of the rhodamine B fluorophore, effective over a wide dynamic temperature range (25 to 91 oC) has been used to map the temperature distribution during the mixing of fluid streams of different temperatures within a microchannel. In addition, this probe was employed to quantify the fluid temperature in a prototype microfluidic system for DNA amplification. FLIM has been demonstrated to provide a superior approach to the imaging within microfluidic systems over other commonly used techniques, such as fluorescence intensity and colourimetric imaging.

This study was conducted to: i) determine the effect of residual feed intake (RFI) on first parity fertility and subsequent lifetime productivity of 820 females ranked as low, medium, and high RFI in western Canada and ii) investigate the relationship between RFI measured as a heifer and re-measured as a mature cow. No significant correlations were found between RFI and fertility/productivity traits. A negative trend (P < 0.10) was observed between RFI, RFIfat and MPPAbw, however, this trend was no longer apparent when RFI was adjusted for back fat and feeding event frequency (RFIfat&activity). No significant (P < 0.05) relationship was found between heifer RFI and cow RFI. These results suggest that selection for feed efficient, low RFI heifers i) has no impact on their fertility and productivity as cows, and ii) may not lead to superior feed efficiency as a mature cow as measured by RFI and Cow DMI. / February 2017

Master Thesis: Customer Lifetime Value - Application to Banking Sector Author: Zuzana Lajksnerová Abstract This thesis deals with the calculation of the customer life value for clients of second largest Czech commercial banks - Československá obchodní banka a.s.. The aim is to estimate the revenue each individual brings to the bank within the first year, two, four since creating a profile but also during the whole time he or she remains in their customer base. The first, theoretical part contains of structured overview of the modeling approach from available literature. In the second part, we develop two different models, which are subsequently applied on client data of 2.7 million ČSOB customers. The prediction takes into account both revenues from each product and probability that customer will use this product. For estimation of the two models we use several different econometric methods, thus linear regression, bootstrap, probit and multinomial logit. The results show that demographic data such as age and gender, as well as product related variables, greatly inuence the final level of value of customer. It is also shown that higher potential earnings are associated with men, people of working age and loyal customers.

No description available.

During the use phase of products, a series of obsolescing factors contribute to why a product is disposed of. Currently the visual state of a product is considered primarily in terms of aesthetic obsolescence which is synonymous with influential factors such as changes in fashion or personal preferences in style. The physical condition of a product is not commonly understood within the context of product replacement and the physical changes due to use are not understood fully. The research contributes to and provides original empirical research findings for the current literature on product lifetime extension, material semantics, the circular economy, emotionally durable design and material culture. Through an initial exploratory study (Photographic Analysis (PA) Study) of previously unexplained types of wear and damage that occur on portable electronic devices a taxonomy of damage (TOD) was established which provided the nomenclature for further studies. The second study (Retrospective Assessment (RA) Study) established the attitudes to wear based on the wear type, location, material and the stage during ownership that the wear occurred at. The RA Study highlighted the differences in the attitudinal responses to differing types of wear and damage and identified the differences in the temporal assessments of wear and damage. A third study (Real Time Assessment (RTA) Study) aimed to confirm or repudiate the findings found in the RA Study. The focus during the study was attitudes to the wear and damage in relation to the differences in materials, the location of the wear and the type of wear and damage was also looked at and led to a fuller understanding of how products and materials are perceived during the use phase; a stage of the product lifetime that is not currently well understood in terms of users aesthetic or cosmetic sensibilities. The final study (Semantic Perception of Materials (SPM) Study) focused on the visual and tactile perceptions of materials. The study established attitudinal perceptions of wear and damage of materials with a quantitative research methodology which has produced a better understanding of material semantics within the context of electronic objects. Through the four studies, discussion topics arose and major findings of the doctoral study were drawn out and seen to be interesting enough for further research and study. These discussions include the importance of including cosmetic obsolescence into the lexicon of product obsolescence and product lifetime extension literature, the differences in the perceptions of materials when they are within the context of a product or being assessed as samples, how differing product contexts affect user perceptions of wear and damage on materials and the potential inclusion of a material wear index that could inform the material selection process that goes further than the technical aspects outlined in current material selection tools and literature.

Traditionally, marketers have assumed that investing in the quality of relationships with customers would generate superior profitability to the selling firm. The assumption is that coordination and collaboration between buyer and seller create value for both firms by reducing costs and expanding revenue opportunities. However, such value creation mechanism does not work every time. Closer relationships require customer specific investments and a higher level of service that may create more cost to the seller than the potential gain in revenue, negatively impacting profitability of the selling firm. This research explores the effect of buyer-seller relationship quality on value creation for the selling firm, emphasizing the understanding of costs associated with serving the relationship. Two studies were conducted: Study 1 qualitatively examines the cost associated with serving customers and estimates actual cost-to-serve for individual customers. Study 2 measures the quality of customer relationships, past customer profitability and customer lifetime value for each customer relationship in the proposed sample. Finally, the effect of RQ on customer lifetime value is evaluated. Results suggest that investing in customer relationships have an effect on the drivers of customer cost and profitability. However, the net effect on customer value is not as clear as it depends on the trade offs of the different drivers of cost and profit.

Wireless sensor networks (WSNs) are emerging as a key enabling technology for applications domains such as military, homeland security, and environment. However, a major constraint of these sensors is their limited battery. In this dissertation we examine the problem of maximizing the duration of time for which the network meets its coverage objective. Since these networks are very dense, only a subset of sensors need to be in "sense" or "on" mode at any given time to meet the coverage objective, while others can go into a power conserving "sleep" mode. This active set of sensors is known as a cover. The lifetime of the network can be extended by shuffling the cover set over time. In this dissertation, we introduce the concept of a local lifetime dependency graph consisting of the cover sets as nodes with any two nodes connected if the corresponding covers intersect, to capture the interdependencies among the covers. We present heuristics based on some simple properties of this graph and show how they improve over existing algorithms. We also present heuristics based on other properties of this graph, new models for dealing with the solution space and a generalization of our approach to other graph problems.