A Federated Learning-Based Approach for Improving Intrusion Detection in Industrial Internet of Things Networks

Abstract

The Internet of Things (IoT) is a network of electrical devices that are connected to the Internet wirelessly. This group of devices generates a large amount of data with information about users, which makes the whole system sensitive and prone to malicious attacks eventually. The rapidly growing IoT-connected devices under a centralized ML system could threaten data privacy. The popular centralized machine learning (ML)-assisted approaches are difficult to apply due to their requirement of enormous amounts of data in a central entity. Owing to the growing distribution of data over numerous networks of connected devices, decentralized ML solutions are needed. In this paper, we propose a Federated Learning (FL) method for detecting unwanted intrusions to guarantee the protection of IoT networks. This method ensures privacy and security by federated training of local IoT device data. Local IoT clients share only parameter updates with a central global server, which aggregates them and distributes an improved detection algorithm. After each round of FL training, each of the IoT clients receives an updated model from the global server and trains their local dataset, where IoT devices can keep their own privacy intact while optimizing the overall model. To evaluate the efficiency of the proposed method, we conducted exhaustive experiments on a new dataset named Edge-IIoTset. The performance evaluation demonstrates the reliability and effectiveness of the proposed intrusion detection model by achieving an accuracy (92.49%) close to that offered by the conventional centralized ML models’ accuracy (93.92%) using the FL method.

Review on Approaches of Federated Modeling in Anomaly-Based Intrusion Detection for IoT Devices

“…The Unisa Malware Dataset (UMD) was utilized by [110], while the CICIDS2017 dataset found attention from [81,111]. Multiple authors were involved in studying the TON_IoT dataset [55,83,84,90,93,94,112], while on N-BaIoT dataset [112], MNIST dataset [91], Edge-IIoTset dataset [77,92], CIC_IoT dataset [80], CSE-CIC-IDS dataset [113]. The SCADA dataset [88,89], NF-BoT-IoT-v2 dataset [114], while BoT-IoT dataset [79,86,95], MQTT dataset [115], and the Power Demand dataset [85].…”

Employing Hybrid ANOVA-RFE with Machine and Deep Learning Models for Enhanced IoT and IIoT Attack Detection and Classification

“…This approach was further investigated by Rashid et al [12] in the context of industrial IoT systems, with their experiments on the Edge-IIoTset dataset yielding an accuracy of 93.92%.…”

Next–Generation Intrusion Detection for IoT EVCS: Integrating CNN, LSTM, and GRU Models

“…Our objectives are centered on developing an ensemble IDS model using convolutional neural network (CNN) [3], long short-term memory (LSTM), and gated recurrent unit (GRU) models, evaluating its performance with the "Edge-IIoTset" dataset [4], optimizing it for resource efficiency, and benchmarking it against existing solutions. We evaluate the ensemble model's efficacy in enhancing detection accuracy, its performance using comprehensive datasets, its feasibility in resource-constrained environments, and its adaptability to evolving cyber threats.…”

“…Advanced Data Processing Techniques: Sophisticated preprocessing techniques are employed to manage complex IoT security data, enhancing model learning efficiency. 4.…”