Assessing the Effectiveness of Siamese Neural Networks to Mitigate Frequent Retraining in IoT Device Identification Models

Internet-of-Things Traffic Analysis and Device Identification Based on Two-Stage Clustering in Smart Home Environments

“…Hence, as also shown in Figures 11 and 12, we can conclude that two-stage clustering can more appropriately extract features that are important to describe IoT traffic, and the time series representations based on two-stage clustering can express not only whether but also whatever an IoT device communicates. Table 4 shows a comparison of the identification performance outcomes between the proposed method and the conventional methods [6][7][8][9][10][11][12][13][14]. Although it is difficult to strictly compare among them due to the lack of detailed data, the identification accuracy of the proposed method is higher than some of the conventional methods [6][7][8] but slightly lower than the others [9][10][11][12][13][14].…”

“…Table 4 shows a comparison of the identification performance outcomes between the proposed method and the conventional methods [6][7][8][9][10][11][12][13][14]. Although it is difficult to strictly compare among them due to the lack of detailed data, the identification accuracy of the proposed method is higher than some of the conventional methods [6][7][8] but slightly lower than the others [9][10][11][12][13][14]. This could be because some traffic features were lost due to feature extraction by clustering.…”

“…Trad et al proposed a method to mitigate frequent retraining for IoT device identification models [13,14]. A Siamese neural network (SNN) was trained to generate embeddings corresponding to the similarities of feature values.…”

“…Therefore, the database was extended to include the new device type, and the model could recognize the new device without the updating. Based on the results in [14], this method yielded an 85.8% F-measure even when 28 unknown IoT device data were added after training. However, this method also processed traffic data at once, whereas all traffic data should be shared with the cloud servers in the real environments.…”

“…For this purpose, it is important to automatically identify connected IoT devices in some way. Several methods for identifying IoT devices or their behavior based on traffic analysis using machine learning have been proposed [6][7][8][9][10][11][12][13][14]. Although these methods automatically identify IoT devices and their behavior from traffic data, they do not consider how they work in smart home environments.…”