Improved Neural Network–Based Joint Spectrum Sensing and Allocation for CR-IoT

Abstract

The rapid expansion of the Internet of Things (IoT) and the increasing demand for wireless communication have intensified the need for efficient spectrum management in cognitive radio networks (CRNs). Traditional approaches to spectrum sensing and allocation often operate in isolation or rely on static methods, which fail to address the dynamic and evolving nature of modern wireless environments. As IoT devices proliferate and spectrum resources become increasingly congested, there is a pressing need for more adaptive and efficient spectrum management solutions. Our approach addresses this need by offering an adaptive framework that responds to the changing spectrum landscape, thereby optimizing spectrum usage and reducing interference. This research suggests improved NN joint spectrum sensing for CR-IoTNet (INJSS-CR). This approach leverages cognitive radio (CR) technology to enhance spectrum utilization and mitigate the impact of spectrum shortages. CR technology enables secondary users (SUs) to detect and access unused spectrum through spectrum sensing. Within the CR-IoTNet framework, joint spectrum sensing and allocation are performed to serve SU-IoT devices via an interference-free channel (IFC). The system comprises multiple primary user base stations (PU-BSs) and SU devices functioning as IoT smart objects. Additionally, we integrate an improved neural network (INN) to adapt to dynamic network conditions and monitor primary user (PU) spectrum utilization using a comprehensive multiclass (J × 8) − D feature set. This combination of advanced techniques and CR technology aims to optimize spectrum management and support the growing IoT ecosystem. In particular, the INJSS-CR obtained the greatest accuracy of 0.9492 at a training rate of 80%.