A Study on IoT Device Authentication Using Artificial Intelligence

Internet of Things (IoT) devices are largely embedded devices which lack a sophisticated user interface, e.g., touch screen, keyboard, etc. As a consequence, traditional Pre-Shared Key (PSK) based authentication for mobile devices becomes difficult to apply. For example, according to our study on home automation devices which leverage smartphone for PSK input, the current process does not protect against active impersonating attack and also leaks the Wi-Fi password to eavesdroppers, i.e., currently these IoT devices can be exploited to enter into critical infrastructures, e.g., home networks. Motivated by this real-world security vulnerability, in this paper we propose a novel proximity-based mechanism for IoT device authentication, called Move2Auth, for the purpose of enhancing IoT device security. In Move2Auth, we require user to hold smartphone and perform one of two hand-gestures (moving towards and away, and rotating) in front of IoT device. By combining (1) large RSS-variation and (2) matching between RSS-trace and smartphone sensor-trace, Move2Auth can reliably detect proximity and authenticate IoT device accordingly. Based on our implementation on Samsung Galaxy smartphone and commodity Wi-Fi adapter, we prove Move2Auth can protect against powerful active attack, i.e., the false-positive rate is consistently lower than 0.5%.

This paper addresses the optimization problems in far-field wireless power transfer systems using Reinforcement Learning techniques. The mobile Radio-Frequency(RF) wireless transmitter can charge the nearby harvested energy enabled Internet of Things(IoT) devices. The wireless transmitter intends to find the optimal path to charge all stationary IoT devices in the shortest time. The Reinforcement Learning is applied to determine both the order of the IoT devices to be charged and the shortest path for the robot to drive from one IoT device to another. Due to the high complexity of the task and the charging environment, the traditional Reinforcement Learning even cannot converge. In order to solve this problem, a Finite State Machine Reinforcement Learning is proposed. At first, a heuristic algorithm is utilized to calculate the optimal path between any two IoT devices. Next, using the idea of the Finite State Machine in designing sequential circuit, the proposed algorithm determines the order of the IoT devices to be charged by setting each IoT device as a particular state. As a result, both the order of the IoT devices and the path from one device to another can be figured out. The numerical results prove the superiority of the Finite State Machine Reinforcement Learning in solving the proposed optimization problems. The real wireless power transfer system is also accomplished in our experimental field.

Many Internet of Things (IoT) services utilize an IoT access network to connect small devices with remote servers. They can share an access network with standard communication technology, such as IEEE 802.11ah. However, an authentication and key management (AKM) mechanism for resource constrained IoT devices using IEEE 802.11ah has not been proposed as yet. We therefore propose a new AKM mechanism for an IoT access network, which is based on IEEE 802.11 key management with the IEEE 802.1X authentication mechanism. The proposed AKM mechanism does not require any pre-configured security information between the access network domain and the IoT service domain. It considers the resource constraints of IoT devices, allowing IoT devices to delegate the burden of AKM processes to a powerful agent. The agent has sufficient power to support various authentication methods for the access point, and it performs cryptographic functions for the IoT devices. Performance analysis shows that the proposed mechanism greatly reduces computation costs, network costs, and memory usage of the resource-constrained IoT device as compared to the existing IEEE 802.11 Key Management with the IEEE 802.1X authentication mechanism.

The research focus was motivated by the limited understanding of cognitive technologies and the growing gap between artificial intelligence (AI) and human intelligence. The research is a literature review, and its purpose is to simplify the meaning and processes behind cognitive technologies, notably, the fundamentals of machine learning (ML) and computer vision with the intention to briefly address the alleged threat of AI taking over the job market. The research is a review of peer-reviewed articles retrieved from comparative studies, systematic reviews, meta-analysis, service research, reports, conference proceedings, experimental studies, literature reviews, scientometric analyses, books, and multi-case studies, dating from the years of 2018 to 2024. This literature review defines machine learning (ML), artificial intelligence (AI), computer vision, and convolutional neural networks (CNNs). It also compares machine learning to traditional programming and reveals the types of learning in ML models’ training. ML and its correlation with AI are also discussed and details about theory of mind, self-aware AI, reactive machines, and limited memory AI are shared. The literature expounds computer vision, particularly convolutional neural network (CNN) and CNN layers. Recent cutting-edge applications of artificial intelligence including generative AI models and autonomous systems are also incorporated. Finally, the literature briefly addresses the alleged threat of AI taking over the job market. The findings of this literature review reveal that AI is becoming the new way of operating. The conclusion shows that AI models require significant computation to allow computers to learn autonomously. Thus, understanding mathematical models of data and perfecting the process of writing software could be the key to remaining employable as more jobs are expected to be shifted due to AI and tasks automation. Keywords: Cognitive technology, artificial intelligence, machine learning, computer vision, convolutional neural networks

