| Authors | M. Jeyabharathi1, D. Kumutha2, C. Priya3, R. Dhivya4, K. Tamilarasi5, G. Vijayakumari6 |
| Affiliations |
1Department of Electronics and Communication Engineering, Best Center,Sastra University, India 2Department of Electronics and Communication Engineering, AMC Engineering College, Bangalore, India 3Department of ECE, Karpagam College of Engineering, Coimbatore, TN, India 4Department of ECE, Adhiparasakthi College of Engineering, Kalavai, Tamilnadu, India 5Department of IT, Panimalar Engineering College, Chennai, India 6Department of ECE, New Prince Shri Bhavani College of Engineering and Technology, Chennai, India |
| Е-mail | skvijaykumu@gmail.com |
| Issue | Volume 18, Year 2026, Number 4 |
| Dates | Received 20 March 2026; revised manuscript received 15 August 2026; published online 21 August 2026 |
| Citation | M. Jeyabharathi, D. Kumutha, C. Priya, et al., J. Nano- Electron. Phys. 18 No 4, 04017 (2026) |
| DOI | https://doi.org/10.21272/jnep.18(4).04017 |
| PACS Number(s) | 84.40.Ba |
| Keywords | 5G n78 band, Sub-6-GHz MIMO antenna, Bayesian optimization, Artificial neural networks, AI-assisted antenna design, Mutual coupling reduction, Polarization diversity, FR-4 substrate, Wideband radiator, BO–ANN optimization. |
| Annotation |
Implementing an AI-assisted design framework that combines Bayesian Optimisation (BO) with Artificial Neural Network (ANN)-based surrogate refinement, a compact four-element MIMO antenna for sub-6-GHz 5G n78 terminals is presented. In order to improve polarization diversity and reduce mutual coupling, the design uses an AI-evolved wideband SISO radiator as the unit cell, which is replicated and arranged in an orthogonal configuration. Performance gains are solely attained through AI-tuned geometry evolution, including optimized inter-element spacing, feedline angles, and central junction shaping, as opposed to depending on DGS, EBG, or parasitic structures. In the 3.0-4.2 GHz band, the resulting 60 75 mm2 single-layer FR-4 prototype achieves |Sij| isolation better than 20 dB, an Envelope Correlation Coefficient (ECC) below 0.008, diversity gain near 10 dB, and channel capacity loss below 0.15 bits/s/Hz. Additionally, the antenna shows array gains of 13-15 dBi and stable radiation patterns. These findings show that highly effective, small, and fabrication-friendly MIMO architectures for next-generation sub-6-GHz 5G devices are made possible by the BO-ANN optimization pipeline. |
|
List of References |