Work

Attack Detection and Mitigation in IoT-Fog Architecture: Handling Class Imbalance Problem

Year: 2023
Type: article
Abstract: The occurrence of data breaches and cyberattacks has greatly increased across numerous companies, organizations, & industries as a result of the exploitation of security holes in IoT devices. There are more zero-day threats presently since more IoT devices are being linked and employ the various protocols. In the realms of big data and cyber-security, DL (deep learning) has shown to be the most effective technique. because it can extract and learn deep features from known assaults and identify novel attacks. Adopting the DL based assaults identification is the greatest crisis in IoT -Fog architecture since it endures with poor or low degree of data privacy. Thereby, this paper focuses on both attack detection and mitigation of attacker in network. The process starts with the class imbalance problem solving via advanced class imbalanced processing. Subsequently, as the extraction of handcrafted features give addition information related to attack behavior, this paper intends to extract the features like correlation-based features, raw data, improved entropy-based features, as well as statistical features. Attack detection will take place based on the retrieved features trained with the DL combo architecture; a novel hybrid detection model combining Bi-GRU and LSTM, detecting the presence of attack in network. However, it is important to mitigate the attacker existing in the network, and hence a new entropy based mitigation procedure is followed in this article. Finally, the results and discussion section shows the efficiency of proposed work over the conventional methods in terms of different performance analysis.
Language: English
Open Access status: diamond
OpenAlex ID: W7217281880
FWCI: 0.155
Cites: 23
Cited by: 2
Related to: 10