An Efficient Node Selection Metric for In-network Process Deployment

In-network processing is a powerful technique for reducing network traffic in an ad hoc network where network efficiency is a critical issue. When an in-network process collects data from multiple data sources, the node hosting the in-network process should be carefully selected to reduce network traffic. Existing metrics used to select the host node are unsatisfactory in this case, because they do not consider differences in the amount of data provided by each data source. In this paper, we propose a node selection metric called COLOR to solve this problem. COLOR value is derived from locations of data sources and the amount of data provided by them so that a data source that provides more data than the others has a stronger effect. Moreover, the communication overheads associated with COLOR are small, because parameters involved by COLOR can be collected during a data retrieval phase, which generally occurs in in-network processing. Simulation results show that data retrieval using COLOR produces less network traffic than that retrieved using existing metrics in environments where placements of data sources and the amount of data are nonuniform.