Wireless communication technologies have easily imprinted their significance into humans' daily lives through Internet of Things (IoT). Recent growth and application of IoT in the smart transportation system, smart cities and so on, possess thriving surface for various types of security attacks with societal, environmental and economic impacts. PUFs (Physically Unclonable Functions) are newly emerging primitive for security consisting of a simple architecture for smart IoT devices including constraints on power. Recently, PUF technology has been widely used for the encryption and authentication of IoT devices. Although, the generation process of the secret key by prior proposed research work possesses challenges of reliability with relation to PUF that affect the security concerns of the applications regarding IoT. It is required to resolve the issues relating to reliability and cost efficiency for smart and IoT devices. For this, the nodes used for communication have to be secure over the network, this causes the reduction in security attacks on the PUF. The proposed work uses a lightweight PUF for enhancing the security issues for IoT devices that have constrained resources. This research work proposes DO-PUF (Dynamic obfuscation - PUF) to secure the IoT devices. Dynamic Obfuscation (DO) integrates the previous input along with generated keys and random numbers, which creates a novel key. DO-PUF is evaluated through designing the instance on machine learning model and it is observed that DO is not only secure but efficient as well.