Data fusion prolongs the lifetime of mobile sensing networks

Mobile sensing accelerates the integration of physical world and virtual space by using the pervasive portable devices with sensing and communication abilities. Since the smart devices are often battery powered, prolonging the network lifetime in mobile sensing becomes very important. Previous works forward packets individually, resulting in a large amount of redundant copies. They therefore consume much energy. We notice that in many applications of mobile opportunistic networks, the sensory data are spatial-temporal correlations. The correlated data can be aggregated in the forwarding process, thus reducing the number of copies and saving energy. Considering this fact, we propose two forwarding schemes by integrating data fusion: Epidemic with Part Fusion (EPF) and Epidemic with Complete Fusion (ECF). The part fusion scheme is responsible for aggregating raw correlated data, and the complete scheme can fuse any type of correlated data (raw or fused data). We give the closed form of the dissemination law of raw data and fused data, respectively. The scaling law theoretically guarantees that the two schemes achieve better tradeoff between mean delivery delay and energy consumption. We evaluate the fusion scheme with synthetical and real traces. The experimental results demonstrate that the EPF can save energy by 55%, and the ECF reduces energy consumption by 80% compared with the non-fusion scheme.