Abstract
The paper presents a mathematical model for authentication of peripheral devices in industrial IoT networks operating in decentralized environments. The model is based on the use of SRAM-PUF to generate 512-bit identifier signals, which increases the probability of collisions to 10−154. To correct errors caused by hardware noise, a hybrid LDPC code (768512) is used, providing correction of up to 10% of bit errors while reducing computational costs by 37% compared to classical BCH codes. A decentralized rejection validation protocol from centralized servers using symmetric AES-GCM encryption to protect challenge-response pairs. Experimental results of the model's resistance to brute-force attacks (FAR = 0%), reduced false rejection rate (FRR = 1.8%) and energy efficiency (0.9 mJ per operation) are obtained. The proposed approach ensures scalability for networks with millions of devices and compatibility with modern cybersecurity standards.