The Internet of Things (IoT) has been widely applied in sensing systems, which connect various sensors and gather their information. Data fusion, which is to integrate multiple data representing the same real-world object into a consistent and accurate representation, is an important issue. Most of existing approaches of data fusion involve various analytical procedures whose high computational complexity usually requires massive computing resources. In this paper, an efficient data fusion approach is proposed. Stochastic models are presented to describe the measurements, and quantitative analyses are given to evaluate the efficacy of data fusion. Complex event processing engine is applied to improve the efficiency of data transmission and processing to meet the real-time demands. Finally, simulation experiments are conducted to validate the efficacy of the approach.