Search

Rapid advances in deep learning and computer vision enable traditional cloud-based decision-making through edge computing with the Artificial Intelligent Internet of Things (AIoT) image sensors (AIoT-IS), thus improving the timeliness and security of image recognition. This study is indented to investigate the potential costs and benefits of AIoT-IS applications. This study summarizes AIoT-IS application scenarios for construction safety monitoring and proposes a cost/benefit analysis method for AIoT-IS implementation projects. According to the case study results, AIoTIS achieves significant benefits, with a Net Present Value Index (NPVI) of 19.17% and a Benefit/Cost Ratio (BCR) of 4.65 as applied to construction site safety monitoring. Interviews with domain experts also provided qualitative feedback, pointing to the directions for future research. The proposed method is applicable for the decision-making of AIoT-IS adoption and the feasibility assessment of other innovative construction technologies. [ABSTRACT FROM AUTHOR]

Industry 4.0 introduced new concepts, technologies, and paradigms, such as Cyber Physical Systems (CPSs), Industrial Internet of Things (IIoT) and, more recently, Artificial Intelligence of Things (AIoT). These paradigms ease the creation of complex systems by integrating heterogeneous devices. As a result, the structure of the production systems is changing completely. In this scenario, the adoption of reference architectures based on standards may guide designers and developers to create complex AIoT applications. This article surveys the main reference architectures available for industrial AIoT applications, analyzing their key characteristics, objectives, and benefits; it also presents some use cases that may help designers create new applications. The main goal of this review is to help engineers identify the alternative that best suits every application. The authors conclude that existing reference architectures are a necessary tool for standardizing AIoT applications, since they may guide developers in the process of developing new applications. However, the use of reference architectures in real AIoT industrial applications is still incipient, so more development effort is needed in order for it to be widely adopted. [ABSTRACT FROM AUTHOR]

This study investigates the application of Virtual Reality (VR) in the educational field, particularly its integration with GAI technologies such as ChatGPT to enhance the learning experience. The research indicates that while VR provides an immersive learning environment fostering student interaction and interest, the lack of a structured learning framework and personalized feedback may limit its educational effectiveness and potentially affect the transfer of VR-learned knowledge to physical hands-on tasks. Hence, it calls for the provision of more targeted and personalized feedback in VR learning environments. Through a randomized controlled trial (RCT), this study collected data from 77 university students, integrating experiential learning in VR for acquiring AIoT knowledge and practical skills, and compared the effects of traditional feedback versus GPT feedback on promoting reflective thinking, learning motivation, cognitive levels, and AIoT hands-on abilities among the students. The results show that the group receiving GPT feedback significantly outperformed the control group across these learning indicators, demonstrating the effectiveness of GAI technologies in providing personalized learning support, facilitating deep learning, and enhancing educational outcomes. This study offers new insights into the integration of GAI technology in VR learning environments, paving new pathways for the development and application of future educational technologies. [ABSTRACT FROM AUTHOR]

The construction industry of Ukraine shall not only recover but also upgrade, enhance and reevaluate existing building projects. Further research raises two pertinent issues for Ukraine - retrofitting and reconstructing destroyed infrastructure. The study's priority objective is to restore damaged and ruined buildings rapidly. It may be achieved using the creation of recovery methods in Ukraine and countries in the post-conflict stage of development. The research involves creating technical specifications for the product of a new version of the automated construction management system, which provides working with the software complexes based on the BIM model. The system implies using Building Information Modeling (BIM) and Artificial Intelligence of Things (AIoT) to make organization of reconstruction faster, better and less costly. The research has been held to demonstrate the viability of the approach. In addition, we acquire a reduction of energy consumption and an increase in the lifespan of the building by choosing retrofitting methods. The efficacy of BIM and IoT technologies enables the integration of contemporary demands to diminish design time and costs. These technologies also optimise design solutions by assimilating knowledge from previous building and structure designs. Additionally, they offer essential information support for the entire investment project life cycle. [ABSTRACT FROM AUTHOR]

This study examines how artificial intelligence (AI) can be incorporated into the Internet of Things (IoT), with a particular focus on using the NVIDIA Jetson Nano as a specialized AI platform. The aim is to boost the functional capabilities of IoT devices, allowing them to independently carry out sophisticated data analysis and decision-making tasks to heighten their efficiency and effectiveness. The paper begins by outlining the motivation for this investigation, then proceeds with a literature review to evaluate the prospective applications, challenges, and benefits of AI in IoT environments. It explores in detail the technical attributes of the NVIDIA Jetson Nano, including its compact design, robust computing power, and the availability of software libraries that facilitate the deployment of AI models on devices with limited resources. A prototype IoT device equipped with the NVIDIA Jetson Nano was developed at Tien Giang University. This device is capable of capturing video and executing AI algorithms for real-time analysis, thereby deriving crucial data for IoT decision-making processes. The outcomes and structure of this prototype exhibit the potential for wider application, particularly in facial recognition technologies. The discussion on this subject extends beyond the confines of this document, indicating ongoing developments and enhancements.

High-pressure pipelines are critical for transporting hazardous materials over long distances, but they face threats from third-party interference activities. Preventive measures are implemented, but interference accidents can still occur, making the need for high-quality detection strategies vital. This paper proposes an end-to-end Artificial Intelligence of Things (AIoT) solution to detect potential interference threats in real time. The solution involves developing a smart visual sensor capable of processing images using state-of-the-art computer vision algorithms and transmitting alerts to pipeline operators in real time. The system's core is based on the object-detection model (e.g., You Only Look Once version 4 (YOLOv4) and DETR with Improved deNoising anchOr boxes (DINO)), trained on a custom Pipeline Visual Threat Assessment (Pipe-VisTA) dataset. Among the trained models, DINO was able to achieve the best Mean Average Precision (mAP) of 71.2% for the unseen test dataset. However, for the deployment on a limited computational-ability edge computer (i.e., the NVIDIA Jetson Nano), the simpler and TensorRT-optimized YOLOv4 model was used, which achieved a mAP of 61.8% for the test dataset. The developed AIoT device captures the image using a camera, processes on the edge using the trained YOLOv4 model to detect the potential threat, transmits the threat alert to a Fleet Portal via LoRaWAN, and hosts the alert on a dashboard via a satellite network. The device has been fully tested in the field to ensure its functionality prior to deployment for the SEA Gas use-case. The AIoT smart solution has been deployed across the 10km stretch of the SEA Gas pipeline across the Murray Bridge section. In total, 48 AIoT devices and three Fleet Portals are installed to ensure the line-of-sight communication between the devices and portals. [ABSTRACT FROM AUTHOR]

