Research And Implementation Of Time Synchronization Algorithms In Wireless Sensor Networks

When the wireless sensor network(WSN)is used in the live detection and fault location of the transformer substation,the higher the time synchronization accuracy is,the closer to the real working condition the information we deduced from the collected data is,which ensures the accurate fault location.However,compared with the tradition networks,WSN owns limited calculating ability and power,which makes it significant to design a highly accurate synchronization method within the calculating ability.The traditional Least Square Flooding Time Synchronization Protocol(LS-FTSP)uses the least square method to estimate the clock drift and clock skew.It doesn't take into account the following two points.One is that when the interval between two neighbouring timestamps is large,the observation noise in the old timestamps will exert an influence on the estimation of the parameters.The other is that the least square method is susceptible to the abnormal data,which may bring about an erroneous result or low synchronization accuracy if abnormality occurs.This thesis proposes Finite Weighted Least Square FTSP(FWLS-FTSP)and Abnormality Reduced Least Square FTSP(ARLS-FTSP)to improve the least square method.In FWLS-FTSP,the observation noise is taken into consideration to improve the synchronization accuracy through giving the data with larger noise a smaller weight.In ARLS-FTSP,the abnormal data points are specially dealt with,which also improves the accuracy.Besides,to reduce the influence from the Clear Channel Assessment(CCA)on the hardware platform,this thesis utilizes the SFD(Start of Frame Delimiter)capture mechanism to obtain timestamps from the physical level instead of getting timestamps from the MAC(Media Access Control)level.The simulation result shows that FWLS-FTSP possesses a synchronization error of 1.51?s,an appreciable improvement compared to LS-FTSP's 4.99?s under the same simulation condition.Under the simulation condition with abnormal data,LS-FTSP's synchronization error is 17.53?s,while ARLS-FTSP reduced the error to 5.59?s.CC2530 nodes are used as the platform to verify ARLS-FTSP based on SFD capture.The verification result shows a 0.231?s synchronization error and an obviously slower increment of the clock difference caused by the clock drift.