The exponential growth of the Internet of Things (IoT) has led to an unprecedented number of devices generating and consuming data across diverse sectors, including smart cities, healthcare, and industrial automation. While traditional cloud-based architectures have been instrumental in managing early IoT deployments, they fall short in supporting the ultra-low latency, high scalability, and massive device density required for future IoT ecosystems. Fifth-generation (5G) mobile networks introduce promising capabilities such as Ultra-Reliable Low-Latency Communication (URLLC), Massive Machine-Type Communications (mMTC), and network slicing. However, leveraging these features for practical, real-time IoT deployments requires novel protocol designs that go beyond conventional centralized paradigms. This paper introduces DKEP-5G, a Dynamic Node-based Edge Protocol that enhances data transmission and processing efficiency by integrating predictive AI models, SDN-based traffic management, and federated edge orchestration. The proposed architecture is evaluated through rigorous simulations using NS-3 and OMNeT++, demonstrating improvements of up to 85.3% in latency, 66.3% in energy efficiency, and a 700% increase in supported devices per square kilometer when compared to traditional cloud-based MQTT systems. We also provide statistical and analytical validation of the system, demonstrating DKEP-5G’s readiness for large-scale, delay-sensitive IoT applications.