Since the open nature of Internet of Things (IoT), it is more vulnerable for various attacks that degrades the network topologies and functionalities. Rather than other attacks, Rank Attack (RA) becomes the most dangerous attack mainly affecting the rank values of routing process. Thus, it reaches more attention in “Routing Protocol for Low Power and Lossy Networks (RPL)” as well. Hence, the trust-based secured IoT network is a quite challenging process. The former methods are limited with overhead problem, more energy consumption and RA, which is a basic security attack on routing in IoT networks. In order to tackle and to detect the attack, the secured IoT routing against rank attacks framework is developed. Initially, the data is aggregated from benchmark datasets. Further, the aggregated data is given to the Multiscale Depth wise Separable One-Dimensional Convolutional Neural Network (MDS-1DCNN) model for rank attack detection. In addition to that, the nodes are mitigated while the routing process takes place. To validate the efficiency of the given network model during routing, the Golden Eagle Optimizer (GEO) is utilized to derive the objective function using the constraints such as shortest distance, energy, path loss and delay. Finally, the performance is validated using diverse parameters and compared against classical methodologies. Thus, the results illustrate that the proposed model has effectively detected the RA that facilitates the secured IoT routing against other models.