Multiband antenna design method using predefined shapes and image segmentation techniques for IoT applications

The development of the internet of things necessitates more compact electronic products. This paper proposes a novel approach: designing a predefined shape to function as a antenna without changing its visual appearance. By comparing with the defined threshold, an example of university emblem image is transformed into a binarized image, then define the conductive metal and dielectric substrates. The metallic part are pixelated, ports are added between adjacent pixels, and the initial structure’s port impedance matrix is derived through full electromagnetic simulation, forming a multi-port network model. The target three bands are: WiFi 2.4 GHz (2.4–2.5 GHz), LTE FDD Band 3400 (3.4–3.5 GHz), and FR1 n79 band (4.7–4.9 GHz). Leveraging the port impedance matrix, we employ a genetic algorithm to optimize the pixel connectivity, meeting the defined design specifications. The resulting patch antenna, retaining the appearance of the university emblem, achieves 12.9%/22.8%/7.0% bandwidths, 3.6/4.6/3.0 dBi gains, and > 75% efficiency, thereby validating the proposed method.