Wireless sensor networks（WSNs）always have to face the dilemma between limited battery energy and large quantity of data transmission. Breaking data stream into blocks has proved to be an effective way to increase the utilization of the network. However, there is still potential improvement to be achieved to fuse the redundant information, which can be intra-frame or inter-frame. In this article, a new approach-Dynamic Data Fusion Mechanism (DDFM) is described and analyzed. DDFM is based on LEACH protocol, using a shorter data pattern to represent a block in transmission so as to reduce the flow. Compared with other aggregation methods, DDFM better guarantees the QoS. The specific message formats, the algorithm and the parameters such as frame size and block size are devised and tested in the following passages. And from the simulation, we prove the effectiveness of DDFM in realizing data fusion with remarkable compression ratio and thus expanding the lift-time of the networks.