Quantum LSTM-driven IoT framework for real-time monitoring of dissolved oxygen in aquaculture systems: a case study in Andhra Pradesh, India

The graphical abstract depicts a workflow where water quality data is collected and preprocessed, processed through a Water Quality Index (WQI), and analyzed using a QLSTM model to classify dissolved oxygen levels into hypoxia, anoxia, and normoxia categories. Dissolved oxygen (DO) monitoring in real-time is an important parameter to maintain aquatic health and promote the sustainability of aquaculture systems. Conventional deep learning methods are promising, but tend to possess major drawbacks, such as poor performance with time-series data of a long-range dependency, data imbalance sensitivity, and insufficient stability of the method in the noised and incomplete Internet of Things (IoT) setting. To overcome these challenges, this article develops a framework for continuous monitoring of DO with the help of IoT and an advanced Quantum Long Short-Term Memory (QLSTM) model to classify DO in aqua ponds. The time-series data collected from fishponds are labelled using the Water Quality Index (WQI), namely hypoxia, anoxia, and normoxia. The proposed QLSTM model classifies the water quality data into different DO conditions for fish. The proposed model is better in handling long-range time dependencies and well suited for imbalanced and time-sensitive applications. Performance results demonstrated that the performance of proposed QLSTM has attained classification accuracy of 92.4%, precision, recall, and f1-score of 90.5, 91.8, and 91.1%, respectively, with a less-computation time compared to the state-of-the-art models. The outcomes underscore the model efficiency and its applicability in real-time decision-making processes in aquaculture management.