Illegal Intrusion Detection of Internet of Things Based on Deep Mining Algorithm

Abstract

In this study, to reduce the influence of The Internet of Things (IoT) illegal intrusion on the transmission effect, and ensure IoT safe operation, an illegal intrusion detection method of the Internet of Things (IoT) based on deep mining algorithm was designed to accurately detect IoT illegal intrusion. Moreover, this study collected the data in the IoT through data packets and carries out data attribute mapping on the collected data, transformed the character information into numerical information, implemented standardization and normalization processing on the numerical information, and optimized the processed data by using a regional adaptive oversampling algorithm to obtain an IoT data training set. The IoT data training set was taken as the input data of the improved sparse auto-encoder neural network. The hierarchical greedy training strategy was used to extract the feature vector of the sparse IoT illegal intrusion data that were used as the inputs of the extreme learning machine classifier to realize the classification and detection of the IoT illegal intrusion features. The experimental results indicate that the feature extraction of the illegal intrusion data of the IoT can effectively reduce the feature dimension of the illegal intrusion data of the IoT to less than 30 and the dimension of the original data. The recall rate, precision, and F1 value of the IoT intrusion detection are 98.3%, 98.7%, and 98.6%, respectively, which can accurately detect IoT intrusion attacks. The conclusion demonstrates that the intrusion detection of IoT based on deep mining algorithm can achieve accurate detection of IoT illegal intrusion and reduce the influence of IoT illegal intrusion on the transmission effect.

Keywords

• data classification
• deep mining algorithm
• extreme learning machine
• feature extraction
• illegal intrusion detection
• IoT