Data centric storage framework for an intelligent wireless sensor network

Ahmed, K 2014, Data centric storage framework for an intelligent wireless sensor network, Doctor of Philosophy (PhD), Electrical and Computer Engineering, RMIT University.

Document type: Thesis
Collection: Theses

Attached Files
Name Description MIMEType Size
Ahmed.pdf Thesis application/pdf 3.26MB
Title Data centric storage framework for an intelligent wireless sensor network
Author(s) Ahmed, K
Year 2014
Abstract In the last decade research into Wireless Sensor Networks (WSN) has triggered extensive growth in flexible and previously difficult to achieve scientific activities carried out in the most demanding and often remote areas of the world. This success has provoked research into new WSN related challenges including finding techniques for data management, analysis, and how to gather information from large, diverse, distributed and heterogeneous data sets. The shift in focus to research into a scalable, accessible and sustainable intelligent sensor networks reflects the ongoing improvements made in the design, development, deployment and operation of WSNs. However, one of the key and prime pre-requisites of an intelligent network is to have the ability of in-network data storage and processing which is referred to as Data Centric Storage (DCS).

This research project has successfully proposed, developed and implemented a comprehensive DCS framework for WSN. Range query mechanism, similarity search, load balancing, multi-dimensional data search, as well as limited and constrained resources have driven the research focus. The architecture of the deployed network, referred to as Disk Based Data Centric Storage (DBDCS), was inspired by the magnetic disk storage platter consisting of tracks and sectors. The core contributions made in this research can be summarized as: a) An optimally synchronized routing algorithm, referred to Sector Based Distance (SBD) routing for the DBDCS architecture; b) DCS Metric based Similarity Searching (DCSMSS) with the realization of three exemplar queries – Range query, K-nearest neighbor query (KNN) and Skyline query; and c) A Decentralized Distributed Erasure Coding (DDEC) algorithm that achieves a similar level of reliability with less redundancy. SBD achieves high power efficiency whilst reducing updates and query traffic, end-to-end delay, and collisions. In order to guarantee reliability and minimizing end-to-end latency, a simple Grid Coloring Algorithm (GCA) is used to derive the time division multiple access (TDMA) schedules. The GCA uses a slot reuse concept to minimize the TDMA frame length. A performance evaluation was conducted with simulation results showing that SBD achieves a throughput enhancement by a factor of two, extension of network life time by 30%, and reduced end-to-end latency. DCSMSS takes advantage of a vector distance index, called iDistance, transforming the issue of similarity searching into the problem of an interval search in one dimension. DCSMSS balances the load across the network and provides efficient similarity searching in terms of three types of queries – range query, k-query and skyline query. Extensive simulation results reveal that DCSMSS is highly efficient and significantly outperforms previous approaches in processing similarity search queries. DDEC encoded the acquired information into n fragments and disseminated across n nodes inside a sector so that the original source packets can be recovered from any k surviving nodes. A lost fragment can also be regenerated from any d helper nodes. DDEC was evaluated against 3-Way Replication using different performance matrices. The results have highlighted that the use of erasure encoding in network storage can provide the desired level of data availability at a smaller memory overhead when compared to replication.

Degree Doctor of Philosophy (PhD)
Institution RMIT University
School, Department or Centre Electrical and Computer Engineering
Keyword(s) data centric storage
sector based distance routing
distributed erasure encoding
similarity searching,
skyline query
K-NN query
range query
grid coloring algorithm
Version Filter Type
Access Statistics: 369 Abstract Views, 401 File Downloads  -  Detailed Statistics
Created: Fri, 14 Nov 2014, 13:36:42 EST by Denise Paciocco
© 2014 RMIT Research Repository • Powered by Fez SoftwareContact us