A Key-Value-based Persistence Model for Sensor Networks


Marcello A. de Sales, Jr.

Oral Defence Date: 

Thursday, December 17, 2009 - 16:10


SCI 241


4:10 pm


Professors Puder and Murphy


Sensor Networks are becoming important to different scientific and industrial communities because of what they produce: the raw data of diverse domains. In order to make use of the collected data, researchers may have to dissect the characteristics of the sensor network in question, regarding different properties such as the purpose and location of the observed data, as well as how the data is described. For this reason, this work's first contribution is a set of data persistence taxonomies based on the state of the art of Data Persistence for Sensor Networks, which can be used to classify the properties of the produced raw data. In order to evaluate the proposed taxonomies, a data persistence component for NetBEAMS was designed and implemented. NetBEAMS is a component-based sensor network infrastructure developed to improve the operation of the SF-BEAMS environmental sensor network. SF-BEAMS is deployed in Tiburon, California, and managed by the Romberg Tiburon Center (RTC). Based on an empirical analysis regarding proposed taxonomies, a Key-Value Data Model is proposed as an alternative to the Relational Data Model traditionally used. Furthermore, the mongoDB database, a schema-less document-oriented database, was selected for evaluation. Results based on the experiments suggest a novel approach to provide External or Data-Centric persistence for networks. Similarly, the literature supports programming languages as a better abstraction when it comes to data access and modification by non-technical users such as biologists. Finally, this report lends itself to future work in the area of data persistence in sensor networks.

Marcello A. de Sales, Jr.

Environmental Sensor Network, Data Persistence, Taxonomies,KVP Data Model, Document-Oriented Databases, KVP Databases, SF-BEAMS, NetBEAMS