**6.2. Future work**

For the future work, at first, we will discuss the random pushing rate to adapt the various changes of data generation rate and information content. For example, in an MRSN, when it is of the state of affairs, nodes generate much more event data than in normal case, that means data generation rate becomes larger. In this case we should decrease the random pushing rate to control the amount of data transmission. On the other hand, from the view point of information entropy, if the self information of generated data is high, it means the generated data are rare generating data. However, when a node applies the normal data aggregation and aggregates the data with normal data, the aggregated data cannot reflect the real situation which may lead bad result. In this case, we can increase the random pushing rate to send high self information data immediately without data aggregation. When the self information of generated data decreases, we decrease random pushing rate to control the quantity of data transmission. Secondly, in wireless sensor network, data are transmitted to sink node by multi-hopping way, which causes the uneven energy consumption on nodes at different locations. Hence, to keep all nodes in the network having the same energy consumption is our another future work.
