Efficient SQL querying on embedded devices using pre-compilation

Microprocessors and embedded devices are used for data collection and analysis applications in infrastructure and en- vironmental monitoring, medical technology, wearable com- puting, and sensor network and mobile systems. Such appli- cations demand low energy solutions without using too much of a device's extremely limited RAM (1KB-100KB) and code space. Previously available database software for embedded devices and sensor networks relied heavily on data trans- mission across networks for centralized data processing. Re- cently, relational database systems for resource-constrained devices have been developed to execute queries on a per- device basis, which saves network transmission overhead. This work extends the applicability of such systems by lower- ing the code space and execution time requirements further through serializing queries at application build time and re- moving the query translation component from the device. By eliminating the need for complex query translation sys- tems on device, our technique can reduce ROM usage by as much as 50% while improving memory utilization. Our ex- periments demonstrate that pre-compiling can reduce query initialization times by 90% compared to typical parsing tech- niques. This translates to a further savings of up to 50% in on-device total execution times. The technique developed is applicable to a wide variety of embedded systems and