چكيده به لاتين
Authenticating of integrated circuits and secure hardware-based encryption protocols play a crucial role in hardware security. The use of smart cards, RFID tags, and the Internet of Things (IoT) is expanding, and secure communication between these devices is of paramount importance. Additionally, among the constraints of applications with limited resources, power consumption, and integrated circuit area, the need for lightweight encryption protocols arises. Physically Unclonable Functions (PUFs) play a significant role in achieving this goal. Considering the challenges of CMOS technology, especially in dimensions below 100 nanometers, recent efforts have focused on designing PUF circuits using nanoscale magnetic elements to improve boot-up speed, data security, and reduce static power consumption. Therefore, in achieving optimal and secure encryption algorithms in IoT applications with limited resources, this thesis will focus on the examination and design of PUF circuits based on nanomagnetic elements.
To this end, two structures of all-magnetic PUFs are proposed, where the first structure is a Delay-based PUF, and the second structure is Memory-based PUF. The Delay-based PUF structure, maintaining desirable characteristics such as randomness and uniqueness, has improved reliability, power consumption, and area overhead up to 270 times compared to other Delay PUF circuits. Additionally, the Memory-based PUF structure, while providing similar security characteristics compared to the referenced structures, has significantly improved power consumption and the area of each memory cell by 3 and 600 times, respectively. Another innovation in the thesis is the combination of a Memory-based PUF structure with a Compute-In-Memory (CIM) circuit to increase security resulting from key generation and encryption computations within the memory itself without disturbance to stored data. This has led to a thriple-mode structure. Considering that the implementation of the entire encryption algorithm, from reading data from memory to key generation and encryption operations, is performed within the memory structure, eliminating the need for separate computation circuits and additional buffers to preserve data, power consumption and area have significantly reduced compared to other structures with separate circuits, resulting in a noticeable increase in security.
