21090 research outputs found
Sort by
A Study of the Opportunities and Challenges of Using Edge Computing to Accelerate Cloud Applications
No full textI explore the viability of using edge clusters to host latency-sensitive applications and to run services that can improve end-to-end communication performance across both wide area networks (WANs) and 5G environments. The study examines the viability of using edge clusters in three scenarios: accelerating TCP communications through TCP splitting in 5G deployments, hosting an entire application-level service or the latency-sensitive part of an application on an edge cluster, and deploying a TCP splitting service on edge clusters to support WAN communication. I explore these scenarios while varying packet drop rates, communication stacks, congestion control protocols, and TCP buffer sizes.
My findings bring new insights about these deployment scenarios. I show that edge computing, especially through TCP splitting, can significantly improve end-to-end communication performance over the classical communication stack. TCP splitting over the 5G communication stack does not bring any benefit and can reduce throughput. This is because of the unique characteristics of the 5G communication stack. Furthermore, over the classical communication stack, TCP splitting brings higher benefit for flows larger than 64 KB. These findings provide valuable insights into how edge clusters can accelerate TCP communication in different network environments and identify high-impact research ideas for future work
MATERIAL WORKS: Optimizing Material Circularity through Reversing Architecture
No full text“Material Works” posits the largest abandoned industrial landscape in the Niagara region — the former General Motors Plant site — as a key infrastructure to circulate the existing building stocks for city scale reuse in St. Catharines, Ontario. The city’s building permits issued in 2023-2024 were analyzed to examine buildings registered for demolition and to build a database of materials to be reused if deconstructed. This initial study informed the design of the facility,
in its scale, aesthetics and programmatic organization.
The half-demolished GM structure, with a former building footprint of 38,000m2, is transformed to a circularity hub to address the city’s potential reusable building material stock. The architecture provides spaces for people to train in deconstruction, salvaged materials to be processed for resale, and designers to demonstrate their potential for architectural re-application. With circulation of materials as the central motif, agencies essential in facilitating circular activities are imagined to co-exist in one physical site to develop approaches to create more sustainable, closed-loop metabolic systems of materials.
The building industry constitutes nearly a quarter of the global waste stream, and with Ontario’s landfills projected to reach capacity by 2032, the movement and uncertain destination of materials remain critical environmental concerns. In the quest for a sustainable architectural future, where construction’s inherent destructiveness contrasts with the demand for densification, focus shifts toward assessing the residual value of existing urban building stocks within the Anthropogenic landscape. This paradigm shift from the current linear to a circular construction model protects the architectural heritage of our urban fabric from rapid erasure while optimizing resource efficiency.
This thesis explores design interventions and industry practices that replace the 21st century’s planned-obsolescence-thinking with reuse, contributing to the discourse of material circularity to address the environmental and cultural resiliency in architecture
Heat Production and Transfer in Earth’s Continental Crust
No full textPlanetary differentiation through tectonism reflects heat production and transfer, reducing internal energy and shaping planetary interiors. Geologic heat originates from two primary sources: primordial heat from planetary formation and the radiogenic decay of isotopes like U, Th, and K. Shearing can generate heat on local scales, but heat transfer predominantly occurs through conduction, where energy flows from hotter to cooler regions via atomic vibrations. In tectonically active areas, advection of magma and convection in fluids are more efficient mechanisms. Local heat transfer often relies on one dominant process, while large-scale systems involve a mix of conduction, convection, and advection. The causes and proportions of each heat source and mantle and crustal radiogenic contributions remain challenging to quantify.
Understanding the re-distribution of heat-producing (i.e., radioactive) elements during metamorphism and crustal differentiation is achieved via combining natural observations with trace element and accessory mineral modelling. Six potential end-members were considered for the protolith of mid-crustal tonalite-trondhjemite-granodiorite packages—often considered the product of lower-crustal melting. Model results suggest heat-producing elements partition subequally between solid and melt at typical pressure-temperature conditions for crustal differentiation. Accessory minerals like apatite, feldspar, amphibole, and epidote are primary repositories for radioactive elements, with their stability in pressure-temperature space governing heat removal from the lower crust. Observations from the Archean Kapuskasing uplift reveal a similar partitioning pattern during mafic rock melting, supporting the notion that radiogenic heat equally influences mantle and crustal processes.