With the rapid development of Internet of Things (IoT) technology, the number of IoT users is growing year after year. IoT will become a part of our daily lives, so it is likely that the security of these devices will be an important issue in the future. Quantum computing is maturing, and the security threat associated with quantum computing will be faced in the transmissions of IoT devices, which mainly use wireless communication technologies. Therefore, to ensure the protection of transmitted data, a cryptographic algorithm that is efficient in defeating quantum computer attacks needs to be developed. In this paper, we propose a device authentication and secure communication system with post-quantum cryptography (PQC) for AIoT environments using the NTRU and Falcon signature mechanism, which can resist quantum computer attacks and be used in AIoT environments to effectively protect the confidentiality, integrity, and non-repudiation of transmitted data. We also used Raspberry Pi to simulate AIoT devices for implementation. [ABSTRACT FROM AUTHOR]

This research is driven by the imperative to improve the milk pasteurization process's efficiency and safety within the Industry 5.0 framework. It introduces the artificial intelligence of things (AIoT) into the pasteurization system and focuses on deploying the C4.5 algorithm, anticipated to render precise real-time decisions by analyzing critical temperature and duration data essential for the low-temperature long-time (LTLT) pasteurization technique. Test results indicate that the C4.5 model is highly effective, demonstrating nearly perfect metrics with a precision of .78, a recall of .77, and an F1-score of .73, while achieving an overall algorithm accuracy of .76. In terms of heating control, the double jacket method tested on a 300-L system proved the control design's capability to quickly reach and maintain desired temperatures, hitting the set-point within 70 min. However, it was observed that systems with larger capacities took longer to reach the set-point, suggesting a potential for further optimization. The findings affirm that integrating AIoT into control systems can significantly enhance process efficiency and effectiveness, consistent with the goal of incorporating sophisticated technological solutions in the food sector. Practical Applications The article “Enhancing Low-Temperature Long-Time Milk Pasteurization Process with a C4.5 Algorithm-Driven AIoT System for Real-Time Decision-Making” explores an AIoT system's potential to revolutionize the dairy industry's pasteurization methods. This system promises enhanced efficiency by optimizing energy and time, thereby reducing costs. It ensures superior quality control through continuous monitoring, maintaining nutritional integrity and safety. The high automation level minimizes human error and operational demands, while predictive maintenance forecasts preresource increase. Moreover, it guarantees product traceability and compliance with food safety regulations. The generated data offers insights for further supply chain optimizations. This integration of AI and IoT thus presents significant advancements in process management, quality assurance, and scalable operations for the dairy industry. [ABSTRACT FROM AUTHOR]

The objective of this paper is to research the synergistic interaction between Industrial Internet of Things (IIoT) and the application of Artificial Intelligence (AI) in IIoT ecosystems (AIoT). For achieving this purpose, a method of integrative study of leading scientific publications on the topics 'Industrial Internet of Things' and 'Artificial Internet of Things' and the PRIZMA flow chart for filtering the list of found documents are applied. The inability of full coverage of the AIoT topic in one article has necessitated the usage of an approach for its content analysis in several thematic groups. The IIoT and AIoT concepts has been explored by various theoretical and practical aspects in many scientific papers published mainly over the past five years, but continues to arouse the interest of researchers. The novelty of the manuscript could be viewed in the attempts of systematizing the key characteristics, advantages, and risks in the practical use of these technologies in the form of smart manufacturing machines, robotic devices, autonomous vehicles, and software for predictive maintenance, and in outlining the possibility for green and sustainable AIoT platforms creation. It is concluded that a key requirement for the implementation of AIoT solutions is the consideration of the various threats and ethical aspects of their design, which contributes to the correct definition of the 'social value' of such business initiatives. [ABSTRACT FROM AUTHOR]

Abstract Integrated air–ground network enhances AIoT performance by improving spectral efficiency, achieving high-speed, stable network connectivity, and enabling sensing, learning, and decision-making. However, unmanned aerial vehicles (UAVs) can lead to local spectrum congestion and competition. To address this issue, intelligent signal processing techniques are employed to enhance AIoT system performance and stability through intelligent multi-channel sensing and communication. A novel communication framework inspired by brain cognition for UAV communication in heterogeneous environments is introduced. This framework iteratively determines the importance of signals, effectively eliminating unimportant signals with interference characteristics, and reducing their transmission power. Simulation results demonstrate the superiority of this method in terms of communication performance.

Globally, 50.44% of buildings are over 30 years old. Old buildings have problems such as obsolete safety monitoring systems and poor management efficiency caused by the use of old equipment. For the better management of buildings and their facilities, we developed an efficient management system based on digitized data from sensors in each device that is connected to the system. The system monitored the operation of each device in real time and sent messages to maintenance personnel in case of any abnormality in each device. This system ensured the normal operation of the building and saved costs and energy via efficient functions. The system was developed and implemented for one year. The operation outcome showed the completeness, relevance, standardization, and visualization of collected data and information. It also enabled the effective scheduling of operation and maintenance by querying the data from each device for intelligent building management. [ABSTRACT FROM AUTHOR]

The rapid increase in new challenges of the combination of the Internet of Things (IoT) and artificial intelligence (AI), which are emerging technologies, can play a compelling role in prompting the development of artificial intelligence Internet of Things (AIoT). Therefore, the demand for AI competencies for everyone will increase. Educational institutes focus on encouraging AI education because the demand for AI-literate workers will increase in the industrial sector. However, teachers' lack of AI knowledge is a significant barrier to AI education. Thus, developing the teacher's AI competencies and educating them about how to use and teach students is critical. In this study, we proposed artificial intelligence of things professional development (AIoT-PD) training to prepare the AI competencies of teachers ready to teach. A quasi-experimental design with a two-day training workshop was conducted among 13 teachers to examine its impact on AI competencies, including AI knowledge, AI skill, and AI attitude. The quantitative data were collected via a pretest and posttest after the training activity, while qualitative data were collected via interviews. This study showed that teachers' AI knowledge significantly improved. These findings revealed the AIoT training workshop's effectiveness in enhancing teachers' AI competencies, which can help them effectively teach students in AI education. [ABSTRACT FROM AUTHOR]

