Abstract—Security is critically important for Internet-of-Things, but existing cryptographic protocols are not lightweight enough for resource-constrained IoT devices. Implementable with simplistic circuits and operable with shallow power, physical unclonable functions (PUFs) leverage small but unavoidable physical variations of the circuit to produce unique responses for individual PUF instances, rendering themselves good candidates as security primitives for IoT devices. Component-differentially-challenged XOR PUFs (CDC XPUFs) are among the PUFs which were shown to be highly secure to machine learning modeling attacks. However, no study of implementation and experimentation has been carried out. In this paper, we report our implementations of CDC XPUFs on FPGAs and experimental studies of the essential properties of CDC XPUFs.