Efficient Data Management for Tiny Embedded Devices

点击翻译按钮获取中文摘要

The objective is to develop a comprehensive data management library for embedded devices that reduces their cost and complexity. Embedded devices are used in a vast variety of areas and represent a multi-billion dollar annual industry. These devices serve critical roles in our economy, industrial processes, environment, and personal lives. Developing applications for embedded devices is challenging due to their limited resources which restricts the type of operations that can be performed and motivates customized solutions that are costly to develop. A data management library that handles issues of data storage, consistency, transactions, and performance would greatly simplify embedded applications and make their development more economical. The scientific approach is to develop database algorithms specifically optimized for embedded devices that use flash memory. Our previous work constructed an optimized flash sorting algorithm, a low-memory flash translation layer and wear leveling scheme, and a complete sensor architecture that has been used for environmental monitoring and reducing water used in irrigation. The work will develop additional algorithms for joins and data aggregation and investigate novel approaches to query processing and optimization. A key goal is to integrate the database layer with the flash storage system layer as there is no holistic system that combines them. This project will research the limits on how efficient data processing can be on devices with minimal resources. The significance of the research is that constructing the data management library will provide an efficient data abstraction for embedded devices like SQL relational databases provided for computers. This will improve development costs and times and increase reliability. The research will be driven by specific applications, including irrigation management where water savings can be as high as 50%, agricultural and vineyard monitoring to improve yield and quality, and environmental monitoring and scientific data collection. These existing research partnerships will improve the research while contributing to the application areas.