INCREMENTAL DEEP NEURAL NETWORK INTRUSION DETECTION IN FOG BASED IOT ENVIRONMENT: AN OPTIMIZATION ASSISTED FRAMEWORK

Abstract

IoT has gained more attention on sharing data and directing respective tasks. IoT also has the potential to enhance the lifestyle of people. Fog computing has been evolved to process information to make its service more efficient. However, the evolution of various technologies also raises difficulties in network security. This paper intends to introduce a new intrusion detection model in fog-based IoT considering the phases like Preprocessing, Feature Extraction, and Intrusion Detection. Initially, data normalization is carried out in the raw data (considered as pre-processing). Then, the pre-processed data is subjected to the feature extraction phases, where the proposed entropy-based features, gain information, and gain ratio features are extracted. Subsequently, the extracted features are subjected to the attack detection phase, where the Incremental Deep Neural Network (DNN) will detect the presence of an intruder in the network based on the data attributed. The training of DNN is made optimal via tuning the weights. This optimization is carried out by a Modified Electric Fish Optimization Algorithm (M-EFO). At last, the supremacy of the proposed developed model is examined via evaluation over extant techniques.