RF-PSF: Zero-Trust Radio Frequency Process Specific Functions as Process Distinction Method

In this work, we propose a manufacturing process technology distinction method, RF-PSF, that applies artificial intelligence to the transmitted RF signal to exploit process-specific inherent properties embedded in it. By performing the design and simulation of a QPSK transmitter in 14 nm, 22 nm, and 65 nm process technologies, we have shown that on average ~ 90% accuracy can be achieved for process distinction with > 99% in the best-case scenario. The effect of sampling rate and quantization, two practical limitations in RF systems, have also been explored. This work establishes RF-PSF as a process distinction method towards zero-trust radio architectures.