Abstract: Internet of things has been increasing at a tremendous rate since last few years. Thus, it has been an active area of research. Data from a single node can provide enough information for various conditions on the field. However, the data from the IoT is used to get an inference on a high level which can be used to take corrective actions. To take the corrective actions, the data should be mined properly which in turn depends on correct data from the sensor. Data fusion techniques are used to provide correct data from the sensor to the data mining algorithm. Data fusion in IoT is not a well-researched topic. In this design, various data fusion techniques are examined, and a hierarchical approach for data fusion in IoT is proposed. The design uses Fuzzy Kalman Filter for state estimation, and Dempster- Shafer method for decision fusion to create a dynamic context-aware system. The proposed design is also scalable for a higher number of nodes in the network which is not found in all implementations. The research also offers a design metrics which can be used for comparison of different data fusion.