Collaborative data reduction for energy efficient sensor networks

When we consider the limited power of wireless sensors, it is necessary to reduce the dimension of data conveyed between sensors, because high dimensional data transmission requires much power consumption of sensors. For data reduction in a network, in-network data aggregation methods and collaborative compression methods were reported. However, the in-network data aggregation methods typically lead to timedelay while performing data aggregation and compression, whereas the current collaborative methods primarily consider data redundancy between two sensors, not among multiple sensors. This paper proposes a new collaborative data reduction method to remove the redundancy existing in the data coming from mutiple sensors as well as from a single sensor. Our method uses a tree-based data propagation model to characterize the collaboration structure among multiple sensors. Our method also seperates the data aggregation process from the collaboration process in order to relieve the time-delay problem during aggregation processes. Thus, the time-delay is occured only during the periodic collaboration phase but not during the real-time data transmission. Our experimental results show that our method can reduce data transmission more effectively than the in-network aggregation without losing important information and without causing delay in aggregation.