Integrated air–ground network enhances AIoT performance by improving spectral efficiency, achieving high-speed, stable network connectivity, and enabling sensing, learning, and decision-making. However, unmanned aerial vehicles (UAVs) can lead to local spectrum congestion and competition. To address this issue, intelligent signal processing techniques are employed to enhance AIoT system performance and stability through intelligent multi-channel sensing and communication. A novel communication framework inspired by brain cognition for UAV communication in heterogeneous environments is introduced. This framework iteratively determines the importance of signals, effectively eliminating unimportant signals with interference characteristics, and reducing their transmission power. Simulation results demonstrate the superiority of this method in terms of communication performance. [ABSTRACT FROM AUTHOR]

The telecom sector is currently undergoing a digital transformation by integrating artificial intelligence (AI) and Internet of Things (IoT) technologies. Customer retention in this context relies on the application of autonomous AI methods for analyzing IoT device data patterns in relation to the offered service packages. One significant challenge in existing studies is treating churn recognition and customer segmentation as separate tasks, which diminishes overall system accuracy. This study introduces an innovative approach by leveraging a unified customer analytics platform that treats churn recognition and segmentation as a bi-level optimization problem. The proposed framework includes an Auto Machine Learning (AutoML) oversampling method, effectively handling three mixed datasets of customer churn features while addressing imbalanced-class distribution issues. To enhance performance, the study utilizes the strength of oversampling methods like synthetic minority oversampling technique for nominal and continuous features (SMOTENC) and synthetic minority oversampling with encoded nominal and continuous features (SMOTE-ENC). Performance evaluation, using 10-fold cross-validation, measures accuracy and F1-score. Simulation results demonstrate that the proposed strategy, particularly Random Forest (RF) with SMOTE-NC, outperforms standard methods with SMOTE. It achieves accuracy rates of 79.24%, 94.54%, and 69.57%, and F1-scores of 65.25%, 81.87%, and 45.62% for the IBM, Kaggle Telco and Cell2Cell datasets, respectively. The proposed method autonomously determines the number and density of clusters. Factor analysis employing Bayesian logistic regression identifies influential factors for accurate customer segmentation. Furthermore, the study segments consumers behaviorally and generates targeted recommendations for personalized service packages, benefiting decision-makers. [ABSTRACT FROM AUTHOR]

Agriculture is an inevitable area for research that must be prioritized as the world population is continuously growing. To meet agricultural product demand, agricultural processes from production to consumption must be optimized and require automation. Digitalization of the agriculture process needs data security in each phase. This paper discusses the layer-wise common attacks in IoT-based agriculture systems and their traditional and Blockchain-based security solutions. Furthermore, this paper reports the current trending technology used by researchers from academia and industry in agriculture and future directions. [ABSTRACT FROM AUTHOR]

Abstract Sedentary, inadequate sleep and exercise can affect human health. Artificial intelligence (AI) and Internet of Things (IoT) create the Artificial Intelligence of Things (AIoT), providing the possibility to solve these problems. This paper presents a novel approach to monitor various human behaviors for AIoT‐based health management using triboelectric nanogenerator (TENG) sensors. The insole with solely one TENG sensor, creating a most simplified system that utilizes machine learning (ML) for personalized motion monitoring, encompassing identity recognition and gait classification. A cushion with 12 TENG sensors achieves real‐time identity and sitting posture recognition with accuracy rates of 98.86% and 98.40%, respectively, effectively correcting sedentary behavior. Similarly, a smart pillow, equipped with 15 sensory channels, detects head movements during sleep, identifying 8 sleep patterns with 96.25% accuracy. Ultimately, constructing an AIoT‐based health management system to analyze these data, displaying health status through human‐machine interfaces, offers the potential to help individuals maintain good health.

High traffic flow in a confined tunnel makes fire safety a critical issue. This paper proposed a digital twin framework for tunnel fire safety management in real-time, driven by dynamic sensor data and AIoT technologies. A deep learning model trained by the Transformer network and simulation dataset is used to predict real-time fire location and size. Then, the AI model is integrated into a 3D digital twin platform developed by the game engine Unity 3D. The performance of the proposed digital twin framework is demonstrated using numerical experiments and large-scale tunnel fire tests. Results show that the established AI model achieved promising accuracy in predicting fire location and power for both numerical and experimental data. The digital twin platform can also visualize the 3D fire scene that supports evacuation, firefighting, and emergency rescue. This research demonstrates the feasibility of using a 3D environment and digital twin in real-time fire safety management.

Abstract In complex industrial environments such as the Internet of Things in coal mines, large mechanical and electrical equipment can generate powerful impulsive noise, which can cause sudden errors. Because it is difficult to establish an accurate channel model, the performance of current error control techniques is limited. To enhance the reliability of information recovery in the Internet of Things in coal mines, the traditional method of shortening the communication distance between sensors is often utilized, but this can be costly. Therefore, this article proposes an intelligent signal processing method against impulsive noise interference that draws on the concept of the Artificial Intelligence of Things (AIoT) and incorporates deep learning technology. This method replaces the traditional sensor signal processing module with a Convolutional Neural Network (CNN), which learns the intricate mapping relationship between transmitted information and sensor signals in impulsive noise environments. Simulation results demonstrate that the proposed method outperforms the traditional sensor signal processing method in three impulsive noise environments by achieving a lower Bit Error Rate (BER). Moreover, this method adopts an improved lightweight neural network, which is more conducive to the deployment of mobile terminals in the Internet of Things.

In smart enterprise systems, augmented IoT can efficiently improve the decision-making, handling, and generation of a huge amount of information during communication. However, Augmented Internet-of-Things (AIoT) leads to various security and trust issues when transmitting information through intermediate devices. In a case where malicious devices can easily integrate with legitimate devices, it can further affect and interfere with the overall performance of the network system. Though various security surveys have been illustrated and schemes have been proposed by scientists, however, all of them are in their early stages. This paper proposes a trusted decision-making mechanism called Active Recognition Trust (ART), using AIoT for handling smart enterprise systems. The proposed mechanism integrates active recognition and associated reference mechanisms to improve the efficiency and effectiveness of the secure transmission process by computing a trust value for each device using impact factors of function fusion systems before information exchanges. Simulation results show that the proposed mechanism can efficiently enhance performance while improving the accuracy of recognizing legitimate devices by reducing or eliminating interference from malicious devices. The proposed mechanism is evaluated using the transmission ratio, identification accuracy, average trust, and run cycle compared to the existing mechanisms. Further, the proposed mechanism achieves approximately 89% better improvement than the baseline approach.

