Data aggregation is an important mechanism to reduce energy consumption in Wireless Sensor Networks (WSNs). By investigating spatial and/or temporal correlation of raw data, sensor nodes can aggregate raw data to a meaningful digest instead of directly sending raw data to sink, this process is considered as data aggregation. Several aggregation works focus on the raw data, they use raw data to cluster the nodes or to do aggregation. While analysis of datasets of real projects shows that some nodes perform similar evolution. Thus we propose a raw data-independent aggregation, i.e., Similar-evolution Based Aggregation (Simba), to consider the evolution of data rather than the raw data. Simba creates a group out of isolated nodes, nodes in the group can cooperatively execute data aggregation, this process reduces the energy consumption on each node. Besides, similar evolution of nodes guarantees the recover accuracy. Our experiments demonstrate that Simba can save more than 91% energy comparing no aggregation, and save more 30% energy than original aggregation functions, and Simba can recover data with higher fidelity comparing with the works relying on raw data.