Examining the crustal heat-production record provides insights into continental growth, thickness, and preservation. A database of crustal rocks, including trace element compositions and crystallization ages, reveals trends in heat production over time with implications for basalt formation and crustal evolution. While Archean mantle melting produced less heat-producing basalt than today due to higher degrees of partial melting, the total crustal heat-production rate has remained relatively constant. However, modern crust exhibits more significant variability, suggesting recent enrichment in heat-producing elements contrasts with a more homogenized Archean crust. Crustal growth pulses marked by mafic-to-felsic transitions imply a cyclic nature of crustal differentiation, with crustal thickness remaining stable due to self-organizing thermal processes.
Trace element substitution in autocrystic zircon is a valuable tool for reconstructing deep geologic processes, but some influences on these proxies remain underexplored. Elevated titanium rims on volcanic zircon, often attributed to magma recharge, could also result from adiabatic ascent. Modelling shows that decompression melting and system expansion during ascent drive cooling, while subvolcanic boiling induces crystallization and latent heat release. Zircon growth during ascent can record these processes, and high-titanium rims may form in a single magma pulse without recharge, emphasizing the importance of multiple geochemical tools to interpret magma evolution.
Ultra-high temperature (UHT) metamorphism represents the thermal extreme of crustal processes, yet its mechanisms and energy sources remain debated; common explanations include mafic underplating or mantle upwelling. The Frontenac Terrane in southeastern Ontario records UHT conditions during the Mesoproterozoic. Back-arc sedimentation between 1390–1200 Ma preceded Shawinigan (~1180–1160 Ma) and Ottawan (~1060 Ma) orogenic events. Granitic and minor mafic intrusions during Shawinigan times triggered regionally advective UHT metamorphism preserved through subsequent reheating events. The Frontenac Terrane provides critical insights into the Grenville Province assembly, with felsic intrusions providing a plausible mechanism for UHT conditions during orogenesis.
Heat production and transfer fundamentally shape Earth's structure and behaviour. Radiogenic heat from U, Th, and K and primordial heat drive crustal and mantle dynamics. Although incompatible with most rock-forming minerals, heat-producing elements are preferentially incorporated into accessory minerals like apatite and zircon, depleting their source rocks. Increased melting reduces system-wide heat production as these elements concentrate in the melt. Advective heating is crucial for crustal growth, enabling magmas to ascend nearly adiabatically and providing a heat source for crustal reworking. These processes dictate metallogenic fluid generation, crustal heterogeneity, and long-term crustal stabilization
Microstructure control and property enhancement of NiTi-stainless steel dissimilar joints
No full textDissimilar joining between Nickel-Titanium (NiTi) and stainless steel (SS) is of significance in many areas especially biomedical applications, however, achieving reliable NiTi-SS joints is highly challenging due to the formation of brittle intermetallic compounds (IMCs) in the fusion zone (FZ) or the interface. Two strategies can be summarized to address this issue: (1) restricting the mixing of molten metals and (2) replacing the most harmful Laves (Fe,Cr)2Ti with ductile phases. The former one poses large processing complexity and may lead to NiTi plastic deformation degrading the functional properties. The latter struggles to eliminate brittle IMCs entirely in the FZ and may introduce toxic elements. This research investigated both aspects to control the microstructure and properties of NiTi-SS joints by leveraging the flexibility of laser beam and the thermomechanical process of resistance welding.
The combination of laser beam defocus and large offset enabled the laser weld-brazing of NiTi and SS wires. This approach successfully eliminated the IMCs network in the FZ, shifting the conventional and complex FZ brittleness issue to a focus on controlling the brazed interface. Additionally, laser welding mode significantly influenced the macrosegregation and porosity in the FZ of NiTi-SS joints. Low laser power density and long welding time mitigated the macrosegregation and porosity by weakening the laser keyhole effect and prolonging the molten pool duration. In NiTi-SS laser weld FZ, large pores were caused by the instability or collapse of the laser keyhole, while small pores originated from the Ni vaporization.
