Open Collections

Hunter, Gerren Brook

Due to the hardware constraints of embedded systems, software optimizations and binary size reductions may improve performance. Data processing is challenging, as the current state-of-the-art for embedded databases is custom-coded solutions that lack the simplicity of use and deployment compared to relational systems. This work details improvements to EmbedDB, an embedded key-value, time-series database specialized for sensor nodes and Internet of Things devices. By modifying EmbedDB to efficiently retrieve buffered records and creating an analogous distribution strategy to SQLite, the aim is to make EmbedDB an attractive option for systems constrained by hardware limitations. To do this, this work details improvements by preventing frequent flushes to the file system, ease of distribution by amalgamating source code, and exploring compiler optimizations benefiting from source code contained inside a single translation unit. Further exploring compiler auto-tuning techniques may reduce binary size or optimize performance.

Text

University of British Columbia. COSC 449

Vancouver : University of British Columbia Library

10.14288/1.0442409

Science, Irving K. Barber Faculty of (Okanagan); Computer Science, Mathematics, Physics and Statistics, Department of (Okanagan)

Undergraduate

DSpace