Search UNB Scholar

About UNB Scholar

UNB Scholar is an institutional repository initiative of UNB Libraries intended to collect, preserve, showcase, and promote the open access scholarly output of the UNB community. Use UNB Scholar to explore specific collections, or search all content in the repository. Material submitted to the repository will also be freely discoverable online through Google and other major search engines.

UNB Libraries welcomes submissions to UNB Scholar. Find out more about depositing your publications!

Communities in UNB Scholar

Select a community to browse its collections.

Recent Submissions

Item
Design of efficient and privacy-preserving similarity query over encrypted data in Cloud
(University of New Brunswick, 2022-02) Zheng, Yandong; Lu, Rongxing
Similarity query, retrieving objects similar to a sample of interest, has enabled plentiful customized services in a variety of applications, such as disease diagnosis, locationbased services, recommendation system, signal processing, etc.; and attracted considerable attention from industry and academia. Meanwhile, big data era has stimulated the continuing explosive growth of data volumes in these applications and has been leading data owners to outsource data and the corresponding similarity query services to a powerful cloud for releasing the burden of local data storage and computation. Due to privacy concerns, data are usually outsourced in an encrypted form, and as a result, similarity queries have to be performed over encrypted data, which is more challenging to implement than similarity queries over plaintext data. Targeting at similarity queries over encrypted data, many solutions have been proposed by designing some customized protocols based on selected encryption techniques, e.g., homomorphic encryption. Based on the type of data, three categories of extensively studied privacy-preserving similarity queries include similarity query for eHealthcare data, similarity query for set data, and similarity query for time series, which are also the focus of our dissertation. Although these three categories have been widely studied, existing solutions still have some limitations in query efficiency, security, and practicality. To address these issues, in the dissertation, we design several efficient and privacy-preserving similarity query schemes for encrypted eHealthcare data with the distance measure of Euclidean distance; for encrypted set data with the distance measure of Jaccard similarity; and for encrypted time series with the distance measure of time warp edit distance (TWED), in the cloud outsourced environment. Specifically, our main contributions of the dissertation can be summarized as i) Design an efficient and privacy-preserving kNN query scheme for outsourced eHealthcare data by employing the k-d tree data structure as index and designing two privacy-preserving protocols for the encrypted data comparison and Euclidean distance computation in the cloud based on a homomorphic encryption technique. The proposed scheme can not only support efficient and privacy-preserving kNN queries over encrypted eHealthcare data but also dynamically update encrypted eHealthcare data in the cloud. ii) Propose a new efficient, privacy-preserving, and practical Euclidean distance based similarity range query scheme over encrypted eHealthcare data in the cloud by integrating the modified asymmetric scalar-product-preserving encryption (ASPE) scheme and Quadsector tree structure, where the modified ASPE scheme enables the cloud server to determine whether a data point satisfies the current similarity range query request through the encrypted data under a single-server setting; and the Quadsector tree used to index the dataset reduces the average computational complexity of query processing sublinear to the size of the dataset. iii) Formalize and design a similarity query based healthcare monitoring scheme over the digital twin cloud platform. In this scheme, we deploy a partition-based tree (PB-tree) to represent the healthcare center’s data and introducing the modified ASPE scheme to design a privacy-preserving PB-tree based similarity range query (PSRQ) algorithm. iv) Propose an efficient and privacy-preserving set similarity query scheme under a single-server setting, which achieves high efficiency in the set similarity query while preserving the data privacy. In the proposed scheme, we design a symmetric-key predicate encryption scheme to achieve privacy-preserving similarity queries over binary vectors and employ B+ tree as the index structure to improve the query efficiency. v) Propose an efficient and privacy-preserving similarity range query scheme for time series data by organizing time series data into a k-d tree and applying the modified ASPE scheme and a symmetric homomorphic encryption to preserve the privacy of k-d tree based similarity queries. vi) Analyze the security of all proposed similarity query schemes in the dissertation and conduct extensive experiments to validate their efficiency. The results demonstrate that all proposed schemes are privacy-preserving, protecting the privacy of datasets and query requests against the honest-but-curious cloud server; and are computationally efficient.
Item
Binary logistic models with partially crossed random effects
(University of New Brunswick, 2021-11) Zhang, Zizhe; Yan, Guohua; Ma, Renjun
