Performance and accuracy trade-off analysis of techniques for anomaly detection in IoT sensors

Detecting abnormal sensors via machine learning: An IoT farming WSN-based architecture case study

“…However, choosing the wrong algorithm may delay or even compromise the detection process. Some studies also compare anomaly detection algorithms [11] [12]. However, their analysis does not cover important aspects such as the impact of running anomaly detection algorithms in edge devices regarding temperature and power consumption.…”

Evaluating Energy and Thermal Efficiency of Anomaly Detection Algorithms in Edge Devices

“…However, choosing the wrong algorithm may delay or even compromise the detection process. Some studies also compare anomaly detection algorithms [19] [20] [6]. However, their analysis does not cover important aspects such as the impact of running anomaly detection algorithms in edge devices regarding temperature and power consumption.…”

“…In order to provide efficient and reliable feedback, data processing becomes a priority task, while other tasks are often overlooked. As a consequence, sensors may be compromised by failures or even manipulated by attackers to induce erroneous decision-making [5] [6].…”

IoT Big Data Value Map

“…In supervised learning, a user defines a target value and classifies each instance with explicitly predefined labels and then these techniques put the future instances into proper classes. On the other hand, in unsupervised learning techniques, instead of users, it is the machine that automatically classifies instances based on similarities [5]. K-Nearest Neighbors (kNN), decision trees, Bayesian classifiers, logistic regression, Support Vector Machines (SVM), and Artificial Neural Networks (ANN) are some algorithms among the supervised learning techniques.…”