Extensive internet of things (IoT) devices being in-fected by malware are an increasingly important viable objective in IoT cyberattacks e.g., botnet, virus, trojan, etc. The botnets got leverage from unsecured IoT devices (e.g., CCTV, Raspberry Pi, Arduino Uno, ESP8266, etc) that operates by using the Internet traffic. In recent year, the high-profile IoT device’s vendor and the researcher from all over universities are exploring the robustness of IoT devices against botnet attacks. This research uses a deep learning approach to prevent botnet attacks on IoT networks. The deep residual one-dimensional CNN (1DCNN) model as the proposed method is used for botnet traffic detection. Two algorithms are provided: data processing for the N-BaIoT dataset and IoT botnet detection training and testing. For data processing, training, and testing, the datasets were evaluated, and the model was optimized with different optimizers. This research used RMS Prop, ADaDelta, AdaGrad, AdaMax, and Adam as optimizers and the CNN was compared with LSTM, CNN with RNN, and Deep residual 1DCNN, respectively. The results showed that Deep Residual 1DCNN with Adam has the highest training accuracy of 88.67%, 88.67% for validation accuracy, and 88.53%for test accuracy.

Background: Metabolic syndrome is a cluster of metabolic disorders including elevated blood pressure, high plasma glucose, excess body fat around the waist, and abnormal cholesterol or triglyceride levels. These conditions cause serious complications such as heart disease, stroke and type 2 diabetes. In Japan, specific health checkups and specific health guidance which focused on metabolic syndrome has been performed since 2008. Those who fall under certain criteria need to receive a medical treatment guidance from doctor, public health nurse or dietitian. Those who received health guidance receives a reassessment of improvement of their life-style 3-6 months later. However, the efficacy of this approach has not been elucidated. In addition, many persons who have metabolic syndrome do not receive this instruction. Recently, the image analysis technology using the artificial intelligence (AI) progresses rapidly. The smart device application “Asken” has an AI-powered photo analysis system which analyzes the photo of the entire meal, and delivers individualized messages and dietary feedbacks. In this study, we utilized the Internet of Things (IoT) device which includes Asken app, body composition analyzer and sphygmomanometer that can connect wirelessly. Objective: Our aim is to assess the efficacy of specific health guidance adding on IoT device. This is a multicenter, unblinded, non-randomized controlled study. Results: At the end of January 2020, we recruited 219 participants including 105 participants with IoT devices. We used 48 participants (32 with IoT and 16 without IoT) who had finished a reassessment 3 to 6 months after initial guidance. Results: Age, body weight (BW), body mass index (BMI), blood pressure (BP), fasting plasma glucose (FPG), hemoglobin A1c (HbA1c), total cholesterol (T-Chol), high density lipoprotein cholesterol (HDL), low density lipoprotein cholesterol (LDL), non-HDL cholesterol (n-HDL), and triglyceride (TG), did not differ between IoT-use and control group. 6 months after initial guidance, the quantity of decrease of BW in IoT-use group was significantly larger than control (-2.5 ± 4.1 kg vs. 0.6±4.4, p = 0.03). In addition, the quantities of decrease of both T-Chol and n-HDL in IoT-use group were also significantly larger than control (T-Chol, -5.9 ± 32.0 vs. 14.3 ± 31.6, p = 0.02; n-HDL, -7.6 ± 29.0 vs. 9.4 ± 27.5, p = 0.01). Conclusion: Using IoT device might be useful for body weight loss and the improvement of mild hypercholesterolemia in those with metabolic syndrome.

