Fast Node Clustering Based on an Improved Birch Algorithm for Data Collection Towards Software-Defined Underwater Acoustic Sensor Networks

In this paper, we take an in-depth study on the data collection scheduling issue in underwater acoustic sensor networks (UASNs). To address this issue, we propose the use of software-defined networking (SDN) technique, based on which the paradigm of software-defined underwater acoustic sensor networks (SD-UASNs) is proposed. With SD-UASNs, we propose an efficient data collection scheme on account of node clustering. We take the data scale into account, propose a fast node clustering approach based on an improved Birch algorithm including node pre-clustering and node re-clustering. In particular, for the node pre-clustering procedure, we improve the cluster feature tree of the Birch algorithm and limit the nodes in each cluster to be within an optimal distance. Based on the proposed pre-clustering approach, we propose a greedy data collection scheme that can heuristically seek optimal solutions when the compressed sampling is taken into account. Evaluation results demonstrate that our scheme performs better in network efficiency than some popular schemes.