Comparative Study and Optimization of Error Diffusion Matrices for High-Fidelity OAM Beam Generation with a Digital Micromirror Device

Orbital-angular-momentum (OAM) beams generate multiple independent orthogonal modes. This property offers a potential increase in the capacity of optical communication systems. This paper systematically evaluates the reconstruction performance of nine error-diffusion matrices for binary holographic encoding of Laguerre-Gaussian (LG) and Bessel modes on a digital micromirror device (DMD). Among the tested matrices, the Sierra diffusion matrix achieved the best balance between high fidelity and computational efficiency. The fidelity obtained by reconstructing high-order LG modes and Bessel modes using Sierra matrix is between 0.96 and 0.99. We further applied a genetic algorithm (GA) to iteratively optimize the weighting coefficients of the Sierra matrix. This optimization led to significant improvements in wavefront reconstruction accuracy. Simulations show an average PSNR improvement of 0.84 dB and a marked reduction in ring artifacts. This work lays a practical foundation for high-precision, high-speed multiplexed OAM beam generation. It may expand the capacity and efficiency of next-generation optical communication networks. • Comparison of nine error-diffusion matrices for digital micromirror device based OAM beam encoding. • Selected Sierra matrix, achieving fidelity over 0.99 for diverse Laguerre–Gaussian and Bessel modes. • Genetic algorithm optimization of Sierra weights boosts average PSNR by 0.84 dB. • Genetic algorithm based optimization reduces artifacts in intensity and phase reconstruction.