Cost-Effective Space Partitioning Approach for IoT Data Indexing and Retrieval
Ibtissem Kemouguette, Zineddine Kouahla, Ala‐Eddine Benrazek, Brahim Farou, Hamid Séridi · 2021
IoT technology implies a huge group of connected devices to capture a tremendous quantity of information. Computing and storing this huge amount of data requires scalable and efficient indexing solutions. Among the latest indexing structures in the BCCF-tree metric space, one framework is focused on providing recursive clustering of the space using the k-means algorithm. This solution suffers from data overlap between space partitioning at large scale. This study aims to optimize the structure of the BCCF. We propose to replace the k-means algorithm with an algorithm for estimating the overlap rate between the partitions of the space. Experimental results show good performance of the construction and search algorithms on several types of datasets.