Enhancement of collaborative communication using optimized dynamic density graph convolutional networks in IoT-enabled intelligent surfaces

Abstract

Reconfigurable intelligent surface (RIS) is defined as a software-controlled passive devices acts as a relay system, reflecting receiving signals from the source to destination in a cooperative mode with optimal signal strength. An IoT-dependent network's configurable and flexible RIS allows for stand alone or cooperative arrangements with significant advantages over traditional networks. Despite its potential, efficient implementation of RISs in Internet-of-Things (IoT)-based networks remains a difficulty because to the complexity associated with phase shift optimisation and symbol identification. In this manuscript, Enhancement of Collaborative Communication using Optimized Dynamic Density Graph Convolutional Networks in IoT-Enabled Intelligent Surfaces (ECC-DDGNN-IoT IS) is proposed. It focuses on optimising the RIS phase shifts through Dynamic Density Graph Convolutional Networks (DDGNN) approaches. This optimisation increases the signal quality and overall system performance in cooperative communication circumstances. This model addresses the complexity of Maximum Livelihood (ML) detection at destination. A DDGNN-based symbol identification method is introduced, along with DDGNN-assisted phase optimisation of the RIS. This technique decreases computational load on the receiver while retaining good detection accuracy. Therefore, Artificial Protozoa Optimizer (APO) is proposed to optimize the weight parameter of DDGNN model to accurately shift the phase of RIS. This model is implemented in MATLAB platform. The proposed ECC-DDGNN-IoT-IS method attains high accuracy, low RMSE and computational complexity compared to the existing techniques, such as deep-learning enabled IoT depend RIS for cooperative communications (DNN-IoT-RISCC), Deep Learning dependent Detection on Reconfigurable intelligent surface Assisted RSM and RSSK (BDNN-RIS-RSSK), and semi-federated learning in massive IoT networks for collaborative intelligence (SFL-CI-MIoT) respectively.