Research On Time Synchronization Algorithm In Wireless Sensor Networks

Recent advances in micro-electro-mechanism system, on-chip sensor, wireless com-munication and network technologies have led to active research in wireless sensor net-works. As a support technology of wireless sensor networks, the time synchronizationmechanism has also attracted serious concern and become a hot spot. The research inthis thesis focuses on three shortcoming aspects of the time synchronization mechanism,including the precision adaptive time synchronization algorithm design for cluster-basedmulti-user wireless sensor networks, the acceleration algorithm design for distributedconsensus time synchronization and the hierarchical consensus time synchronization al-gorithm design.Most proposed time synchronization algorithms are designed under the single-userassumption and achieve fixed precision performance. However, when serving in multi-user wireless sensor networks, the precision-fixed time synchronization algorithms willencounter energy ine?ciency problems. By in-cluster synchronization error decompo-sition and reformulation, the approximately proportional relationship of synchronizationerrors and synchronization periods and the condition to hold the proportional relationshipis found. Based on that, a lightweight precision adaptive time synchronization algorithmis proposed for cluster-based multi-user wireless sensor networks. The algorithm can findthe goal synchronization period faster and more accurately than the existed adaptive timesynchronization algorithms.The planar distributed consensus time synchronization algorithm has excellent ro-bustness and scalability performance. However, the slow convergence rate prevents thealgorithm being implemented in large-scale networks. By mapping the iteration processof the distributed consensus time synchronization algorithm into Markov domain, thedependence of the algorithm convergence rate on the neighbor-size distribution and thenetwork size in circulant networks and co-uniform networks is analyzed. Two acceler-ation algorithms, based on adjusting neighbor-size distribution, are proposed to reducethe convergence iterations of the distributed consensus time synchronization algorithm.Besides, how to decide the neighbor size for the acceleration algorithms to balance the ac-celeration performance and the node transmission power consumption is also discussed. In order to reduce the communication overhead of the planar distributed consensustime synchronization algorithm, two hierarchical consensus time synchronization algo-rithms, cluster-tree based and passive clustering based respectively, are proposed. Bycombining clustering technologies with the consensus time synchronization, the hier-archical consensus time synchronization algorithms can significantly reduce the com-munication overhead. Besides, since the passive clustering technology can shorten thecritical-path-length of the consensus synchronization route, which is benefit for reducingnetwork synchronization errors and accelerating the convergence of the consensus timesynchronization algorithm, the passive clustering based consensus time synchronizationalgorithm also achieves better network synchronization error and synchronization con-vergence rate performance than the planar distributed consensus time synchronizationalgorithm.