Proactive Data Allocation in Distributed Datasets based on an Ensemble Model

Abstract

The evolution of the Internet of Things (IoT) drives the incorporation of numerous devices into a huge infrastructure where various services can be provided. Devices are located close to users being capable to interact with them and their environment to collect data. The collected data are transferred to the Cloud through the ‘intervention’ of the Edge Computing (EC) infrastructure. Multiple nodes are present at the EC that can undertake the responsibility of keeping some processing activities close to end users, thus, minimizing the latency in the provision of responses. In this paper, we elaborate on a model that supports an efficient data management mechanism to proactively decide the location where data should be stored. We aim at concluding a number of datasets exhibiting a high accuracy as exposed by their solidity. The proposed approach deals with the similarity of the collected data and the already formulated datasets before we decide the final allocation. Any decision is made upon the synopses of the discussed datasets avoiding the processing of huge volumes of data. Additionally, we elaborate on an ensemble scheme for matching the incoming observations and the available synopses. The performance of the proposed scheme is depicted by the relevant numerical outcomes.