Broadband-Based IoT Communication Solutions for Agribusiness and High-Speed Industrial Internet of Things (IIoT): Advanced Machine Learning Approach

The Internet of Things (IoT) has transformed various sectors, including agriculture and industrial automation, by enhancing efficiency and productivity, including data-informed decision-making. Innovative communication solutions play a crucial role in the success of IoT applications in rural and remote areas, where connectivity/broadband is limited or unavailable. Agriculture and industrial IoT (IIoT) in this region will not thrive because of poor network coverage or unreliable broadband connectivity. This paper introduces a comprehensive strategy for implementing broadband-based IoT (B-IoT) communication solutions for agribusiness and IIoT by leveraging an advanced machine learning (AML) framework known as the parameterized analytical modeling-based deep neural network-enhanced parametric rectified linear unit (PDNN-ePReLU) model. This approach incorporates a novel parameterized-based DNN algorithm and a new activation function $F_{A}$ to enhance Cognitive Radio (CR) on TV White Spaces (TVWS) for backhauling and last-mile connectivity. This includes the enhancement of the fifth-generation (5 G) reduced-capability (RedCap) devices on sub- 7 GHz unlicensed or private frequency bands for IoT access networks. The proposed solution consists of an input layer comprising specific network parameters, hidden layers, and an output layer focusing on coverage, throughput, latency, and energy efficiency (EE), which the PDNN-ePReLU model aims to optimize. This model employs CR to utilize unused TVWS frequencies dynamically and improves the network performance by predicting optimal transmission and adaptively allocating resources. This paper outlines a detailed implementation framework showcasing the PDNN-ePReLU model’s training, validation, and test phases while presenting the potential benefits for agribusiness and IIoT in underserved areas. MATLAB simulation results indicate that the optimized CR significantly enhances the coverage, throughput, latency, and EE over long distances, outperforming legacy CR and traditional networks (Wi-Fi, LTE, etc.). Furthermore, the optimized 5G RedCap surpasses its legacy (non-optimized) counterparts, offering a robust and scalable solution for B-IoT in agribusiness and IIoT applications.