Industry 4.0 introduced new concepts, technologies, and paradigms, such as Cyber Physical Systems (CPSs), Industrial Internet of Things (IIoT) and, more recently, Artificial Intelligence of Things (AIoT). These paradigms ease the creation of complex systems by integrating heterogeneous devices. As a result, the structure of the production systems is changing completely. In this scenario, the adoption of reference architectures based on standards may guide designers and developers to create complex AIoT applications. This article surveys the main reference architectures available for industrial AIoT applications, analyzing their key characteristics, objectives, and benefits; it also presents some use cases that may help designers create new applications. The main goal of this review is to help engineers identify the alternative that best suits every application. The authors conclude that existing reference architectures are a necessary tool for standardizing AIoT applications, since they may guide developers in the process of developing new applications. However, the use of reference architectures in real AIoT industrial applications is still incipient, so more development effort is needed in order for it to be widely adopted.

Agriculture Internet of Things (AIoTs) deployments require design of high-efficiency Quality of Service (QoS) & security models that can provide stable network performance even under large-scale communication requests. Existing security models that use blockchains are either highly complex or require large delays & have higher energy consumption for larger networks. Moreover, the efficiency of these models depends directly on consensus-efficiency & miner-efficiency, which restricts their scalability under real-time scenarios. To overcome these limitations, this study proposes the design of an efficient Q-Learning bioinspired model for enhancing QoS of AIoT deployments via customized shards. The model initially collects temporal information about the deployed AIoT Nodes, and continuously updates individual recurring trust metrics. These trust metrics are used by a Q-Learning process for identification of miners that can participate in the block-addition process. The blocks are added via a novel Proof-of-Performance (PoP) based consensus model, which uses a dynamic consensus function that is based on temporal performance of miner nodes. The PoP consensus is facilitated via customized shards, wherein each shard is deployed based on its context of deployment, that decides the shard-length, hashing model used for the shard, and encryption technique used by these shards. This is facilitated by a Mayfly Optimization (MO) Model that uses PoP scores for selecting shard configurations. These shards are further segregated into smaller shards via a Bacterial Foraging Optimization (BFO) Model, which assists in identification of optimal shard length for underlying deployment contexts. Due to these optimizations, the model is able to improve the speed of mining by 4.5%, while reducing energy needed for mining by 10.4%, improving the throughput during AIoT communications by 8.3%, and improving the packet delivery consistency by 2.5% when compared with existing blockchain-based AIoT deployment models under similar scenarios. This performance was observed to be consistent even under large-scale attacks.

The telecom sector is currently undergoing a digital transformation by integrating artificial intelligence (AI) and Internet of Things (IoT) technologies. Customer retention in this context relies on the application of autonomous AI methods for analyzing IoT device data patterns in relation to the offered service packages. One significant challenge in existing studies is treating churn recognition and customer segmentation as separate tasks, which diminishes overall system accuracy. This study introduces an innovative approach by leveraging a unified customer analytics platform that treats churn recognition and segmentation as a bi-level optimization problem. The proposed framework includes an Auto Machine Learning (AutoML) oversampling method, effectively handling three mixed datasets of customer churn features while addressing imbalanced-class distribution issues. To enhance performance, the study utilizes the strength of oversampling methods like synthetic minority oversampling technique for nominal and continuous features (SMOTE-NC) and synthetic minority oversampling with encoded nominal and continuous features (SMOTE-ENC). Performance evaluation, using 10-fold cross-validation, measures accuracy and F1-score. Simulation results demonstrate that the proposed strategy, particularly Random Forest (RF) with SMOTE-NC, outperforms standard methods with SMOTE. It achieves accuracy rates of 79.24%, 94.54%, and 69.57%, and F1-scores of 65.25%, 81.87%, and 45.62% for the IBM, Kaggle Telco and Cell2Cell datasets, respectively. The proposed method autonomously determines the number and density of clusters. Factor analysis employing Bayesian logistic regression identifies influential factors for accurate customer segmentation. Furthermore, the study segments consumers behaviorally and generates targeted recommendations for personalized service packages, benefiting decision-makers.

AbstractCurrent methods used in cell culture monitoring, characterisation and handling are manual, time consuming and highly dependent on subjective observations made by human operators, resulting in inconsistent outcomes. This project focuses on developing an automated system for cell growth analysis, utilising Artificial Intelligence of Things (AIoT) for use in bio-manufacturing and bio-research. The proposed AIoT system applies a U-Net convolutional neural network (CNN) model for fibroblast cell segmentation to monitor confluency and incorporates a mechanical robotic arm for automated sample handling. Intel Movidius Neural Compute Stick 2 (NCS2) and OpenVINO Toolkit were used to allow for standalone deployment on an UP2 Squared and a Raspberry Pi board that is integrated with a digital microscope system. The robotic arm was programmed to pick, place and sort the cell samples within the working environment. The results obtained from the CNN model development achieved an accuracy of 95% and an intersection over Union (IoU) of 66%. The OpenVINO Toolkit successfully optimised power-consumption and accelerated the segmentation on a 2K image to be completed in less than 13 seconds. The AIoT cell detection and characterisation system is able to automatically analyse the cell culture while reducing manual sample handling by laboratory personnel. Eventually, it is hoped that this AIoT automated cell detection and characterisation system will have a positive impact and contribute towards the implementation of the Industrial Revolution IR4.0 in bio-based research and industries.

ChoiWith the rapid development of the Internet of Things and artificial intelligence (AI), the requirement for sensing technologies for smart bearings has increased dramatically. The general bearing sensors can only recognize the basic information from temperature or vibration, far from satisfying the self-diagnosis and self-maintenance. Recently, self-powered sensing technologies based on triboelectric nanogenerators have paved a new route for fabricating smart bearings. In this study, the triboelectric principle is applied to a commercial metal-polymer plain bearing (MPPB) bearing, which can achieve self-sensing, self-diagnosis, and self-maintenance. The geometrical structure of the triboelectric MPPB (T-MPPB) is designed to balance the output efficiency and external load, and the super durability and load capability are verified. Besides, the mechanism behind the output change trend under boundary and hydrostatic fluid lubrication is revealed for the first time. Furthermore, the deep learning algorithm can classify the lubrication states with highly accurate performance. The proposed T-MPPB has the potential to achieve self-maintenance with the lubricating pump according to the lubrication condition classified by the AI. This research not only establishes the feasibility of designing self-powered smart MPPB but also demonstrates a way for identifying lubrication states, thus achieving self-diagnosis and self-maintenance ability by self-powered sensors. [ABSTRACT FROM AUTHOR]

