Adaptive beamforming using neural networks for conformal antenna array uses

In communication applications, the radiation pattern of conformal antenna arrays is a critical design consideration. By electronically steering the main radiating beam in both azimuth and elevation, a hemispherical coverage can be achieved, enabling a pencil-shaped radiation pattern. This paper presents a neural network-based synthesis approach for adaptive beamforming in conical and cylindrical conformal antenna arrays, including three-dimensional array synthesis examples. The array factor (AF) is derived by adjusting the antenna array’s dimensions using generalized analytical techniques, and key performance metrics such as sidelobe level (SLL), half-power beamwidth (HPBW), and directivity are evaluated. Mutual coupling effects are neglected, assuming isotropic radiators. The proposed array demonstrates significant improvements over existing conical, coaxial cylindrical, and cylindrical configurations in terms of size, 3D scanning capability, HPBW, SLL, and directivity. Using MATLAB simulations, we apply neural network-based beamforming to conformal arrays, focusing on cylindrical, coaxial cylindrical, and conical geometries. The neural network effectively learns the radiation characteristics of these arrays and demonstrates strong performance. Building on prior work, we examine three new scenarios: A uniform circular array (UCA) mounted on an empty coated cylinder, a concentric circular array (CCA) on a multilayered coated cylinder and a conical array formed by stacking uniform circular arrays with linearly decreasing diameters along the z-axis. The proposed method is validated through numerical examples and can be extended to other 3D array geometries, such as spherical configurations or aperiodic conformal phased arrays.