Design and Optimization of a Configurable Pixelated Antenna Using a Genetic Algorithm

Modern antenna design standards are evolving, and there has been a significant interest in algorithm-driven antenna design. Predefined, conventional structures are often implemented and optimized through optimization techniques. However, recent trends show that optimized pixelated antennas can outperform classical designs due to greater design flexibility. This study proposes a novel approach towards designing a pixelated antenna with no conventional shape, which could be configured to resonate at any frequency using a genetic algorithm. To achieve this, a radiating structure composed of an 8x8 grid of metallic pixels was generated, with each pixel being independently controlled by a binary parameter. This structure was then mirrored across both the X and Y axes for symmetry. As this was a binary optimization task, a genetic algorithm was used to find the optimal pixel configuration that achieves resonance at a desired frequency while minimizing return loss. The antennas were simulated in CST Microwave Studio while the Genetic Algorithm was run on MATLAB to generate the antenna parameters. The generated structures show excellent performance with return loss (S11) values of −18.05dB at 2.45GHz, −28.93dB at 3GHz, −16.49dB. at 4GHz, −28.78dB at 5GHz, and −22.34dB at 6GHz, respectively. A further fine-tuning process was also explored. The process was shown to be able to efficiently generate new antenna structures as per user demand.