Browsing by Subject "Data modeling"
Now showing 1 - 2 of 2
Results Per Page
Sort Options
Item Broadband electromagnetic data interpolation based on a hybrid polar-exponential fitting algorithm(2006-08) Starosta, Matthew Samuel, 1981-; Ling, HaoA hybridized electromagnetic fitting algorithm using polar and exponential basis functions is presented with comparisons to previous interpolation methods. In the first chapter an introduction and overview of model-based parameter estimation is given. Previous interpolative techniques and models are discussed and compared with sample results in the second and third chapters. The second chapter is devoted to the rational function model and its uses modeling data resembling resonance phenomena. The third chapter discusses the uses of adaptive feature extraction (AFE) and its effectiveness in modeling scattering responses. Chapter four presents the hybrid fitting algorithm and results on synthetic and modeled data. The final chapter contains conclusions and suggestions for future work.Item Supporting device-to-device search and sharing of hyper-localized data(2015-05) Michel, Jonas Reinhardt; Julien, Christine; Garg, Vijay; Lam, Simon; de Veciana, Gustavo; Vishwanath, SriramSupporting emerging mobile applications in densely populated environments requires connecting mobile users and their devices with the surrounding digital landscape. Specifically, the volume of digitally-available data in such computing spaces presents an imminent need for expressive mechanisms that enable humans and applications to share and search for relevant information within their digitally accessible physical surroundings. Device-to-device communications will play a critical role in facilitating transparent access to proximate digital resources. A wide variety of approaches exist that support device-to-device dissemination and query-driven data access. Very few, however, capitalize on the contextual history of the shared data itself to distribute additional data or to guide queries. This dissertation presents Gander, an application substrate and mobile middleware designed to ease the burden associated with creating applications that require support for sharing and searching of hyper-localized data in situ. Gander employs a novel trajectory-driven model of spatiotemporal provenance that enriches shared data with its contextual history -- annotations that capture data's geospatial and causal history across a lifetime of device-to-device propagation. We demonstrate the value of spatiotemporal data provenance as both a tool for improving ad hoc routing performance and for driving complex application behavior. This dissertation discusses the design and implementation of Gander's middleware model, which abstracts away tedious implementation details by enabling developers to write high-level rules that govern when, where, and how data is distributed and to execute expressive queries across proximate digital resources. We evaluate Gander within several simulated large-scale environments and one real-world deployment on the UT Austin campus. The goal of this research is to provide formal constructs realized within a software framework that ease the software engineering challenges encountered during the design and deployment of several applications in emerging mobile environments.