Date of Publication :25th July 2024

Abstract:Security has been always a serious concern in Internet-of-Things (IoT) that is characterized by highly interconnected devices and appliance thereby forming a complex form of network. At present, there are various form of solution that are chosen as security measures to safeguard IoT devices and services from various form of lethal threats. However, all the solutions are mainly in form of encryption, which has quite a limitation towards resisting dynamic threats. It is noted that Artificial Intelligence (AI) based scheme has been making a potential contribution towards solving the similar security threats in IoT in predictive manner. Ir-respective of archives of literatures where AI has been implemented in the form of machine learning and deep learning methods; however, no conclusive information is yet stated towards its effectiveness. Hence, this paper reviews the existing AI based approaches and contributes towards this issue by offering a com-pact, precise, and effective findings stating the true effectiveness of existing learn-ing- based methods. Further, it is noted that existing system suffers from various limitation that are associated with issues with various learning approaches as well as dataset too. The impact of such limitation is that reduced strength to identify and resist dynamic forms of complex security threats in IoT. Finally, the paper contributes towards offering a suggested methodology where adoption of hybrid learning approach can be used to address the identified gaps in existing learning approaches towards strengthening IoT security.

Reference :