No abstract

No abstract

To date, physical unclonable functions (PUFs) have been extensively examined, and are used to distinguishgenuine and counterfeit products. Moreover, they are also attracting attention as one of themethods to solve the security problems of Internet of Things (IoT) devices. However, most PUFs arebased on integrated circuit (IC) memory and use digital modulation for authentication. This studyproposes a new PUF that uses analog circuits and analog values for authentication. The advantage ofanalog circuits is that they can handle analog values. Moreover, their characteristics do not changewhen the surrounding environment is adjusted. Research on analog PUFs that evaluate stable signalsand DC voltages has been proposed to date. This study uses an astable multivibrator to analyze PUFsfor unstable signals. For analysis, we examine the conventional method of calculating the hammingdistance of digital values and the method using machine learning(ML). Consequently, we were ableto identify individuals with unsteady signals from analog values when using ML.