Using low-rank adaptation to fine-tune pre-trained neural network models has attracted widespread attention due to its advantages of low resource requirements, high precision, and no additional inference delay. However, most existing methods are designed for large language models based on the Transformer structure and lack adaptation to convolutional neural networks (CNN) widely used on Internet of Things (IoT) devices. In addition, IoT devices are usually deployed outdoors and collect large amounts of data that are affected by environmental conditions. Providing a highly robust CNN model is a prerequisite for providing high-quality services. To this end, this paper proposes a highly robust seed-filter adaptation method (SeFA) for pre-trained CNNs. SeFA introduces an adaptation branch with the same structure as the backbone network. In the adaptation branch, some filters are first designated seed filters and grouped. Then, additional filters are generated from the grouped seed filters and nonlinear transformation functions (NLFs) with different hyperparameters. The seed filters' parameters are updated during model training, and the NLF hyperparameters are randomly initialized and frozen. Both grouping seed filters and configuring NLFs with non-learnable hyperparameters can improve the model's robustness. This is because grouping seed filters can generate diverse filters on demand without increasing the model's complexity, and the NLFs' rules can regularize the model. It is worth mentioning that the number of fine-tuning parameters of the pre-trained model that can adapt to downstream tasks can be flexibly controlled by specifying the number of seed filters. The key idea of this paper is to propose SeFA with flexible, controllable, learnable parameters, high robustness, and adaptability to pretrained CNN models, thereby facilitating fine-tuning on resource-constrained IoT devices and providing highly robust visual services. Experimental results on the CIFAR-10, CIFAR-10-C, CIFAR-100, CIFAR-100-C, and Icons-50 datasets demonstrate that the proposed SeFA outperforms other state-of-the-art methods. Specifically, on the ResNet-152 model and the CIFAR-10-C dataset, the accuracy of our SeFA is about 7% higher than that of the full fine-tuning method. In addition, we verify the efficiency of SeFA-based models for fine-tuning and inference on the device.

A network-independent tool-based usable authentication system for Internet of Things devices

Ever-increasing advancement in the field of Artificial Intelligence (AI) has caused a spike in the implementation of the Internet of Things (IoT) devices in the industrial as well as the commercial environment. This also demands its assimilation in the sector of power systems, mainly the security and safety part of the power system which faces the most negligence. This paper underlines the economic losses caused by the laxness shown in this department of power and proposes the practices to minimize such losses using AI integrated into IoT devices in the grid and other areas of the power sector. The research focuses on the safety of humans from fires or intruders as well as from any part of the grid which can be a fire hazard if neglected.

According to the World Health Organization, about 15% of the world's population has some form of disability. Assistive Technology, in this context, contributes directly to the overcoming of difficulties encountered by people with disabilities in their daily lives, allowing them to receive education and become part of the labor market and society in a worthy manner. Assistive Technology has made great advances in its integration with Artificial Intelligence of Things (AIoT) devices. AIoT processes and analyzes the large amount of data generated by Internet of Things (IoT) devices and applies Artificial Intelligence models, specifically, machine learning, to discover patterns for generating insights and assisting in decision making. Based on a systematic literature review, this article aims to identify the machine-learning models used across different research on Artificial Intelligence of Things applied to Assistive Technology. The survey of the topics approached in this article also highlights the context of such research, their application, the IoT devices used, and gaps and opportunities for further development. The survey results show that 50% of the analyzed research address visual impairment, and, for this reason, most of the topics cover issues related to computational vision. Portable devices, wearables, and smartphones constitute the majority of IoT devices. Deep neural networks represent 81% of the machine-learning models applied in the reviewed research.

