Abstract:As widely known, most of the Internet of Things (IoT) devices own small storage and constrained computing power, and hence, their poor security evaluation capabilities make them vulnerable to several types of network attacks. Given this setting, anomaly network traffic detection techniques based on deep learning (DL) offer some practical solutions, and they have brought new opportunities to the security of the IoT. However, existing DL models for anomaly network traffic detection need better flexibility and classification accuracy. Also, the scale of those models needs to be optimized, as a sheer majority of them need to be more suitable for deployment on terminal devices of IoT. Therefore, we propose an anomaly network traffic detection model in this work LMCA, standing for Lightweight Model Integrating adjusted MobileNet and Coordinate Attention mechanism. Combining the adjusted MobileNet model and the coordinate attention mechanism, it constructs a lightweight anomaly network traffic detection model and effectively extracts traffic data's local, global, and spatial–temporal features, which would be easy to deploy on IoT terminals. LMCA has a small scale and good performance, making it suitable for IoT environments. Moreover, we use an original traffic feature extraction method to reduce redundant features and speed up neural network convergence. This work also solves a problem so that the original MobileNet model could perform better on a small dataset, extending the anomaly traffic detection for IoT. To simulate the IoT environment, we used the wired network dataset CICDS2017 and the wireless network dataset AWID. Experimental results demonstrate that the proposed work outperforms other existing methods, the accuracy reached 99.96% on the CICIDS2017 dataset and 99.98% on the AWID dataset.

Efficient Log Anomaly Detection Based on Dimension Reduction and Attention Aware TCN

LightLog: A Lightweight Temporal Convolutional Network for Log Anomaly Detection on the Edge

TinyLog: Log Anomaly Detection with Lightweight Temporal Convolutional Network for Edge Device

TCN-based Lightweight Log Anomaly Detection in Cloud-edge Collaborative Environment

TCN-Log2Vec: A Comprehensive Log Anomaly Detection Framework Based on Optimized Log Parsing and Temporal Convolutional Network

Reducing Events to Augment Log-based Anomaly Detection Models: An Empirical Study

ETCNLog: A System Log Anomaly Detection Method Based on Efficient Channel Attention and Temporal Convolutional Network

Federated Anomaly Detection on System Logs for the Internet of Things: A Customizable and Communication-Efficient Approach

Anomaly Detection in Unstructured Logs Using Attention-based Bi-LSTM Network

Natural Language Processing-based Model for Log Anomaly Detection

Log anomaly detection method based on CNN and LSTM fusion

Improving Performance of Log Anomaly Detection with Semantic and Time Features Based on BiLSTM-Attention

Log-Based Anomaly Detection with Multi-Head Scaled Dot-Product Attention Mechanism

DualAttlog: Context aware dual attention networks for log-based anomaly detection

Log2graphs: An Unsupervised Framework for Log Anomaly Detection with Efficient Feature Extraction

Efficient Log-based Anomaly Detection with Knowledge Distillation

Log-based Anomaly Detection based on EVT Theory with feedback

LMCA: a lightweight anomaly network traffic detection model integrating adjusted mobilenet and coordinate attention mechanism for IoT

Improving Log-Based Anomaly Detection with Component-Aware Analysis

Log-based Anomaly Detection Without Log Parsing

LogAnomaly: Unsupervised Detection of Sequential and Quantitative Anomalies in Unstructured Logs