Contents I. Introduction
With the fourth industrial revolution, the Internet of Things (IoT) has rapidly advanced and widely gained acceptance by various applications, such as healthcare, agriculture, smart home, mobility, and industrial processing [1], [2]. As the data transfer among IoT devices drastically increases, however, the existing von Neumann architecture encountered speed and energy limits due to memory bottlenecks. In-memory computing has been developed to resolve memory bottlenecks by processing data entirely in memory without transferring data to a processing unit. Content addressable memory (CAM), a specialized memory for high-speed searching, is one of the in-memory-computing techniques required to perform the various functionality of IoT devices. CAMs are generally used for applications requiring high-speed table lookup and operate as network routers for pack forwarding or classifications [3]. Besides, the recent development in data processing has explored new CAM applications, such as neuromorphic associative memory, pattern matching, and reconfigurable computing [4]. However, as the number of connections among IoT devices increases, more security threats occur from unconstrained access by the public [5]. Thus, IoT devices need to perform additional cryptographic functions for handling these security threats. Conventional cryptography that stores random keys generated from a random number generator in nonvolatile memory has limitations in both cost and security level because it requires two separate chips and has a risk of information leakage while transmitting data between the two chips [6]. These limitations of conventional cryptography led to physically unclonable function (PUF) being actively studied [7]. Since PUF acts as both a random key generator and storage of the generated keys using physical characteristics of integrated circuits (IC), it is cost-effective while showing a higher level of security than conventional cryptography. As a result, PUF has become a lightweight solution to fulfill the security requirements of IoT devices [8].