Both IMCs control strategies were investigated in resistance spot welding (RSW) of NiTi and SS for the first time. The use of Nb interlayer resulted in a unique sandwich-structured joint, where two FZs were separated by solid-state Nb, suppressing the mixing of dissimilar molten metals. Nb-containing eutectics formed at both interfaces, enhancing the joint strength with a 38% increase in fracture load and a remarkable 460% increase in energy absorption. In another approach, increasing Ni concentration via a melted Ni interlayer effectively replaced Fe2Ti with relatively ductile Ni3Ti in the FZ. However, high Ni content also induced large pores and cracks, limiting the effectiveness of this strategy in NiTi-SS RSW.
A novel processing approach leveraging interfacial liquid control was proposed, achieving a solid-state joined interface in NiTi-SS fusion welding (e.g., resistance microwelding) without any additional interlayers. The produced NiTi-SS joints showed superior strength, superelasticity and corrosion resistance compared to NiTi joints or base metal. The ultrathin reaction layer at the solid-state joined interface contributed to a strong metallurgical bonding, while Joule heating effects and interfacial reactions enhanced superelasticity and corrosion resistance of the joint. Notably, a face-centered-cubic (FCC) reorientated layer (ROL) was found between SS and IMC layer at the controlled ultrathin interface. The formation of this ROL was uncovered based on an epitaxial growth model. This ROL introduced a strong crystallographic mismatch with the textured SS, resulting in the fracture at this interface. These phenomenal findings offer valuable insights for studying material interface and controlling dissimilar-metal welding process
Engineering cell-penetrating peptide mediated protein-bound nanoparticles for delivering siRNA and chemotherapeutics
No full textProteins serve as the “workers” of biochemistry, orchestrating nearly all biological functions. Functional endogenous proteins are often related to the pharmacokinetics and pharmacodynamics of drugs and nanomedicines, particularly in processes such as drug absorption, biodistribution, and metabolism. That means the innate interactions between proteins and drugs/nanoparticles exist, but the discovery and application of these interactions are underappreciated so far. By imitating the protein binding behaviors and interactions, some proteins may hold significant promise in drug and nanoparticle delivery due to their biocompatibility and functionalities. This thesis presents a methodology for engineering biomimetic protein coronas to camouflage cationic peptide/siRNA (P/si) nanocomplexes by utilizing proteins derived from the innate P/si protein corona (P/si-PC), which was also applied to the peptide-based lipid nanoparticles (pLNP). By leveraging these protein corona species, an efficient method for producing protein-bound chemotherapeutic nanoparticles in aqueous phases using microfluidic technology was developed.
For cationic nanoparticles, the spontaneous nanoparticle-protein corona formation and aggregation in biofluids can trigger unexpected biological reactions. This thesis presents a biomimetic strategy for camouflaging the P/si with single or dual proteins, which exploits the unique properties of endogenous proteins and stabilizes the cationic P/si for safe and targeted delivery. An in-depth study of P/si-PC formation and protein binding was conducted. The results provided insights into the biochemical and toxicological properties of cationic nanocomplexes and the rationales for engineering biomimetic protein camouflages. Based on this, the human serum albumin (HSA) and apolipoprotein AI (Apo-AI) ranked within the top 20 abundant protein species of P/si-PC were selected to construct biomimetic HSA-dressed P/si (P/si@HSA) and dual protein (HSA and Apo-AI)-dressed P/si (P/si@HSA_AI), given that the dual-protein camouflage plays complementary roles in efficient delivery. A branched cationic cell-penetrating peptide (CPP, b-HKR) was tailored for siRNA delivery, and their nanocomplexes including the cationic P/si and biomimetic protein-dressed P/si were produced by a precise microfluidic technology. The biomimetic anionic protein camouflage greatly enhanced P/si biostability and biocompatibility, which offers a reliable strategy for overcoming the limitation of applying cationic nanoparticles in biofluids and systemic delivery.
