A distributed, energy-efficient and QoI-aware framework for in-network processing

In-network processing (INP) is a promising method that allows aggregation of data while it is being transferred along the communication paths as a means to optimize the utilization of network resources without violating the quality of information (QoI) requirements. Given the large amount of data existing in dynamic environments, the optimization of INP requires a distributed framework that can adapt easily to network changes and user requirements. In this work, we develop the principle for designing a distributed mechanism in order to determine and control INP. Specifically, the proposed framework can decide, in a distributed way, which nodes along the communication paths optimally perform INP, with consideration of operational energy consumption and QoI requirements for achieving global optimal INP. The significance of the proposed distributed method is that it requires each node to make independent decisions locally for data aggregations, thus naturally enhance robustness and efficiency against network and data load dynamics. Extensive numerical results are presented to confirm the validity of the proposed approach.