Optimization of dynamic scattering arrays

This thesis addresses the optimisation of Dynamic Scattering Arrays (DSAs) for 6G wireless systems. DSAs move signal processing from the digital baseband into the electromagnetic domain, reducing power consumption and hardware cost. The main objectives are to: (1) investigate beamforming toward a fixed far-field target and maximise directivity in the steering direction; and (2) quantify robustness under manufacturing tolerances via noise injection. We replace classical quasi-Newton solvers with the Adam optimiser, which handles the bounded, non-convex optimisation landscape imposed by realistic varactor models for the antenna configurable loads. The work is organised into three use cases: single-target superdirective beamforming, multi-user MISO zero-forcing, and MIMO over-the-air SVD precoding. The simulation engine supports all three scenarios and includes robust optimisation with noise injection for the single-target case.