With the unprecedented technological advances witnessed in the last two decades, more devices are connected to the Internet, forming what is called the Internet of Things (IoT). The IoT devices with heterogeneous characteristics and the quality of experience (QoE) requirements may engage in the dynamic spectrum market due to the scarcity of radio resources. We propose a framework to efficiently quantify and supply radio resources to the IoT devices by developing intelligent systems. The primary goal of this paper is to study the characteristics of the next generation of cellular networks with non-orthogonal multiple access (NOMA) to enable connectivity to clustered IoT devices. First, we demonstrate how the distribution and QoE requirements of IoT devices impact the required number of radio resources in real time. Second, we prove that using an extended auction algorithm by implementing a series of complementary functions enhance the radio resource utilization efficiency. The results show a substantial reduction in the number of sub-carriers required when compared with conventional OMA and the intelligent clustering is scalable and adaptable to the cellular environment. Ability to move spectrum usages from one cluster to other clusters after borrowing when a cluster has fewer users or move out of the boundary is another soft feature that contributes to the reported radio resource utilization efficiency. Moreover, the proposed framework provides IoT service providers cost estimation to control their spectrum acquisition to achieve the required quality of service with a guaranteed bit rate (GBR) and non-GBR.