Deep Learning is a high-performance machine learning approach that combines supervised machine learning and feature learning. It is built of a sophisticated models with numerous hidden layers and neurons to create advanced image processing models. DL has proven its effectiveness and resilient in different fields including big data, computer vision, image processing, and many others. In agriculture, rice leaf infections are a frequent and pervasive issue that lower crop and output. This research proposed a reduced form of Capsule Network (Caps NET), a form convolutional neural network, for the classification of rice leaf disease. The goal of the suggested Caps NET model was to assess the suitability of various feature learning models and enhance deep learning models' capacity to learn about rice leaf disease classification. Caps NET was fed images of both healthy and infected leaves. High classification performance was obtained with the ideal configuration (FC1 (960), FC2 (768), and FC3 (4096)), which had 96.66% accuracy, 97.25% sensitivity, and 97.49% specificity.
Read MoreDoi: https://doi.org/10.54216/JISIoT.140201
Vol. 14 Issue. 2 PP. 01-07, (2025)
This study presents a new adaptive routing protocol for fire emergencies, leveraging a newly created dataset and a hybrid deep learning approach to optimize decision-making and data routing strategies. The developed protocol integrates a hybrid of Convolutional Neural Networks (CNNs) with Bi-Directional Long Short-Term Memory (BiLSTMs) deep learning models to predict fires at early stages, effectively managing the dynamic and unpredictable nature of fire emergencies to prevent data loss and ensure packet delivery to the base station. Exhaustive validation was conducted utilizing the standard protocol to ensure the reliability and effectiveness of the proposed approach. Experimental results demonstrate the superior performance of the proposed hybrid-deep learning model and the significant enhancements in routing efficiency and monitored data preservation for the developed protocol compared to the standard protocol. The findings are useful in providing a reliable solution for adaptive routing during emergencies.
Read MoreDoi: https://doi.org/10.54216/JISIoT.140202
Vol. 14 Issue. 2 PP. 08-24, (2025)
Current industrial control systems are increasingly integrating with corporate Internet technology networks in order to fully utilize the abundant resources available on the Internet. The growing connection between industrial control systems and the internet has made them a desirable choice. Industrial control systems are in need of significant protection due to being a common target for a range of cyber-attacks. The use of the Internet of Things is currently increasing across industries due to its efficiency, and the Internet of Things is facing a security challenge. This document gives an overview of the intrusion detection system and the methods of the intrusion detection system. The purpose of this document is to examine intrusion detection methods and present the best method based on studies. Experimental results show that this system uses a combination of machine learning methods for high performance.
Read MoreDoi: https://doi.org/10.54216/JISIoT.140203
Vol. 14 Issue. 2 PP. 25-35, (2025)
Wireless sensor networks have been identified as one of the most important technologies. A vast amount of research and development has been devoted to this area in the past decade. Nowadays, they have been applied in various fields including environment monitoring, smart building, medical care, and etc. With the advances in electronics, wireless communications, and sensor technology, more and more new opportunities have been created for the research in wireless sensor networks. However, the successful implementation of WSN faces many challenges, such as limited power, limited memory, and limited computing capability. Among them, limited power is the most critical restriction because it is usually impossible for the battery-powered sensor nodes to be recharged. Therefore, one of the main areas of interest for wireless sensor network research is how to reduce power consumption. The proposed system classifies sensor nodes into two operational modes, optimizes node deployment, and finds optimal node placements using a genetic algorithm (GA) to minimize the energy consumption of the WSN. The system's successful testing on a simulated WSN meant for radiation site identification revealed its potential for practical real-world applications.
Read MoreDoi: https://doi.org/10.54216/JISIoT.140204
Vol. 14 Issue. 2 PP. 36-43, (2025)
Artificial intelligence (computer-based intelligence) is advancing significantly in all areas and applications of life at a high speed. The use of modern technologies has become a necessity in daily life, and smart systems have entered daily life, especially in the design of smart homes. Smart homes linked to man-made intelligence mimic the way residents live and facilitate many activities and services. Although some studies have shown how smart homes use computer-based intelligence, few applications have been reported for integrating smart technologies into installation and use of the Internet of Things. In this research, the basic problems in adaptive smart home systems, such as the development of the smart home and its synchronization with the Internet of Things, and “what is the relationship between analysis and adaptation in smart homes with simulation of intelligence algorithms” were addressed to be the focal point of this paper. Moreover, this study aims to depict the capabilities and elements of artificial intelligence in improving the performance of smart homes. In order to understand how to use artificial intelligence to build smart homes, the precise situation of applying artificial intelligence in smart home elements and the way applications are used in homes was determined. We simulated a multi-service smart home environment by designing an efficient, multi-purpose artificial intelligence algorithm to improve the control level and enhance the performance of smart home services.
Read MoreDoi: https://doi.org/10.54216/JISIoT.140205
Vol. 14 Issue. 2 PP. 44-61, (2025)