Abstract

Smart environments can be created with the help of the Internet of Things (IoT) technologies that have spread at such a pace that allows intelligent monitoring and management of vital resources. Nevertheless, an attainment of sustainable development in such settings is of great concern because of the dynamic resource needs, heterogeneity of systems and the constraints of the management strategies which are either statical or rule based. To combat such requirements, the paper will set forth a novel computational paradigm, integrating Artificial Intelligence and Genetic Algorithms, which were called a GA-Optimised Deep Neural Network (GA-DNN), to act as a sustainable resource management tool in a smart IoT-based setting (Zhang & Tao, 2020). The proposed solution incorporates a deep neural network which is used to model complex nonlinear relationships of the data provided by IoT sensors and other system control variables, the genetic algorithm is used to optimise the network parameters and hyper parameters to improve prediction efficiency and decision efficiency. The framework will be used to help optimise toward a multi-objective where the use of energy, resource utilisation efficiency, system latency, and sustainability influence can be taken into account. To test the effectiveness of the proposed GA-DNN model, a simulated multi-zone smart environment based on the simulated conditions of the real-world IoT deployment is used. The results of the experiment confirm that the suggested solution provides considerable increases in resource efficiency and sustainability indicators over current deep learning or rule-based methods of IoT management, and low compute latency that can be used in real-time work. The results point to the possibility of hybrid AI-based computational models as an efficient and scalable solution to the next generation of sustainable smart IoT environments.