Performance Evaluation of SVM-based Diagnosis Training Model for IOT Sensors
Since vast amounts of sensor data are collected every second in IoT, immediate fault diagnosis is very difficult. Thus, the efficient offloading process eliminating worthless data is very important. This study introduces IoT fault diagnosis issues in IoT edge computing, and focuses on efficient data screening for IoT sensors.
This study proposes SVM-based filtering model for efficient sensor data offloading to reduce data and network overhead costs of remote cloud servers. The proposed model exploits adaptive sensor candidate filtering using fault relevance training. This study also evaluates the performance of the proposed filtering model through Google CoLab, and the test results show that the prediction accuracy can be increased up to 91% in a public database.
Keywords - AI-based Training Model, Support Vector Machine, IoT Fault Diagnosis, CoLab, Performance Evaluation.