Reconfigurable MIMO antenna for Sub-6 GHz 5G cognitive radio with machine learning based metaheuristic optimization

Abstract

Abstract The growth of 5G and cognitive radio (CR) has increased the need for frequency-reconfigurable multiple-input multiple-output (MIMO) antennas that can access spectrum dynamically while controlling interference. This paper presents a wideband reconfigurable MIMO antenna for Sub-6 GHz 5G CR applications. The proposed antenna supports interweave and underlay operating modes, which improves spectrum utilization, coverage adaptability, and communication flexibility. Reconfigurability is achieved using varactor diodes for continuous frequency tuning and PIN diodes for band switching. In wideband mode, the antenna operates from 2.59 to 5.09 GHz, with a simulated reflection coefficient close to -35 dB and a peak gain of about 5.6 dBi. The measured reflection coefficient reaches approximately -40 dB and shows close agreement with the simulated response, confirming the practical feasibility of the design. A machine-learning-assisted optimization framework using particle swarm optimization (PSO), Newton Raphson optimization (NRO), and Walrus Optimization (WO) is used to improve bandwidth, gain, and impedance matching. Electromagnetic simulations and measurements confirm strong $$S_{11}$$ , gain, isolation, and radiation performance. These results show that the proposed design is suitable for next-generation adaptive wireless systems and provides a scalable framework for reconfigurable MIMO antenna design in 5G and cognitive radio environments.