To Evaluate and Analyze the Performance of Anomaly Detection in Cloud of Things

The Internet of Things (IoT) has become a household name in our day-to-day life. The IoT concept offers several opportunities to improve human life as well as save the time, but it also has the potential to increase the vulnerability of personal information once it is collected. In terms of security and privacy, data collected in IoT is typically offloaded to the cloud, which leaves data vulnerable to a variety of attacks. The integration of IoT and cloud can be considered as cloud of things (CoT). In this research, we look for abnormalities in security measures that were once extensively used, such as wired or wireless networks. The huge number of sources that generate these data, as well as the connectivity and traffic patterns in CoT. This research focuses on evaluating the performance of anomaly detection in the CoT using three primary machine learning algorithms: KNN, CNN, and Naive Bayes. The evaluation of the anomalies on the BotIoT dataset which has been created using the simulation of IoT devices and routers. The best accuracy is achieved by CNN is 99.94% with 99.5% F1-score. Our evaluation demonstrates the promising results in terms of RoC curve and confusion matrix.