In 2020, the world experienced the threat of the COVID-19 epidemic, and seniors and chronic disease patients generally reduced their exercise and social activities to avoid increasing the risk of infection, which could lead to increased loneliness and even many diseases. Combining golf croquet games with a mobile application (App) and AIoT companion robots, this research constructs a home-based intelligent exercise system, uses the technology acceptance model (TAM), deduces users' intention to use this system based on perceived usefulness and perceived ease of use, and adds the needs of love and belonging, esteem, cognitive, aesthetic, and self-actualization in Maslow's hierarchy of needs theory (HNT) to conduct an analysis of system needs. This study collected empirical data, totaling 320 participants including seniors and their caregivers, from elderly care centers in northern Taiwan using a cross-sectional survey and purposive sampling. Based on regression and variance analysis, the results show that participants have a high level of acceptance of this system, believing that it is easy to learn and operate and can increase interaction with others, improve self-confirmation, satisfy the thirst for knowledge, increase the feeling of happiness, and fulfill self-actualization needs. In the future, by collecting and recording the process of seniors using the App, so as to find out their health problems as soon as possible, expand their daily life through this exercise, and achieve the goal of happy living and better healthcare. [ABSTRACT FROM AUTHOR]

In complex industrial environments such as the Internet of Things in coal mines, large mechanical and electrical equipment can generate powerful impulsive noise, which can cause sudden errors. Because it is difficult to establish an accurate channel model, the performance of current error control techniques is limited. To enhance the reliability of information recovery in the Internet of Things in coal mines, the traditional method of shortening the communication distance between sensors is often utilized, but this can be costly. Therefore, this article proposes an intelligent signal processing method against impulsive noise interference that draws on the concept of the Artificial Intelligence of Things (AIoT) and incorporates deep learning technology. This method replaces the traditional sensor signal processing module with a Convolutional Neural Network (CNN), which learns the intricate mapping relationship between transmitted information and sensor signals in impulsive noise environments. Simulation results demonstrate that the proposed method outperforms the traditional sensor signal processing method in three impulsive noise environments by achieving a lower Bit Error Rate (BER). Moreover, this method adopts an improved lightweight neural network, which is more conducive to the deployment of mobile terminals in the Internet of Things. [ABSTRACT FROM AUTHOR]

The autonomous fault diagnosis of mechanical systems is crucial to addressing smart manufacturing product issues. In this article, we propose intelligent diagnosis and prediction technologies based on acoustic emission (AE) for mechanical motors. The integration of practical technologies, such as acoustic analysis, artificial intelligence (AI), edge computing (EC), electromagnetics, communication, and other theory-based subjects, is convenient for achieving flexible changes made in response to the edge operation trend. The proposed model, developed using acoustic information links with machine learning (ML) platforms to collect acoustic information via feature extraction (FE), is novel in that it can detect system health and prevent system failures. It can inspire innovative design concepts once the above model is combined with the EC migration module. In addition, in this paper, we discuss the embedded system in smart manufacturing applications, including AE, to establish an ML framework that is trained using audio emission data. The valuable results from the proposed algorithm experiments show that the audio judgment accuracy rate can be above 90%. At the current stage, the metric accuracy and precision of mechanical motor discrimination can reach 93.5% and 0.97, respectively. In this paper, we present an analytical method for performing motor axis misalignment judgment based on tiny machine learning (TinyML) techniques, which will enable the IoT field to move toward smart energy savings. [ABSTRACT FROM AUTHOR]

In smart enterprise systems, augmented IoT can efficiently improve the decision-making, handling, and generation of a huge amount of information during communication. However, Augmented Internet-of-Things (AIoT) leads to various security and trust issues when transmitting information through intermediate devices. In a case where malicious devices can easily integrate with legitimate devices, it can further affect and interfere with the overall performance of the network system. Though various security surveys have been illustrated and schemes have been proposed by scientists, however, all of them are in their early stages. This paper proposes a trusted decision-making mechanism called Active Recognition Trust (ART), using AIoT for handling smart enterprise systems. The proposed mechanism integrates active recognition and associated reference mechanisms to improve the efficiency and effectiveness of the secure transmission process by computing a trust value for each device using impact factors of function fusion systems before information exchanges. Simulation results show that the proposed mechanism can efficiently enhance performance while improving the accuracy of recognizing legitimate devices by reducing or eliminating interference from malicious devices. The proposed mechanism is evaluated using the transmission ratio, identification accuracy, average trust, and run cycle compared to the existing mechanisms. Further, the proposed mechanism achieves approximately 89% better improvement than the baseline approach. [ABSTRACT FROM AUTHOR]

been rapidly developed due to their flexibility and scalability to be adopted in several fields for modeling realworld applications like object detection, image classification, etc. However, their high accuracy incurs intensive computations. Therefore, it is crucial to carefully choose a suitable computer platform and implementation methodology for CNN network architectures while achieving increased efficiency. Parallel architectures are prevalent in CNN implementation. Herein, we present a new Single Instruction Multi Data (SIMD) parallel implementation of the proposed CNN to speed up the execution process and make it suitable to deploy on low-cost, low-power consumption platforms. The proposed implementation produces an improved model of deep CNN executable on a cost-efficient platform and portability to work autonomously with multi-core processing units while maintaining working accuracy. Raspberry Pi 3 B is a low-power target device for implementing our model. The proposed approach is characterized by high diagnostic accuracy of up to 96.35 % while incurring power consumption of 3.65 Watts, achieving power reduction between 19.17 % and 68.45 % compared to the prior work. Meanwhile, it has a fine inference time for the selected platform. The outstanding results of this study reflect the success of employing parallel architectures to utilize the quad courses of the ARM processor on the target platform. The presented model can be an efficient medical assistant to provide automated detection and diagnosis for myopia ocular disease. Thus, it can be a promising healthcare toolkit that reduces the effort of the medical staff and increases the quality of the provided medical services for myopia patients [ABSTRACT FROM AUTHOR]

Edge computing technology solves the shortcomings of high latency, mobility and location awareness in remote cloud computing, but it also brings many security challenges to the AIoT. In view of the open and heterogeneous characteristics of edge network, a more secure edge computing intrusion detection method is studied in this paper. The main idea is to combine edge computing with AIoT to realize edge smart interconnection. This method optimizes the weight value in machine learning by increasing the screening and mitigation of cloud server training samples, so as to provide efficient and accurate edge intrusion detection behavior. The experimental results show that the proposed method can more effectively improve the detection accuracy and reduce the detection time. [ABSTRACT FROM AUTHOR]

