Wireless sensor networks (WSNs) should collect accurate measurements to reliably capture the state of the environment that they monitor. However, measurement data collected from one or more sensors may drift or become erroneous due to hardware failures or sensor degradation. In WSNs with remote deployments, detecting those measurement errors through a centralized reporting approach can result in a large number of message transmissions, which in turn dramatically decreases the battery life of sensors in the network. In this thesis, we address this issue through three main contributions. First, we propose a protocol in which sensors detect errors in a peer-to-peer (P2P) fashion, and that extends the life of the WSN by minimizing the number of messages transmitted. Second, we propose an e ective anomaly detection approach that has low memory and processing requirements, allowing for easy deployment on low-cost sensor hardware. Third, we develop a trace-driven, discrete-event simulator that allows us to evaluate the developed protocol and approach. In doing so, we use three datasets from real WSN deployments, which include indoor air temperature, sea surface water temperature and seismic wave amplitude sensors. Our results show that our P2P protocol can accurately detect errors and simultaneously extend the e ective WSN lifetime dramatically compared to the centralized protocol.
【 预 览 】
附件列表
Files
Size
Format
View
An energy-efficient P2P protocol for validating measurements in wireless sensor networks
PDF
download
878987797
028-85220240
