{"title":"Botnet Attack Detection in IoT Networks using CNN and LSTM","authors":"Anuj Sharma, Prasoon Mishra, Dr. G. Geetha","doi":"10.1109/ICECAA58104.2023.10212330","DOIUrl":null,"url":null,"abstract":"Botnet attacks are a major concern for IoT devices, but using deep learning (DL) to identify them requires significant memory space and network traffic, making it difficult to implement on devices with limited memory. One can use dimensionality reduction methods to decrease the number of features in IoT network traffic data. The Bot-IoT dataset is a dataset that is accessible to the public, and it can be utilized to identify botnet attacks in IoT networks., with millions of samples of botnet attack traffic classified into DDoS, DoS, reconnaissance, and information theft scenarios. Dimensionality reduction techniques like principal component analysis (PCA) and autoencoder can help reduce the feature dimensionality of the dataset. Autoencoder, an unsupervised deep learning technique generates a hidden layer's latent-space representation of the input data. The reduced feature set can be used by deep learning algorithms like Long Short Term Memory (LSTM) and Convolutional Neural Network (CNN) to detect botnet attacks. Performance measurements like accuracy, precision, recall, and confusion matrix can be used to evaluate the effectiveness of the approach. In summary, the proposed approach uses dimensionality reduction techniques like PCA and autoencoder to reduce the feature dimensionality of the Bot-IoT dataset, making it feasible to use DL algorithms like LSTM and CNN to identify botnet attacks. Performance metrics can be used to evaluate the effectiveness of the approach.","PeriodicalId":114624,"journal":{"name":"2023 2nd International Conference on Edge Computing and Applications (ICECAA)","volume":"145 1","pages":"0"},"PeriodicalIF":0.0000,"publicationDate":"2023-07-19","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"2023 2nd International Conference on Edge Computing and Applications (ICECAA)","FirstCategoryId":"1085","ListUrlMain":"https://doi.org/10.1109/ICECAA58104.2023.10212330","RegionNum":0,"RegionCategory":null,"ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"","JCRName":"","Score":null,"Total":0}
引用次数: 0
Abstract
Botnet attacks are a major concern for IoT devices, but using deep learning (DL) to identify them requires significant memory space and network traffic, making it difficult to implement on devices with limited memory. One can use dimensionality reduction methods to decrease the number of features in IoT network traffic data. The Bot-IoT dataset is a dataset that is accessible to the public, and it can be utilized to identify botnet attacks in IoT networks., with millions of samples of botnet attack traffic classified into DDoS, DoS, reconnaissance, and information theft scenarios. Dimensionality reduction techniques like principal component analysis (PCA) and autoencoder can help reduce the feature dimensionality of the dataset. Autoencoder, an unsupervised deep learning technique generates a hidden layer's latent-space representation of the input data. The reduced feature set can be used by deep learning algorithms like Long Short Term Memory (LSTM) and Convolutional Neural Network (CNN) to detect botnet attacks. Performance measurements like accuracy, precision, recall, and confusion matrix can be used to evaluate the effectiveness of the approach. In summary, the proposed approach uses dimensionality reduction techniques like PCA and autoencoder to reduce the feature dimensionality of the Bot-IoT dataset, making it feasible to use DL algorithms like LSTM and CNN to identify botnet attacks. Performance metrics can be used to evaluate the effectiveness of the approach.