A new quad-band polarization-insensitive metasurface for RF energy harvesting and its design optimization

In avoiding the inconvenient and costly maintenance and replacement of the conventional batteries in large-scale and low-power wireless sensor applications, radio frequency (RF) energy harvesting shows exclusive potential due to its abundant availability and resilience in environmental conditions. This study introduces firstly a novel metasurface (MS) absorber topology employing four sequentially rotated spoof-local-surface-plasmons (SLSPs) resonators to achieve a multi-band energy harvesting and a polarization insensitivity. To realize an efficient and accurate design optimization of the proposed MS absorber, a customized multi-stage collaborative machine learning-assisted optimization methodology, incorporating three different fidelity levels, the nonlinear autoregressive multi-fidelity Gaussian Process (NARGP), and the multi-task Gaussian Process (MTGP), is then proposed. A prototype MS absorber is finally optimized and fabricated. Both the numerical and the experimented results validate the presented works.