Research On Intrusion Detection Technology For Wireless Internet Of Things

In recent years,with the rapid development of wireless communication technology,Io T technology has been widely used in many fields.Attacks against the Internet of Things are also emerging.How to effectively detect various attack behaviors to ensure the security of the wireless Internet of Things has become one of the current research hotspots.This paper proposes an intrusion detection model suitable for the wireless Internet of things to detect various attacks in the wireless Internet of things.The main work of this paper is as follows:1.A wireless Io T intrusion detection model based on ensemble learning is proposed.The MBNMI algorithm is used to reduce the dimension of features,reducing the amount of data storage and calculation;the first-level detection of traffic data is carried out through the improved random forest algorithm.By screening the decision trees in the original random forest,the decision trees with high accuracy and low similarity are used to form the final random forest;for the traffic detected as suspicious at the first level,the original features are encoded by the Light GBM algorithm to generate new features,and the new features are spliced with the original features and then detected and classified by logistic regression.Through experimental tests,the intrusion detection model has good performance.2.A data-augmented wireless Io T intrusion detection model is proposed.This paper proposes a data augmentation scheme based on the fusion of WGAN-GP and DE-GAN algorithms.Based on the WGAN-GP algorithm,the random noise is formed into a noise vector containing the original data information through the pre-trained decoding encoder,and the new noise is used as the input of the generative adversarial network generator,thereby improving the performance of the generative adversarial network.In order to further improve the training speed and stability of the model,Ada Belief is used instead of Adam as the optimizer of the generative adversarial network.The generated data is tested and evaluated by the C2 ST method.The test results show that the generated data is superior to the data generated by the GAN algorithm.Finally,the enhanced data is used to train the intrusion detection model,and the NSL-KDD test data set is used for testing.The results show that the intrusion detection model has better detection performance.