Reinforcing IoT Security through Machine Learning Based Spam Detection
R. Chatrapathi, M.O. Ramkumar, Dontabhaktuni Jayakumar, R.Pooja Sri · 2023
Millions of linked sensors and actuators make up the Internet of Things (IoT), a fast-growing network that transmits data over wired or wireless communication channels. IoT has grown significantly in the last ten years alone, with an estimated 43 billion connected devices predicted to be in use by 2023. Machine learning algorithms may be quite useful in this situation for establishing biotechnology-based security and authorization as well as for spotting abnormalities to improve the security and usability of IoT devices. Attackers frequently consider machine learning techniques to be possible weaknesses in sophisticated IoT devices, though. In order to secure IoT devices, the research suggests employing machine learning to identify spam. A machine-learning framework for spam detection in an IoT is the suggested method for achieving this goal. This framework assesses five alternative machine learning models using various measures and a wide range of input feature sets. A spam score is generated by each model using the improved input characteristics. Based on a number of variables, this score shows the degree of trustworthiness of an IoT device. The IoT-23 dataset is used to validate the proposed method. The findings show that the suggested solution is superior to the current approaches in terms of effectiveness.