Framework for Intelligent Radio Grid For Legacy radio sets Using CNN Based Signal Classifier And Software Defined Radios

The escalating demands of modern communication necessitate intelligent radio grids for both military and civilian applications. However, ongoing and legacy radio sets lack compatibility due to parameter disparities, which demand extensive hardware replacements for migration to Mobile Ad-Hoc Networks(MANET) and cognitive domains. This paper presents a comprehensive framework for implementation of an intelligent radio grid, aimed at enhancing the efficiency and reliability of wireless communication systems while using contemporary and legacy radio sets. Our framework relies on a hybrid model which uses Convolutional Neural Network(CNN) -based model for RF classification and YoLO for image classification. Further the model is integrated with Software Defined Radio(SDR) to dynamically adjust radio parameters to enable interoperability and establishment of MANET in real-time. The paper proposes solution for interoperability among radios of different make and models, paving the way for MANET implementation and establishment of a smart radio grid using legacy equipment. The implementation encompasses critical aspects including spectrum sensing, signal classification, automatic parameter adaptation, countering spoofing attacks and fingerprint based authentication. Experimental results showcase 0.92 accuracy of our model to classify radios and fully interoperability between different band, make and model radio sets, which helps in improving operational efficiency, reducing costs, and enhancing overall system capacity.