Currently, commercially applied lipid nanoparticles (LNPs) for RNA delivery, such as in siRNA and mRNA vaccines, utilize similar lipid compositions and ratios, raising the risk of unintentional patent infringement. This research attempted to engineer a novel peptide-based LNP formulation stabilized and functionalized by artificial protein corona that constitutes HSA and lipoprotein (Apo-AI; apolipoprotein E, Apo-E). The cationic peptide (b-HKR) enabled efficient siRNA condensation and reversible protein binding. Combining b-HKR and the artificial protein corona offers an alternative to the commonly used ionizable lipids, PEG-lipids, and excipients (such as sucrose), providing both pH-responsive functionality and storage stability. The in vitro results showed that the dual protein (HSA and Apo-AI) functionalized pLNP (pLNP@HSA_AI) is optimal for enhanced stability and RNAi efficacy. In contrast, single protein-functionalized pLNPs encountered a dilemma: pLNP@HSA improved stability but showed almost no RNAi efficacy, while the pLNP@AI exhibited remarkable RNAi efficacy but aggregated upon the addition of Apo-AI. The dual protein (HSA and Apo-E) functionalized pLNP (pLNP@HSA_E) also showed promise in addressing this dilemma, although the use of Apo-E is less cost-effective than Apo-AI due to its limited availability.
The use of endogenous proteins, particularly albumin, for the targeted delivery of chemotherapeutics has proven practical. However, how to effectively produce the protein-bound chemotherapeutics nanoparticles in a complete aqueous phase (without the use of organic solvents) is worth pursuing to eliminate the solvent-related safety risks. In this research, the protein-bound Dox (Dox) nanoparticles were successfully produced through a one-step microfluidic mixing process in aqueous phases, in which the nanoparticle formation was instantaneously mediated by a self-assembled nano-peptide (np). The np-mediated HSA-bound Dox (D-np-HSA) and dual proteins (HSA; Apo-AI)-bound Dox (D-np-HSA-AI) nanoparticles exhibited efficient drug encapsulation and pH-triggered drug releases. In vitro cellular studies showed that the nanoparticles (D-np-HSA and D-np-HSA-AI) exhibited superior efficacy in killing tumor cells (A549 and MCF7) while being less toxic to normal cells (NIH3T3) compared to free Dox. Notably, D-np-HSA-AI was less prone to induce drug resistance, and cell lines that developed resistance to free Dox remained sensitive to D-np-HSA-AI. Besides, the results revealed that drug resistance development of A549 is associated with cellular phenotypic (size, morphology, and dividing speed) changes. Cellular (cytoplasmic and nuclear) proteomics was conducted by comparing the protein species, abundances, and relation networks of normal, Dox-induced, and nanoparticle (D-np4-HSA-AI) induced A549 cells, which aimed to provide potential protein biomarkers associated with drug resistance and druggable protein/gene targets for overcoming the drug resistance
Low-power and Radiation Hardened TSPC Registers
No full textBattery-operated systems require power and energy-efficient circuits to extend their battery life. Flip-flops (FFs) are a basic component of digital circuits, and their power consumption and speed significantly impact the overall performance of a digital system. A clock network in a complex System-on-Chip (SoC) consumes a substantial amount of power. Additionally, often pipelines are used to enhance the system throughput, which puts additional burden on the clock network. Arguably, a flip-flop with fewer clock transistors will reduce its power burden on the clock network. This research proposes three very low-power Single-edge Triggered (SET) True Single-phase Clock (TSPC) FFs with only two and three clock transistors. Moreover, a scan-chain of 256 FFs and AES-128 encryption engine were designed as a benchmark to further investigate the power savings of the proposed FFs. Additionally, we have also designed three very low-power Dual-edge Triggered (DET) latch-multiplexer
type TSPC FFs with only eight and ten clock transistors to sample the data at both positive and negative clock edges.
