Abstract:Device type identification (DTI) is a significant system to recognise several device types based on internet of things (IoT) management. If an infected IoT device is not isolated from the network for a certain amount of time, it cause cross-contamination and introduce malware in the whole network. To overcome this problem, a graph sample and aggregate-attention network optimised with arithmetic optimisation algorithm fostered IoT device type identification (GrSAgAN-AOA-DTI-IoT) is proposed for enhancing IoT security. The network traffic feature vector contains maximum, minimum, mean, variance, and kurtosis, which are extracted by TF-IDF. These extracting features are supplied to the IoT device type identification phase. The proposed GrSAgAN-AOA-DTI-IoT approach is activated in Python. The GrSAgAN-AOA-DTI-IoT method attains higher accuracy 29.36%, 32.67%, 36.14% and 21.33% and lower computational complex 16.39%.11.39%, 8.36% and 14.31% compared to the existing methods.

Aggregate-attention network optimised and graph sample with arithmetic optimisation algorithm fostered IoT device type identification for enhancing IoT security