Time Synchronization In Network-centric Sensor Networks
Raje, Sejal D
MetadataShow full item record
Time synchronization is a crucial component of infrastructure for wireless sensor networks (WSN). Most applications of WSNs make extensive use of time synchronization, for accurate timestamping of events, coordinate activities of the network, data fusion, object tracking or TDMA scheduling. The unique requirements of sensor networks in terms of precision, lifetime, energy and scope of the synchronization achieved, calls for design of new synchronization methods for WSNs. In this thesis, we describe the design of three novel techniques for achieving synchronization in network centric scenario of sensor networks: sequential least squares, Kalman filter and fuzzy logic systems. The strength of these techniques for prediction of parameters in a dynamic real-time case is exploited in the current problem. These techniques achieve a highly accurate, long-term and adaptive synchronization by forming a time conversion scale between clocks of two nodes in the network, one of which is the sender of timestamps and other is the receiver. A comparison of these approaches is also presented. An outline of a protocol based on these techniques is suggested, which is highly energy-efficient, lightweight, multimodal, tunable, scalable and which can provide relative or absolute notion of time.