Privacy-Preserving Fuzzy Commitment Schemes (PPFCS) have emerged as a promising solution for secure Internet of Things (IoT) device authentication, addressing the critical need for privacy and security in the rapidly growing IoT ecosystem. This paper presents a novel PPFCS-based authentication mechanism that protects sensitive user data and ensures secure communication between IoT devices. The proposed scheme leverages error-correcting codes (ECC) and cryptographic hash functions to achieve reliable and efficient authentication. The PPFCS framework allows IoT devices to authenticate themselves without revealing their true identity, preventing unauthorized access and preserving users’ privacy. Furthermore, our PPFCS-based authentication mechanism is resilient against various attacks, such as replay, man-in-the-middle, and brute-force attacks, by incorporating secure random nonce generation and timely key updates. We provide extensive experimental results and comparative analysis, demonstrating that the proposed PPFCS significantly outperforms existing authentication schemes in terms of security, privacy, and computational efficiency. As a result, the PPFCS offers a viable and effective solution for ensuring secure and privacy-preserving IoT device authentication, mitigating the risks associated with unauthorized access and potential data breaches in the IoT ecosystem.

Artificial intelligence, fueled by machine learning and deep learning techniques, is revolutionizing various domains. Reinforcement learning (RL) stands out as a potent method for training agents to navigate complex environments and make informed decisions. Our focus is on applying RL techniques, specifically Convolutional Neural Networks (CNNs) combined with policy gradient methods, to enhance the gameplay experience of Khmer chess. Our goal is to surpass the performance of traditional chess engines. The system employs deep neural networks to train AI agents, enabling self-play iterations for strategy refinement. Specifically, we utilize RL technology to iteratively enhance game strategies based on self-matching results, ultimately improving the system's chess proficiency. Our approach entails developing a CNN-based RL system tailored for Khmer chess, encompassing strategies, value evaluation mechanisms, and rule adaptations specific to the game. We utilize deep neural networks to facilitate agent training through self-play iterations, leveraging RL techniques for continual strategy refinement. To enhance training efficiency, we introduce a segmentation method for Khmer chess stages, optimizing the neural network's learning process by mapping game situations to optimal actions based on cumulative rewards. Furthermore, we integrate RL principles to guide action selection towards maximizing reward values, employing Deep Q-Learning with policy gradient for optimal decision-making. With the experimental validation demonstrates the efficacy of our CNN-based RL system in enhancing Khmer chess gameplay. The system exhibits self-improvement, adaptability, and human-like gameplay characteristics, enriching player experience and entertainment value. Moreover, the proposed approach showcases improved training efficiency compared to conventional RL-based chess systems, highlighting its efficacy and scalability for AI-driven game enhancements.

With artificial intelligence (AI) rapidly reshaping global industries, its role in healthcare, and particularly in mental health, is becoming more prominent. The rising adoption of AI for enhancing clinical decision-making and therapeutic processes highlights the need to examine its use in mental health environments. To analyze the AI implementation for detecting mental health disorders is the objective of this review paper, focusing on AI contributions to risk assessment, personalized treatment, and early diagnosis of the disorders. We analyzed published studies from the year 2000 to 2024 using the PRISMA methodology, which explored machine learning (ML) models, deep learning (DL) models, natural language processing (NLP) technique, convolutional neural networks (CNNs), and reinforcement learning in the mental healthcare sector. The results show that AI-powered chatbots and data-driven therapy platforms can enhance access to mental health services, improve treatment, and increase disease detection accuracy. But these advancements also raise ethical concerns related to algorithmic biasness, data security, and the chances of depersonalization of care. Though AI implementation is highly promising, future research is essential to optimize its usage, ensure inclusivity across diverse populations, and address legal and ethical considerations. This paper explores the dynamic landscape of AI mental disorder treatment and underscores the necessity for reliable, transparent and fair AI systems development.

FitCNN: A cloud-assisted and low-cost framework for updating CNNs on IoT devices

We examine data-intensive real-time applications, such as forest fire detection, medical emergency services, oil pipeline monitoring, etc., that require relatively low response time in processing data from the Internet of Things (IoT) devices. Typically, in such circumstances, the edge computing paradigm is utilised to drastically reduce the processing delay of such applications. However, with the growing IoT devices, the edge device cluster needs to be configured properly such that the real-time requirements are met. Therefore, the cluster configuration must be dynamically adapted to the changing network topology of the edge cluster in order to minimise the observed overall communication delay incurred by edge devices when processing data from IoT devices. To this end, we propose an intelligent assignment of IoT devices to edge devices based on Reinforcement Learning such that communication delay is minimised and none of the edge devices is overloaded. We demonstrate, with some preliminary results, that our algorithm outperforms the state-of-the-art.