A novel platform of dispersed streaming is developed by the fog paradigm for the applications associated with the internet of things. The sensed information's of the IOT plat form is collected from the edge device closer to the user from the lower plane and moved to the fog in the middle of the cloud and edge and then further pushed to the cloud at the top most plane. The information's gathered at the lower plane often holds unanticipated values that are of no use in the application. These unanticipated or the unexpected data's are termed as anomalies. These unexpected data's could emerge either due to the improper edge device functioning which is usually the mobile devices, sensors or the actuators or the coincidences or purposeful attacks or due to environmental changes. The anomalies are supposed to be removed to retain the efficiency of the network and the application. The deep learning frame work developed in the paper involves the hardware techniques to detect the anomalies in the fog paradigm. The experimental analysis showed that the deep learning models are highly grander compared to the rest of the basic detection structures on the terms of the accuracy in detecting, false-alarm and elasticity.

Deep Learning Fog Computing Anomalies Detection Accuracy in Detecting False-Alarm and Elasticity