Internet of Things (IoT) is a novel networking communication field in which the nodes act as physical objects. These nodes might be stationary or mobile. The majority of networking communication problems occur due to nodes’ mobility. Due to the mobility of IoT networks, excessive control overhead messages are generated. These messages are generated due to the frequent exchange of some control information, such as device identity, location information, device mobility, device direction, and others. In general, network clustering helps to alleviate the problem of generating high control overhead messages. Therefore, many proposed or enhanced Cluster-Based Routing (CBR) protocols address the shortcoming of traditional CBR protocols. In CBR protocols, each cluster selects only one node based on some parameters, and this node is called a LEaDeR (LEDR). The LEDR acts as a coordinator for its associated cluster. Furthermore, the rest of the nodes located in the cluster are called Smart DEVices (SDEV). A problem occurs when the SDEVs and the LEDR exchange their control overhead messages with each other. The exchanging of these messages utilizes the available network resources. This article proposes a Clustered IoT (CIoT) routing protocol, the CIoT routing protocol applies in a grid clustered topology. The proposed CIoT protocol aims to efficiently propagate messages between any pair of devices. The CIoT protocol selects only one route among a set of candidate routes, and the selected route is verified and evaluated based on some constraints. Also, this article proposes a Predictive Generated Hello Messages Algorithm (PGHMA). The PGHMA aims to minimize the generated control overhead messages produced by SDEVs and LEDRs. In comparison to other CBR protocols, simulation results demonstrate the effectiveness of the CIoT algorithms by increasing network throughput by 78% and decreasing end-to-end latency by 12.5%. Also, the PGHMA algorithm significantly outperforms other algorithms in terms of reducing the number of generated HELLO messages by 5.9%. Finally, we implement and evaluate our proposed algorithms by comparing them to other related algorithms published in the literature.