ABSTRACT
This chapter presents some work on advancing the physical disorder modeling with the help of Machine Learning techniques. It shows that through modeling the physical disorder, Machine Learning techniques can benefit the performance of physically unclonable functions (PUFs) by filtering out the unreliable challenge and responses (CRPs). The chapter describes the sources of process variations, i.e., physical disorder in modern complementary metal-oxide-semiconductor (CMOS) circuits. While the development of semiconductor technology is advancing into the nanometer realm, one significant characteristic of CMOS fabrication is the random nature of process variability, i.e., the physical disorder of CMOS transistors. The chapter provides an introduction to common PUF terminologies and performance metrics. Environmental variations are detrimental to PUF circuits. Some of the common environmental sources of variations include power supply noise, temperature fluctuations and external noise. These variations must be minimized to improve the reliability of PUF circuits.
