No-Interaction School Attendance Scheme Depending On Machine Learning Concepts and RSSI On 2.4 GHz Frequency Band

Most educators take their students’ attendance in different ways including roll call, passing an attendance sheet. In addition to the time required for this process, current practices are also susceptible to spreading contagious diseases (e.g., COVID-19, COVID-19 Delta Variant). The concept of no-interaction school attendance system can help us not only to evade the waste of valuable class time via attendance automation, but also prevent the spread of contagious diseases. This paper proposes a no-interaction school attendance scheme that utilizes machine learning concepts over Received Signal Strength Indicator (RSSI) values on the Wi-Fi 2.4GHZ frequency band. It frames the problem of attendance recording as binary classification problem where a student is present if he/she is adjacent to his peers that belong to the same class room and absent otherwise. A set of novel feature vectors was computed by considering the sequences of RSSI values as time series input to binary classifiers to predict a student’s location in a classroom. The proposed prediction model achieved a maximum of 96% accuracy.