STFNIoT:Lightweight IoT Intrusion Detection Based on Explainable Analysis Using Spatiotemporal Fusion Networks

Research Square (Research Square), Год журнала: 2025, Номер unknown

Опубликована: Янв. 28, 2025

With the widespread popularity of IoT applications, devices are increasingly becoming targets cyber attacks. Existing intrusion detection systems usually face computing resource limitations and accuracy challenges when facing complex, high-dimensional attack traffic data. Therefore, this paper proposes a lightweight framework STFNIoT based on interpretable analysis spatiotemporal fusion networks, which combines principal component (PCA) deep learning models to address above problems. PCA performs data dimensionality reduction reduce feature redundancy while retaining key information. Subsequently, network(STFN) is used for learning. STFN contains two components: convolutional neural network (CNN) extracting spatial features bidirectional long short-term memory (BiLSTM) capturing time-dependent features, thereby efficiently relationship between devices. In addition, integrates SHAP interpretability algorithm, can intuitively reveal decision-making process model enhance transparency reliability system. Experimental results show that achieves 100%, 97.70% 97.15% in binary, hexaclass multiclass tasks Edge-IIoTset dataset, respectively, significantly improving performance compared with existing methods. modular design effectively reduces computational overhead suitable resource-constrained environments. This study provides an efficient explainable method.

Язык: Английский

Advances in Acoustic Emission Monitoring for Grinding of Hard and Brittle Materials

Journal of Materials Research and Technology, Год журнала: 2025, Номер unknown

Опубликована: Март 1, 2025

Язык: Английский

Design and Performance Optimization of High Efficiency Wireless Sensor Network Data Transmission Algorithm

Applied Mathematics and Nonlinear Sciences, Год журнала: 2024, Номер 9(1)

Опубликована: Янв. 1, 2024

Abstract Wireless Sensor Networks (WSN), as the cornerstone of modern Internet Things (IoT) technology, achieve comprehensive perception and real-time transmission physical world information by densely deploying small lowpower sensor nodes in target areas, greatly promoting interconnectivity between people things, things. However, limited energy communication capabilities make efficient reliable data a major challenge WSN design big environment. To address this challenge, paper proposes an innovative optimization algorithm based on Ant Colony Optimization Neural Network (ACO-NN). This combines global search capability ACO with powerful learning ability neural networks. Specifically, utilizes to explore accumulate pheromones different paths, while using networks evaluate predict path quality, thereby guiding selection paths. The experimental results show that proposed can significantly improve efficiency reliability transmission, reduce consumption, extend network lifespan.

Язык: Английский