To encourage the good health and well-being sustainable development goal, this article presents the design and implementation of real-time indoor air quality index (AQI) prediction using an artificial internet of things (AIoT) electronic nose integrated into a vacuum cleaner robot. The objective of the proposed method is to implement an effective embedded AIoT solution utilizing sensor fusion and the TinyML framework for the purpose of strengthening the environmental health system with suitable current technology. The high-accuracy sensor outputs of total volatile organic compounds (TVOC), humidity, equivalent carbon dioxide (eCO2), and PM2.5 gathered as the dataset are normalized in the data pre-processing state and utilized to create trained models using dense neural networks (DNN) deep learning algorithms. Tiny machine learning is responsible for neural network training, as it is capable of executing AI algorithms on embedded devices with extremely low power consumption and limited RAM and ROM resources. The testing results demonstrate that the predictive model performed well, with 99% accuracy for a maximum absolute regression error less than 15 and an 18.33 mean square error. The embedded device implementation uses Wio terminals with ARM Cortex-M4F microcontrollers for real-time indoor air quality index prediction and visualization. Experimental results demonstrating the average precision of the indoor AQI prediction were obtained at an average accuracy of over 98% with a computation time of 10 milliseconds and an acceptable usage of ROM and RAM resources of 8.5 KB and 1.1 KB, respectively, along with successive performances that satisfied web application data virtualization using the internet of things (IoT). [ABSTRACT FROM AUTHOR]

Abstract There have recently been intensive efforts aimed at addressing the challenges of environmental degradation and climate change through the applied innovative solutions of AI, IoT, and Big Data. Given the synergistic potential of these advanced technologies, their convergence is being embraced and leveraged by smart cities in an attempt to make progress toward reaching the environmental targets of sustainable development goals under what has been termed “environmentally sustainable smart cities.” This new paradigm of urbanism represents a significant research gap in and of itself. To fill this gap, this study explores the key research trends and driving factors of environmentally sustainable smart cities and maps their thematic evolution. Further, it examines the fragmentation, amalgamation, and transition of their underlying models of urbanism as well as their converging AI, IoT, and Big Data technologies and solutions. It employs and combines bibliometric analysis and evidence synthesis methods. A total of 2,574 documents were collected from the Web of Science database and compartmentalized into three sub-periods: 1991–2015, 2016–2019, and 2020–2021. The results show that environmentally sustainable smart cities are a rapidly growing trend that markedly escalated during the second and third periods—due to the acceleration of the digitalization and decarbonization agendas—thanks to COVID-19 and the rapid advancement of data-driven technologies. The analysis also reveals that, while the overall priority research topics have been dynamic over time—some AI models and techniques and environmental sustainability areas have received more attention than others. The evidence synthesized indicates that the increasing criticism of the fragmentation of smart cities and sustainable cities, the widespread diffusion of the SDGs agenda, and the dominance of advanced ICT have significantly impacted the materialization of environmentally sustainable smart cities, thereby influencing the landscape and dynamics of smart cities. It also suggests that the convergence of AI, IoT, and Big Data technologies provides new approaches to tackling the challenges of environmental sustainability. However, these technologies involve environmental costs and pose ethical risks and regulatory conundrums. The findings can inform scholars and practitioners of the emerging data-driven technology solutions of smart cities, as well as assist policymakers in designing and implementing responsive environmental policies.

High-pressure pipelines are critical for transporting hazardous materials over long distances, but they face threats from third-party interference activities. Preventive measures are implemented, but interference accidents can still occur, making the need for high-quality detection strategies vital. This paper proposes an end-to-end Artificial Intelligence of Things (AIoT) solution to detect potential interference threats in real time. The solution involves developing a smart visual sensor capable of processing images using state-of-the-art computer vision algorithms and transmitting alerts to pipeline operators in real time. The system’s core is based on the object-detection model (e.g., You Only Look Once version 4 (YOLOv4) and DETR with Improved deNoising anchOr boxes (DINO)), trained on a custom Pipeline Visual Threat Assessment (Pipe-VisTA) dataset. Among the trained models, DINO was able to achieve the best Mean Average Precision (mAP) of 71.2% for the unseen test dataset. However, for the deployment on a limited computational-ability edge computer (i.e., the NVIDIA Jetson Nano), the simpler and TensorRT-optimized YOLOv4 model was used, which achieved a mAP of 61.8% for the test dataset. The developed AIoT device captures the image using a camera, processes on the edge using the trained YOLOv4 model to detect the potential threat, transmits the threat alert to a Fleet Portal via LoRaWAN, and hosts the alert on a dashboard via a satellite network. The device has been fully tested in the field to ensure its functionality prior to deployment for the SEA Gas use-case. The AIoT smart solution has been deployed across the 10km stretch of the SEA Gas pipeline across the Murray Bridge section. In total, 48 AIoT devices and three Fleet Portals are installed to ensure the line-of-sight communication between the devices and portals.

In recent years, environmental sound classification (ESC) has prevailed in many artificial intelligence Internet of Things (AIoT) applications, as environmental sound contains a wealth of information that can be used to detect particular events. However, existing ESC methods have high computational complexity and are not suitable for deployment on AIoT devices with constrained computing resources. Therefore, it is of great importance to propose a model with both high classification accuracy and low computational complexity. In this work, a new ESC method named BSN-ESC is proposed, including a big–small network-based ESC model that can assess the classification difficulty level and adaptively activate a big or small network for classification as well as a pre-classification processing technique with logmel spectrogram refining, which prevents distortion in the frequency-domain characteristics of the sound clip at the joint part of two adjacent sound clips. With the proposed methods, the computational complexity is significantly reduced, while the classification accuracy is still high. The proposed BSN-ESC model is implemented on both CPU and FPGA to evaluate its performance on both PC and embedded systems with the dataset ESC-50, which is the most commonly used dataset. The proposed BSN-ESC model achieves the lowest computational complexity with the number of floating-point operations (FLOPs) of only 0.123G, which represents a reduction of up to 2309 times in computational complexity compared with state-of-the-art methods while delivering a high classification accuracy of 89.25%. This work can achieve the realization of ESC being applied to AIoT devices with constrained computational resources. [ABSTRACT FROM AUTHOR]

