Software Defined Networks (SDN) as act as a challenging tool for intersecting the dynamic demands of the network management in terms of flexibility and effectiveness. With the exponential enhancement in network traffic, constraints need the most significant Quality of Services (QoS), a requirement for the routing optimization. In the introduction phase of SDN-based Internet of Things (IoT), the sampling-based security method had resulted in poor performance on the attack detection. Hence, this research proposes the Markov Chain-based Artificial Neural Network (MC-ANN) approach for dynamic routing optimization in SDN. The controller of the SDN effectively identifies the routing path with the help of group of link weight as well as establishes the guidelines on SDN-enabled switches. The success of the proposed MC-ANN approach is estimated through various performance metrics like delay, packet loss, and throughput. The proposed MC-ANN approach attains the better throughput of 9.4 Mbps on the 100,000 iterations when compared to the Deep Reinforcement Learning (DRL).