Research On Adaptive Time Synchronization Algorithms For Wireless Sensor Networks

Time synchronization is fundamental and critical for a distributed sensor system to set up a common timescale for wireless sensor networks(WSNs).For years,time protocol for network such as Network Time Protocol(NTP)has kept the network clock perfectly synchronized.However,due to the energy limitation,scalability,large density,and high precision requirements of sensor nodes,NTP is not suitable for WSNs.Therefore,other synchronization protocols have been proposed to meet the demands of WSNs.For instance,Reference Broadcast Synchronization(RBS)and Timing-sync Protocol for Sensor Networks(TPSN)are two widely used algorithms.In this paper,we research on the RBS and TPSN algorithms and propose a new method based on RBS,which introduces an adaptive synchronization interval by exploiting the skew estimation information between two sensors.In the new method,to reduce the number of transmitted synchronization packets between two sensors,we introduce the concept of broadcast step size for time message exchange,where sensors record the arrival time of broadcast and exchange only once in every broadcast step size.A silent interval is derived by each pair of sensors after every synchronization period.Simulation results show that the proposed mechanism can reduce synchronization overhead without much precision loss.Moreover,we propose an adaptive mechanism for a multi-hop TPSN topology to reduce synchronization cost.We firstly study the cumulative errors in a multi-hop TPSN topology.Then by taking into account the clock offset and skew estimation errors in each level,we model these errors using Gaussian distribution.By using the new method,cumulative errors will not blow up according to our simulation.We also found that after 4 hops errors cannot be further lowered even more synchronization packets are sent.The proposed method provides a good solution considering the trade-off between precision and energy efficient in synchronization.