Performance Evaluation of Indices-Based Query Optimization from Flash-Based Data Centric Sensor Devices in Wireless Sensor Networks

Abstract

Flash memory has become a more widespread storage medium for modern wireless devices because of its effective characteristics like nonvolatility, small size, light weight, fast access speed, shock resistance, high reliability, and low power consumption. Sensor nodes are highly resource constrained in terms of limited processing speed, runtime memory, persistent storage, communication bandwidth, and finite energy. Therefore, for wireless sensor networks supporting sense, store, merge, and send schemes, an energy efficient and reliable database-based query optimization technique is highly required with consideration of sensor node constraints. Databases on hard disk drives perform data storage and retrieval using index structures which are still not practiced for sensor devices. In this paper, we evaluate different indices like B-tree, R-tree, and MR-tree by implementing them on log structured external NAND flash memory-based advanced file systems for supporting energy efficient data storage and query optimization from flash based data centric sensor devices in wireless sensor networks. Experimental results show that PIYAS (Rizvi and Chung, 2010) file system along with B-tree indexing deployed on flash memory MLC gives the significant performance in respect of high query throughput optimization and less resources consumption for wireless sensor devices.