Furthermore, high-performance computations in Integrated Circuits (ICs) are increasingly needed for space and safety-critical applications. ICs are subjected to high-energy ionizing particles in the radiant space environment, which will cause the device performance to degrade or even fail. A Single Event Upset (SEU) occurs in the logic circuit when an ion strikes a device’s sensitive node, changing the output from 0 to 1 or from 1 to 0. In radiant applications, ICs contain storage cells like FFs, latches, or Static Random Access Memories (SRAM), and always experience SEU. Although package and process engineering can minimize alpha particles, cosmic neutrons cannot be physically blocked. Therefore, for high reliability systems, soft error tolerant circuit designs are crucial. Traditional Radiation Hardened By Design (RHBD) techniques have some trade-offs between area, speed, power, and energy consumption. Thus, new designs are required to reduce these penalties. This research proposes two high-performance, low-power, low-energy, and low-area RHBD TSPC FFs with only four and five clock transistors suitable for space and safety-critical applications
Applications of Lévy Semistationary Processes to Storable Commodities
No full textVolatility Modulated Lévy-driven Volterra (VMLV) processes have been applied by
Barndorff-Nielsen, Benth and Veraart (2013) to construct a new framework for modelling spot
prices of non-storable commodities, namely energy. In this thesis, we extend this framework to
storable commodities by showing that successful classical models belong to the framework albeit
under some parameter restrictions (a result which to our knowledge is new). Additionally, we
propose a new model for spot prices of storable commodities which is built on the VMLV
processes and their important subclass of so-called Lévy semi-stationary (LSS) processes. The
main feature of the framework exploited in the model proposed in this thesis is the memory of the
VMLV processes which is used judiciously to account for cumulative changes in inventory over
time and the corresponding expected changes in prices and volatility. To the best of our
knowledge, this is the first study which uses the LSS processes to investigate pricing in storable
(as opposed to non-storable) commodity markets to account for the impact of inventory on pricing.
To complement the theoretical development of the new model, we also provide in this thesis a
companion set of calibration and empirical analyses to shed light on the new model’s performance
compared to previously established models in the literature
On the three charge regimes of bipolar charge conditioners
No full textThis is an Accepted Manuscript of an article published by Taylor & Francis in Aerosol Science and Technology on 12 March 2025, available online: https://doi.org/10.1080/02786826.2025.2461161.Many aerosol instruments utilize bipolar charge conditioners (neutralizers), including scanning mobility particle sizers (SMPS). The charge distribution from a bipolar charge conditioner must be known to calculate size distributions from SMPS scans. Therefore, a reproducible and known “steady-state” charge distribution is desired to improve measurement accuracy. In this work, we show that although a steady-state charge develops within a common Kr-85 bipolar charge conditioner (TSI 3077A), gaseous ions are readily convected into tubing downstream of the charge conditioner, causing significant deviation from steady-state in less than a second. The “downstream ions” are predominantly positive since negative ions are more readily lost to tubing walls due to their higher diffusivity. Others have previously studied this potential effect, but there is disagreement among them. This study resolves the disagreements, and to the authors’ knowledge, we are the first to: quantify the surprising significance of this effect (mean charge at 239 nm changes by up to a factor of 4); show it occurs rapidly (milliseconds); and demonstrate that a true steady-state distribution is more asymmetric than classical theory (mean charge at 239 nm is >3x higher in magnitude). Interaction time of the particles with the remaining free ions downstream of the charge conditioner is shown to be the main consideration for this effect, as demonstrated by varying flow rate, tube length and tube diameter and achieving similar charging results as a function of this time. We perform advanced numerical modeling of charging, convection and diffusion of particles and ions, and show good quantitative agreement with experimental data. These results are further supported by similar charging results measured from a bipolar charge conditioner that has a different internal geometry and ion source than the 3077A charger. To increase consideration of this critical charging effect and enhance understanding of bipolar charging in general, we thereby quantify three distinct regimes of bipolar charging, namely: the (1) charging regime (i.e., as charge develops from its initial to steady-state), (2) steady-state regime, and (3) discharging regime (i.e., effect of “downstream ions”). These three regimes of bipolar charging have wide implications for designing instrumentation, interpreting measurements and validating charging models with the aim of improving accuracy in aerosol instruments such as the SMPSTyler Johnson acknowledges and appreciates the significant support of the Postdoctoral Fellowship from Natural Sciences and Engineering Research Council of Canada (NSERC). Robert Nishida and Jason Olfert received funding support from Transport Canada
High-Dimensional Statistical Inference and False Discovery Rate Control with Covariates
No full textIn this thesis, we focus on three statistical problems. First, we consider graph-based tests for differences of two high-dimensional distributions. Second, we investigate the estimation of multiple large covariance matrices and the application to high-dimensional quadratic discriminant analysis. Lastly, we focus on controlling the false discovery rate while incorporating complex auxiliary information.
