Protocolo ciente de correlação espacial para redes de sensores sem fio
FAVARIN, Gilmar. Protocolo ciente de correlação espacial para redes de sensores sem fio. 2011. 94 f. Dissertação (Mestrado em Ciências Exatas e da Terra) - Universidade Federal de São Carlos, São Carlos, 2011.
The usage of wireless sensor network is increasingly being applied to people s everyday lives everywhere: from energy consumption in households and buildings in general, to vital signs in assistive medicine, infrastructure monitoring, chemical or biological product leaking detection in industries, better surveillance, environmental monitoring, among many others. WSN can be deployed in different densities next to several thousands of nodes. However, the development of WSN solutions are limited mainly by energy resource restriction. The great challenge to WSN solutions is to increase the network longevity while guaranteeing data delivery, reliability and accuracy in an environment prone to different types of failures. The largest source of energy consumption is data transmission. Thus, solutions to WSN needs to avoid intense communication keeping energy consumption balance and so the network longevity. In applications in which high density of nodes is necessary, sensing process can produce a large amount of data which are similar or redundant, due to the special proximity among the nodes. This spatial proximity can be explored in routing solutions to reduce the amount of messages transmitted throughout the network. This work presents the Spatial Correlation Aware Routing Protocol - SCARP , which makes use of spatial correlation to reduce the number of network transmissions. With SCARP, the WSN is configured in cells and nodes of each cell are selected, in an alternated way, to transmit similar or redundant data, and so reducing the number of transmitted messages. This traffic reduction results in less energy consumption and longer network longevity. Evaluation results show that SCARP outperforms similar solutions described in the literature, such as DAARP, which uses clustering and aggregation. SCARP has a positive performance even for large node density scenarios.