A UKF-based algorithm for asynchronous data fusion of wireless sensor is proposed. The RNAT mechanism is used to identify redundant nodes in the wireless sensor network, and a data redundancy tree is constructed to implement redundant data removal. According to the result of deduplication, in the environment where the radius of each sensor's detection range is equal, two nodes that detect the target information are arbitrarily selected by the four-circle positioning method to calculate the intersection of the boundaries of two circular detection regions, finding and approximating the target according to the iterative method. The preconditions that the original sensors of different sensors are independent of each other and the different original quantity measurements of the same sensor are independent of each other are set, and the measured values of each channel are calculated. The unmeasured Kalman filter is used to update the measured value in the form of filtering, the Kalman filter gain matrix is introduced, and the asynchronous data positioning result is combined to realize data fusion. The experimental results show that the data utilization after fusion is higher than the current results, and the algorithm has short time-consuming, low energy consumption, and high data fusion precision. The accuracy of the whole fusion is more than 90%.