In the bustling streets of Pakistan, small-scale fruits and vegetables shops stand as vital hubs of daily life. These humble establishments are where people flock to satisfy their everyday needs. However, the traditional methods employed by shopkeepers using manual weighing scales have proven to be time-consuming and limit the shopkeepers' ability to serve multiple customers simultaneously. But in a world rapidly embracing automation and propelled by the wonders of artificial intelligence, a revolution is underway. In this visionary paper, we introduce the concept of an intelligent scale that will revolutionize the retail process. This remarkable scale possesses the power to automate numerous tasks, making the shopping experience seamless and efficient. Imagine a scale that not only determines the weight of the produce but also possesses the ability to detect and identify each fruit and vegetable placed upon it. By harnessing the potential of cutting-edge technology, we fine-tuned pre-trained models, such as YOLOv5n and YOLOv7, on our extensive dataset, consisting of 12 classes and 2622 images. The dataset was collected manually and it closely aligns with real-time scenarios, ensuring that the distribution in our training and validation sets were similar and that it reflected what our models will encounter during testing. As a result, our YOLOv5n and YOLOv7 models have achieved astonishing mean Average Precision (mAP) scores of 0.98 and 0.987, respectively. YOLOv5n demonstrates an impressive processing speed of 20 frames per second (fps) on a CPU, while it reaches an impressive 125 fps on a GPU. Similarly, YOLOv7 achieves a processing speed of 2 fps on a CPU, which escalates to 66.6 fps on a GPU. These extraordinary results testify to the remarkable accuracy and efficacy of our system when subjected to real-world testing scenarios. To ensure accurate weighing, we incorporated a load cell with an hx711 amplifier, providing precise measurements that customers can trust. However, our intelligent scale does not stop there. We understand that determining weight alone is insufficient when it comes to transactions. Hence, a meticulously crafted Python script was developed to map each specific item to its corresponding price based on its weight. With all these incredible features in place, the experience of purchasing from a fruits and vegetables shop is taken to new heights. The intelligent scale is accompanied by a user-friendly graphical user interface (GUI), where customers can conveniently view their order and prices. Once the order is complete, a simple click on the print button generates a neatly printed bill, ensuring a seamless transaction. The implications of this intelligent scale are profound. Shopkeepers can now serve customers faster and more efficiently, effortlessly managing multiple transactions simultaneously. The introduction of automation enhances the overall shopping experience, leaving customers delighted and eager to return. This amalgamation of technology and traditional commerce heralds a new era, where small-scale shops can thrive and adapt to the ever-evolving needs of the modern world. [ABSTRACT FROM AUTHOR]

Energy demand forecasting is crucial for planning and optimizing the use of energy resources in building facilities. However, integrating digital solutions and learning techniques into legacy buildings presents significant challenges due to limited or outdated resources, hampering predictive analytics in these buildings and their circuits. To fill this gap, this article proposes an innovative demand forecasting strategy using an AIoT retrofit architecture based on the SmartLVGrid metamodel. This architecture allows remote monitoring of legacy building circuits, facilitating the collection, processing and storage of data in the cloud. We use several learning algorithms, including linear regression, support vector regressor, random forest regressor, XGBoost regressor, and long short-term memory (LSTM) neural network, to predict energy demand 15 min ahead, identifying potential overruns of contracted demand in accordance with Brazilian regulations. After Bayesian optimization, the LSTM neural network outperformed other models for most of the selected datasets and detected 32 out of 38 demand overruns on the test set. XGBoost and random forest followed closely, detecting 30 demand overruns. Overall, our cost-effective solution optimizes energy usage and efficiently mitigates potential demand exceedances in building installations. This is achieved through a step-by-step approach to upgrading existing aging facilities, which promotes energy efficiency and sustainability. [ABSTRACT FROM AUTHOR]

This study proposes an innovative integration of the Car-to-Car Network-Hierarchical deep neural network (CtCNET-HDRNN) model with Fifth generation (5G) and Dedicated Short-Range Communications (DSRC) systems, streamlining computational efficiency in edge computing. CtCNET-HDRNN is a specialized deep learning model designed for vehicular communication, allowing vehicles to exchange information seamlessly in a connected environment. It harnesses an adaptive learning rate and regularization within the model’s advanced training methodology, ensuring optimal data fit, superior generalization, and efficient convergence. A key novelty lies in the introduction of a Sparse Deep Recurrent Neural Network (SDRNN), which significantly reduces computational complexity by pruning insignificant connections, making it suitable for deployment on resource-constrained edge devices. SDRNN is a variant of recurrent neural networks designed to minimize computational burden while maintaining high performance in time-series data analysis. Furthermore, this research presents an original integration model, adeptly merging the CtCNET-HDRNN model with the Millimeter wave (mmWave) of 5G and Monte Carlo for DSRC systems for seamless data transmission. The mmWave technology offers high-speed communication capabilities, while Monte Carlo enables adaptive collision avoidance and efficient channel access control for vehicular networks. Beyond immediate computational gains, this integrated model also contributes significantly to edge computing research and practical applications, promising enhanced system performance and improved user experience in vehicular communication scenarios. The proposed approach opens new possibilities for efficient and reliable communication in connected vehicles, laying the foundation for safer and smarter transportation systems.

The limited availability of peer-reviewed scientific evidence in the cannabis industry has led many companies to rely on techniques derived from non-peer-reviewed sources for their practices. This study begins by examining the literature on the cultivation of C. sativa, investigating optimal conditions and their effects on growth, and characterising the requirements for greenhouse monitoring and control. A systematic review of current technological approaches is then conducted. The review demonstrates that technology-based control of greenhouse environments has the potential to surpass manual or traditional rule-based management techniques by reducing costs and increasing yields. However, the adoption of these technologies is impeded by the lack of publicly available labelled data on growth, pest and disease, environmental, and yield data of multiple indoor cultivation cycles. Currently, much of the research in this field is conducted privately by companies in the cannabis industry. This study recognises substantial gaps in research surrounding C. sativa cultivation and emphasises the opportunity for new research to address the absence of available C. sativa datasets and peer-reviewed scientific studies outside of private endeavours.

