Optimization Driven Distributed Deep Learning for Aqua Status Prediction in IoT

Abstract

The continuous screening of the quality and characteristics of water in IoT (Internet of Things) is essential due to the increasing requirement on aquaculture in order to maximize the yields. There are various physicochemical parameters used in water quality monitoring, but the analysis of these parameters are needed to obtain the final decision with experts. This paper proposes an aqua status prediction model in IoT using the Fractional Gravitational Search Algorithm based distributed Deep Convolutional Neural network (FGSA-based distributed Deep CNN). Initially, the aqua parameters are analysed using the distributed IoT nodes in the aqua environment. The loss and delay in the transmission of the data related to the aqua status is controlled with the selection of cluster head optimally using the FGSA. The prediction of the aqua status is done with the distributed DeepCNN in the final step. The performance of the proposed method is analyzed with the evaluation metrics, namely accuracy, energy, and throughput. The accuracy, energy, and throughput of the proposed FGSA-based distributed Deep CNN classifier is obtained as 95.4758, 99.4293, and 99.9571, respectively, which is high as compared to the existing methods. This shows the effectiveness of the proposed method in the prediction of the aqua status.