Redução de dados em redes de sensores sem fio baseada em stream de dados

Abstract

In the world there are a variety of phenomena, such as temperature, pressure, and humidity, which can be monitored by specific sensor devices with processing and communication power. These devices, working cooperatively, are known as wireless sensor networks. Each sensor node can monitor and report some phenomena to a special node called sink node, using a wireless communication.Despite their potential applications, wireless sensor networks have particular features imposed by resource restrictions, such as low computational power, reduced bandwidth and especially limited power source. Specifically, when we have a lot of data to be sent, the reduced bandwidth problem is increased since more nodes will try to access the wireless medium generating a packet delay. Thus, some data reduction is necessary where only the data relevant to the application is used.In wireless sensor networks, the monitored phenomena have data stream characteristics (online, imprecise, with noise, and of moderate size). To process a data stream there are some techniques such as sampling, histogram, sliding windows, and sketch. This techniques allow the data processing and reducing where the network requirements, like energy consumption and packet delay.Thus, the general problem treated here is the stream-based data reduction in wireless sensor networks so the network resources are saved, but at the same time we are interested to have a minimum data quality that represents the monitored phenomena. Our solution, proposed for this problem is a generic architecture that can be applied to general applications. This architecture has an API that allows to apply data reduction techniques to stream-based applications in wireless sensor networks. We use this architecture to model some applications that need to reduce the data at the sensor nodes, cluster heads, or routing nodes.The results show that it is possible to use our solution in general applications, leading to reduction in both energy consumption and packet delay without loosing the data representativeness. Furthermore, when the architecture is integrated to the routing phase on real-time applications the results show that it is possible to achieve the deadline and keep the information quality about the monitored phenomena.