Testing whether two samples are from a common distribution is an important problem in statistics. Friedman & Rafsky (1979) proposed a non-parametric multivariate distribution test based on the minimal spanning tree (MST). Recently, this test has been extended under various scenarios. However, as demonstrated in Chapter 2, these extensions are not sensitive to sparse alternatives. To address this, we propose a two-step testing procedure, IM-MST. Specifically, IM-MST incorporates marginal screening while accounting for the dependence structure via energy distance, followed by MST-based tests. IM-MST combines the strength of both non-parametric screening and MST-based tests. Simulation studies and real data applications are conducted to evaluate the numerical performance of the two-step procedure, demonstrating that IM-MST exhibits substantial power gains.
When estimating covariance matrices for data from two related categories, it is reasonable to assume that these covariance matrices share certain structural components. As a result, the precision matrix (the inverse of the covariance matrix) for each category can be decomposed into three parts: a common diagonal component, a common low-rank component, and a category-specific low-rank component. This decomposition can be motivated by a factor model, where some latent factors are common across two categories while others are specific to individual categories. In Chapter 3, we propose a consistent joint estimation method for two precision matrices building on the work of Wu (2017). Furthermore, these estimators are applied to formulate a high-dimensional quadratic discriminant analysis (QDA) rule, for which we derive the convergence rate for the classification error.
In many genetic multiple testing applications, the signs of the test statistics provide important directional information. For example, in RNA-seq data analysis, a negative sign could suggest that the expression of the corresponding gene is potentially suppressed, while a positive sign could indicate a potentially elevated expression level. However, most existing procedures that control the false discovery rate (FDR) ignore such valuable information. In Chapter 4, we extend the covariate and direction adaptive knockoff procedure (Tian 2020) by implementing powerful predictive functions. Through simulation studies and real data analysis, we show that our procedures are competitive to existing covariate-adaptive methods. The companion R package Codak is available
MIRAGE-ANNS: Mixed Approach Graph-based Indexing for Approximate Nearest Neighbour Search
No full textApproximate nearest neighbor search (ANNS) on high dimensional vectors is important for numerous applications, such as search engines, recommendation systems, and more recently, large language models (LLMs), where Retrieval Augmented Generation (RAG) is used to add context to an LLM query. Graph-based indexes built on these vectors have been shown to perform best but have challenges. These indexes can either employ refinement-based construction strategies such as K-Graph and NSG, or increment-based strategies such as HNSW. Refinement-based approaches have fast construction times, but worse search performance and do not allow for incremental inserts, requiring a full reconstruction each time new vectors are added to the index. Increment-based approaches have good search performance and allow for incremental inserts, but suffer from slow construction. This work presents MIRAGE-ANNS (Mixed Incremental Refinement Approach Graph-based Exploration for Approximate Nearest Neighbor Search) that constructs the index as fast as refinement-based approaches while retaining search performance comparable or better than increment-based ones. It also allows incremental inserts. We show that MIRAGE achieves state of the art construction and query performance, outperforming existing methods by up to 2x query throughput on real-world datasets