International Journal of Scientific Research and Engineering Development

This study presents a comprehensive framework for optimizing bandpass filters (BPFs) to meet the demanding requirements of modern communication systems, including 5G, Internet of Things (IoT), and satellite networks. The proposed methodology integrates a hybrid Genetic Algorithm-Machine Learning (GA-ML) approach, advanced substrate materials, and fractal defective ground structures (DGS) to achieve significant performance improvements. By leveraging GA-ML, optimization time was reduced by 75%, and insertion loss was improved from 2.1 dB to 1.8 dB. The use of fractal DGS allowed for a 40% reduction in footprint and 15 dB harmonic suppression, while Rogers RO3003 substrates minimized dielectric loss at millimeter-wave (mmWave) frequencies. Experimental validation confirmed the practicality of these innovations, with IoT BPFs achieving a compact 5 mm² footprint and satellite filters demonstrating robust thermal stability. Despite these advancements, challenges in cost scalability and frequency adaptability remain, particularly for IoT applications and THz designs. This paper underscores the need for cost-effective materials and larger datasets to enhance AI/ML optimization at higher frequencies. Future research directions include the exploration of quantum-inspired algorithms, graphene-based substrates, and the development of standardized benchmarking datasets to drive the evolution of adaptive, spectrum-efficient filters for next-generation 6G and THz communication systems.