IoT, Volume 3, Issue 2 (June 2022) – 4 articles

Global demand for sleep-tracking wearables, or consumer sleep technologies (CSTs), is steadily increasing. CST marketing campaigns often advertise the scientific merit of devices, but these claims may not align with consensus opinion from sleep research experts. Consensus opinion about CST features has not previously been established in a cohort of sleep researchers. This case study reports the results of the first survey of experts in real-world sleep research and a hypothetical purchase task (HPT) to establish economic valuation for devices with different features by price. Forty-six (N = 46) respondents with an average of 10 ± 6 years’ experience conducting research in real-world settings completed the online survey. Total sleep time was ranked as the most important measure of sleep, followed by objective sleep quality, while sleep architecture/depth and diagnostic information were ranked as least important. A total of 52% of experts preferred wrist-worn devices that could reliably determine sleep episodes as short as 20 min. The economic value was greater for hypothetical devices with a longer battery life. These data set a precedent for determining how scientific merit impacts the potential market value of a CST. This is the first known attempt to establish a consensus opinion or an economic valuation for scientifically desirable CST features and metrics using expert elicitation. Full article

Using the Internet of Things (IoT) for various applications, such as home and wearables devices, network applications, and even self-driven vehicles, detecting abnormal traffic is one of the problematic areas for researchers to protect network infrastructure from adversary activities. Several network systems suffer from drawbacks that allow intruders to use malicious traffic to obtain unauthorized access. Attacks such as Distributed Denial of Service attacks (DDoS), Denial of Service attacks (DoS), and Service Scans demand a unique automatic system capable of identifying traffic abnormality at the earliest stage to avoid system damage. Numerous automatic approaches can detect abnormal traffic. However, accuracy is not only the issue with current Intrusion Detection Systems (IDS), but the efficiency, flexibility, and scalability need to be enhanced to detect attack traffic from various IoT networks. Thus, this study concentrates on constructing an ensemble classifier using the proposed Integrated Evaluation Metrics (IEM) to determine the best performance of IDS models. The automated Ranking and Best Selection Method (RBSM) is performed using the proposed IEM to select the best model for the ensemble classifier to detect highly accurate attacks using machine learning and deep learning techniques. Three datasets of real IoT traffic were merged to extend the proposed approach’s ability to detect attack traffic from heterogeneous IoT networks. The results show that the performance of the proposed model achieved the highest accuracy of 99.45% and 97.81% for binary and multi-classification, respectively. Full article

Federated Learning (FL) is a state-of-the-art technique used to build machine learning (ML) models based on distributed data sets. It enables In-Edge AI, preserves data locality, protects user data, and allows ownership. These characteristics of FL make it a suitable choice for IoT networks due to its intrinsic distributed infrastructure. However, FL presents a few unique challenges; the most noteworthy is training over largely heterogeneous data samples on IoT devices. The heterogeneity of devices and models in the complex IoT networks greatly influences the FL training process and makes traditional FL unsuitable to be directly deployed, while many recent research works claim to mitigate the negative impact of heterogeneity in FL networks, unfortunately, the effectiveness of these proposed solutions has never been studied and quantified. In this study, we thoroughly analyze the impact of heterogeneity in FL and present an overview of the practical problems exerted by the system and statistical heterogeneity. We have extensively investigated state-of-the-art algorithms focusing on their practical use over IoT networks. We have also conducted a comparative analysis of the top available federated algorithms over a heterogeneous dynamic IoT network. Our analysis shows that the existing solutions fail to effectively mitigate the problem, thus highlighting the significance of incorporating both system and statistical heterogeneity in FL system design. Full article

The Internet of Things is enriching our life with an ecosystem of interconnected devices. Object cooperation allows us to develop complex applications in which each node contributes one or more services. Therefore, the information moves from a provider to a requester node in a peer-to-peer network. In that scenario, trust management systems (TMSs) have been developed to prevent the manipulation of data by unauthorized entities and guarantee the detection of malicious behaviour. The community concentrates effort on designing complex trust techniques to increase their effectiveness; however, two strong assumptions have been overlooked. First, nodes could provide the wrong services due to malicious behaviours or malfunctions and insufficient accuracy. Second, the requester nodes usually cannot evaluate the received service perfectly. For this reason, a trust system should distinguish attackers from objects with poor performance and consider service evaluation errors. Simulation results prove that advanced trust algorithms are unnecessary for such scenarios with these deficiencies. Full article