Educational studies and behavioural scientists frequently encounter data with binary outcomes that have cross-classified data structures. For example, in a student admission study (success or failure), schools and areas could be treated as crossed random effects since not all students from the same school live in the same area and vice versa. It is crucial to incorporate crossed random effects into the model for data with cross-classified structures; otherwise, data analysis results might be misleading. This thesis proposes a binary logistic model with partially crossed random effects, which is further extended to a baseline-category logit model with partially crossed random effects for multinomial analysis. The random effects in our proposed models are predicted by the orthodox best linear unbiased predictor (BLUP) approach. Our models are robust because they only need to specify the first and second moments of the random effects. The simulation study shows that the estimation algorithm generally performs well. In addition, we apply these models to insurance data about motor vehicle accidents and interpret the estimates for practical references.
Item
Tree characteristics selected by woodpeckers for foraging on snags and declining trees in regenerating Acadian Forest in New Brunswick
(University of New Brunswick, 2022-04) Zhang, Jingyi; Nocera, Joe
The value of dead and decaying wood to woodpeckers as breeding substrates is well understood. Many authors have emphasized its positive influence on woodpecker population density. However, many forest management activities including clearcuts, shelterwood cuts and seed tree cuts that result in dense, old stands with reduced snag densities are thought to negatively impact woodpecker populations. Therefore, I conducted a study on the selection of tree-scale characteristics of snags and declining trees for foraging by woodpeckers in regenerating Acadian Forest in New Brunswick to improve knowledge of the foraging habitat requirements of local woodpeckers. I searched for signs of foraging (excavation holes) by woodpeckers on 129 snags and declining trees sampled in 10 linear transects across a section of the University of New Brunswick (UNB) Woodlot. I developed a negative binomial regression model based on the presence or absence of foraging signs. I tested several variables for their influence on the probability of tree choice by woodpeckers for foraging including tree species, diameter at breast height (dbh), fungus coverage, and decay stage. Results showed that woodpeckers selected significantly larger diameter trees for excavation foraging (z = 3.197, p = 0.001, β = 0.085). Therefore, retaining large-tree and old-growth forests on site after management operations is essential to the conservation of woodpeckers in New Brunswick.
Item
High-resolution laser and far-infrared Fourier transform synchrotron-based spectroscopy of selected molecules
(University of New Brunswick, 2022-02) Zarringhalam, Hanif; Tokaryk, D. W.; Adam, A. G.
In the first part of this thesis, three ruthenium-bearing diatomic molecules have been studied in the visible region of the electromagnetic spectrum. Ruthenium monofluoride, ruthenium monochloride and ruthenium monoxide molecules were created in a molecular beam apparatus. The high-and low-resolution spectra of these small molecules were taken with the laser-induced fluorescence technique. The dispersed fluorescence technique was used to determine the vibrational frequencies of the RuF, RuCl and RuO molecules. The results of the high-resolution analysis of the spectra revealed extensive isotopic structures of the three molecules. Spin-orbit and hyperfine interactions in the ruthenium monofluoride molecule were observed and analyzed. Hyperfine structure in the ruthenium monoxide molecule was also detected and studied. In the second part of the thesis, three medium-sized ring molecules, catechol, furan and pyrrole, belonging to the C2v point group, have been studied in the infrared region. Vibrational bands of pyrrole and furan were collected between 800-900 cm-1 at the Canadian Light Source with an FTIR technique. Observed vibrational bands of pyrrole and furan in that region were studied. A nearby level that had A2 symmetry which could not be accessed from the ground states had perturbed the fully overlapped bands of these molecules. Also, progress has been made in obtaining the first high-resolution rotationally resolved vibrational band of catechol at the Canadian Light Source.
Item
Drivers of springtime flooding in the Upper Saint John River Basin (2001–2018)
(University of New Brunswick, 2022-02) Yu, Xindi; Bourque, Charles P.-A.
In this report, I assessed the role of various environmental variables in controlling springtime flooding in the upper Saint John River (SJR) basin over a seventeen-year period, from 2001–2018. There is ample research in the scientific literature that analyzes flooding, but research specific to the springtime flooding of the upper SJR is noticeably absent. The objectives of my research were to characterize waterflow behavior in the upper SJR as a function of hydrometeorological and landcover data, both acquired from independent data sources available on the world wide web, i.e., Daymet and Global Forest Watch program data, respectively. Data for the hydrometric variables for the region (i.e., discharge rate and stage height at various points along the SJR system) were acquired from Environment and Climate Change Canada. The results showed that repeated forest cover removal in the upper SJR basin had a role in increasing the risk of spring flooding. Near-ground air temperatures and cumulative snow degree-days during the snow-ablation period of each year were equally important given their role in generating meltwater and causing surges in stream and river discharge. The report displays an innovative statistical tool for land managers’ decision-making regarding waterflow dynamics in the upper SJR basin under anticipated climate change.