Research And Implementation Of Clock Synchronization In Wireless Sensor Network

Because of low cost, wide coverage, low power consumption, wireless sensor network(WSN)has found wide applications in many areas, including environmental monitoring, defense andmilitary, health care, home care, etc.Clock synchronization technology is an essentially underlying technique for WSN. In detail,all sensor nodes need to work together to complete the functionality of the system, such as nodelocalization algorithm, data fusion, TDMA scheduling, collaborative sleeping ect, which are allbased on the precise time synchronization. This thesis analyzes and studies the clocksynchronization algorithms in WSN. The main work is given as follows:Fristly, this thesis introduces the research background and significance of time synchronizationin WSN. Some basis theories are introduced briefly, such as the structure of wireless sensor network,key technology and the challenges, etc. Especially, the existing and classical clock synchronizationalgorithms are analyzed and summarized.Secondly, this thesis introduces the traditional TPSN (Timing-Sync Protocol for SensorNetworks) algorithm. That is, through carefully analyzing the error sourcrs in TPSN, TPSN-Balgorithm is proposed. Specifically, it is shown that the error caused by the messaging delay followsthe normal distribution with the mean of0, and thus Bayesian estimation is used to calibrate thepacket receiving time to obtain a more accurate vaule. Theoretically, TPSN-B doesn’t change thepacket messaging protocols, but just builds proper mathematical model to reduce the error ofreceiving time, which consumes negligible energy.Through simulations based on OMNET++, it is shown that, without having negative effect onenergy consumpation, TPSN-B algorithm through Bayesian estimation can effectively reduce thescope of packet receiving time errors, and improve the time synchronization than the original TPSNalgorithm.Finally, a further improved algorithm based on the TPSN-B is proposed. Generally, the error oftime synchronization may be caused by two sources: information transmission delay and theinstability of crystal oscillator frequency. TPSN-B algorithm deals with the error caused byinformation transmission delay, but TPSN-B doesn’t consider the clock drift caused by theinstability of crystal oscillator frequency. As a result, the nodes need to execute timesynchronization process continually in a short period, and will consume a lot of bandwidth and energy. Based on the proposed TPSN-B, this article introduced two kinds of mathematical modelsto estimate the best-fit line between time deviation and local time: the least square method (TPSN-L)and the chebyshev approximation algorithm (TPSN-C). Then we can obtain the value of the clockdrift rate and adjust the local time, without executing time synchronization process constantly. Thesimulation results indicate that, in comparison with TPSN-B, in the improved TPSN-L and TPSN-Calgorithms, energy consumption is reduced without affecting the clock synchronization accuracy.