There have recently been intensive efforts aimed at addressing the challenges of environmental degradation and climate change through the applied innovative solutions of AI, IoT, and Big Data. Given the synergistic potential of these advanced technologies, their convergence is being embraced and leveraged by smart cities in an attempt to make progress toward reaching the environmental targets of sustainable development goals under what has been termed "environmentally sustainable smart cities." This new paradigm of urbanism represents a significant research gap in and of itself. To fill this gap, this study explores the key research trends and driving factors of environmentally sustainable smart cities and maps their thematic evolution. Further, it examines the fragmentation, amalgamation, and transition of their underlying models of urbanism as well as their converging AI, IoT, and Big Data technologies and solutions. It employs and combines bibliometric analysis and evidence synthesis methods. A total of 2,574 documents were collected from the Web of Science database and compartmentalized into three sub-periods: 1991–2015, 2016–2019, and 2020–2021. The results show that environmentally sustainable smart cities are a rapidly growing trend that markedly escalated during the second and third periods—due to the acceleration of the digitalization and decarbonization agendas—thanks to COVID-19 and the rapid advancement of data-driven technologies. The analysis also reveals that, while the overall priority research topics have been dynamic over time—some AI models and techniques and environmental sustainability areas have received more attention than others. The evidence synthesized indicates that the increasing criticism of the fragmentation of smart cities and sustainable cities, the widespread diffusion of the SDGs agenda, and the dominance of advanced ICT have significantly impacted the materialization of environmentally sustainable smart cities, thereby influencing the landscape and dynamics of smart cities. It also suggests that the convergence of AI, IoT, and Big Data technologies provides new approaches to tackling the challenges of environmental sustainability. However, these technologies involve environmental costs and pose ethical risks and regulatory conundrums. The findings can inform scholars and practitioners of the emerging data-driven technology solutions of smart cities, as well as assist policymakers in designing and implementing responsive environmental policies. [ABSTRACT FROM AUTHOR]

Forest fires are the main cause of desertification, and they have a disastrous impact on agricultural and forest ecosystems. Modern fire detection and warning systems rely on several techniques: satellite monitoring, sensor networks, image processing, data fusion, etc. Recently, Artificial Intelligence (AI) algorithms have been applied to fire recognition systems, enhancing their efficiency and reliability. However, these devices usually need constant data transmission along with a proper amount of computing power, entailing high costs and energy consumption. This paper presents the prototype of a Video Surveillance Unit (VSU) for recognising and signalling the presence of forest fires by exploiting two embedded Machine Learning (ML) algorithms running on a low power device. The ML models take audio samples and images as their respective inputs, allowing for timely fire detection. The main result is that while the performances of the two models are comparable when they work independently, their joint usage according to the proposed methodology provides a higher accuracy, precision, recall and F 1 score ( 96.15 %, 92.30 %, 100.00 %, and 96.00 %, respectively). Eventually, each event is remotely signalled by making use of the Long Range Wide Area Network (LoRaWAN) protocol to ensure that the personnel in charge are able to operate promptly. [ABSTRACT FROM AUTHOR]

Current methods used in cell culture monitoring, characterisation and handling are manual, time consuming and highly dependent on subjective observations made by human operators, resulting in inconsistent outcomes. This project focuses on developing an automated system for cell growth analysis, utilising Artificial Intelligence of Things (AIoT) for use in bio-manufacturing and bio-research. The proposed AIoT system applies a U-Net convolutional neural network (CNN) model for fibroblast cell segmentation to monitor confluency and incorporates a mechanical robotic arm for automated sample handling. Intel Movidius Neural Compute Stick 2 (NCS2) and OpenVINO Toolkit were used to allow for standalone deployment on an UP2 Squared and a Raspberry Pi board that is integrated with a digital microscope system. The robotic arm was programmed to pick, place and sort the cell samples within the working environment. The results obtained from the CNN model development achieved an accuracy of 95% and an intersection over Union (IoU) of 66%. The OpenVINO Toolkit successfully optimised power-consumption and accelerated the segmentation on a 2K image to be completed in less than 13 seconds. The AIoT cell detection and characterisation system is able to automatically analyse the cell culture while reducing manual sample handling by laboratory personnel. Eventually, it is hoped that this AIoT automated cell detection and characterisation system will have a positive impact and contribute towards the implementation of the Industrial Revolution IR4.0 in bio-based research and industries. [ABSTRACT FROM AUTHOR]

Deploying artificial intelligence on edge nodes of Low-Power Wide Area Networks can significantly reduce network transmission volumes, event response latency, and overall network power consumption. However, the edge nodes in LPWAN bear limited computing power and storage space, and researchers have found it challenging to improve the recognition capability of the nodes using sensor data from the environment. In particular, the domain-shift problem in LPWAN is challenging to overcome. In this paper, a complete AIoT system framework referred to as LPAI is presented. It is the first generic framework for implementing AIoT technology based on LPWAN applicable to acoustic scene classification scenarios. LPAI overcomes the domain-shift problem, which enables resource-constrained edge nodes to continuously improve their performance using real data to become more adaptive to the environment. For efficient use of limited resources, the edge nodes independently select representative data and transmit it back to the cloud. Moreover, the model is iteratively retrained on the cloud using the few-shot uploaded data. Finally, the feasibility of LPAI is analyzed, and simulation experiments on the public ASC dataset provide validation that our proposed framework can improve the recognition accuracy by as little as 5% using 85 actual sensor data points. [ABSTRACT FROM AUTHOR]

In recent years many amazing deep learning models have been developed, but in the process of practical applications, people often find that these deep learning models have high requirements for hardware storage space and computing power. In Artificial Intelligent of Things (AIoT) scenario, the computing power of the edge or terminal side are relatively limited, therefore, most conventional deep learning models are difficult to be deployed into AIoT devices. It is significant to explore the different performance under different scales of deep learning models. In this paper, we mainly propose a method to analyze the impact of deep learning models with various sizes through various experiments. We employ slimmable network as a Neural Archtecture Search (NAS) tool to realize various model size freely, and evaluate them on the indicators of flops, robustness and accuracy. The experimental results show the variation of flops, robustness and accuracy with the various model sizes, which help understand the impact on performance of deep learning models with different scales in AIoT systems. • Interpreting robustness of deep learning models with different model complexity. • Employing slimmable network to compress neural network. • Understanding the relevance between robustness and accuracy of deep learning models. [ABSTRACT FROM AUTHOR]

Abstract According to data from the Earth's Volcano and Geological Disaster Reduction Center, a country like Indonesia has experienced many natural disasters, one of which is flooding. Floods are an annual natural disaster, especially on mountain slopes. Mountainous areas experience more dangerous than floods than the urban areas because they can cause other natural disasters, such as landslides and damage the hiking trails. The steep and winding roads minimize and limit the number of officers working in the mountains. Therefore, flood detection and monitoring equipment is needed. The proposed system based on AIoT technology provides real-time flood analysis so that the authorities can monitor residents around mountainous areas and provide early warning. This research focuses on the flood observation system as an early warning system to effectively monitor the flood-prone mountain slopes in real time while taking into account the cost, time efficiency, and safety measurement. The proposed system design includes the integration of sensors into the microcontroller, and the communication between the posts using LoRa and SIM900 sends data to the cloud server via the Internet. All sensor readings for each post are displayed on the app, and alerts are sent via SMS and the app.