Joint Estimation And Compensation Of Clock Offset And Clock Skew In Wireless Senor Network

Clock synchronization among different nodes, which have their own autonomous clocks, is a key component of a wireless sensor networks(WSNs). Many coordinated applications require a common time frame for the entire network such as node localization, data fusion, sleeping and wake-up coordination, time-based channel sharing and scheduling, and other time-based tasks. However, every individual node in a WSN has its own clock. Different clocks drift from each other due to imperfections in the oscillator, aging, digital circuit design and other environmental variations. Also, sensor nodes are usually powered with just a battery. Thus, all tasks of a WSN, including synchronization, should be carefully performed to ensure longer operating lifetime。In order to make full use of the broadcast communication feature of WSNs, two synchronized message exchange models are presented based on Pairwise Broadcast Synchronization Protocol(PBS). Different from other synchronization approaches, a subset of sensor nodes are synchronized by overhearing the timing message exchanges of a pair of sensor nodes. Therefore, a group of sensor nodes can be synchronized without sending any extra messages. And an energy-efficient clock synchronization algorithm is proposed. The clock offset and skew estimation problem is formulated as a linear least squares problem using auxiliary variables and can be solved by LS algorithm. Also the closed-form solution is obtained. Then, the mean square errors (MSE) of proposed LS algorithm, Maximum Likelihood Estimation(MLE) and Cramer-Rao lower bound (CRLB) are compared. Finally, simulation results demonstrate that the proposed LS algorithm, is low complexity and energy-efficient for Duty-cycle WSNs, also avoid convex or log-convex of MLE.For the co-collaboration feature of WSNs, clock skew, clock offset and node self-localization are joint estimated. Consider the Time of Arrival(TOA) measurement errors and communication delay, a mathematical model of TOA message with clock skew, clock offset and node localization is first established. The problem is complexity and nonlinear with 2D timing and 2D positioning parameter. So a pseudo-linear analysis method is used and can be solved by Weighted Least Square algorithm. Then, an error compensated CWLS algorithm is presented to reduce the estimation error because of the parameter coupling in WLS. Finally, the theoretical analysis MSE and CRLB are derived. Simulation results show that position estimation accuracy of CWLS algorithm is very close to the CRLB, and better than the LWLS. At the same time, CWLS overcome the defect that clock drift exceed a tolerable border with time increasing. This method allows that time synchronization does not need additional communication expenses, and thus saves energy.Finally, a clock skew prediction model based on its correlation to temperature is proposed, and a constrained least squares method is used to estimate the mathematically correlation. Then a synchronization protocol utilizing Temperature-Aware Compensation for WSNs is designed. Node can continuously adjust its local time according to the environment temperatures changes. The proposed protocol is a effectively complement for TPSN, FTSP and other protocols, thus making estimation of other protocol higher synchronization accuracy and lower computational complexity. The proposed protocol is also able in Duty-cycle WSNs, node can dynamically compensate the clock skew according to the working temperature. Finally, a set of experiments are conducted in STM32108w nodes and demonstrate that the Temperature-Aware protocol can greatly reduce the clock drift and prolong the ersynchronization intervals.