Programmable Logic Functions-Integrated Acoustic In-Sensor Computing

In-sensor computing supports edge computing by reducing data transmission, but logic and arithmetic operations are still underdeveloped in acoustic sensors. Mechanical computing with metamaterials integrates these functions directly into sensors by responding to external stimuli, offering a promising solution. However, current mechanical logic switching depends on structural transformations, limiting logic function density. Therefore, a reprogrammable logic method is proposed using a geometrically imbalanced graded phononic crystal (GiGPnC). By designing graded unit cells, the structure produces two types of asymmetric scattering effects on antisymmetric Lamb waves, and creating constructive and destructive interference at the point defect. These four acoustic frequency responses correspond to all input-output mappings of a two-input one-output system, enabling mechanical computing. Then, reprogrammable realizations of seven basic logic gates and combinational logic are experimentally demonstrated, including a 1-bit half-subtractor and a 4-bit even parity generator, on a single GiGPnC. This frequency-response-based reprogrammable method can be extended to more complex logic functions. This reprogrammable design paradigm is expected for acoustic in-sensor computing centered on mechanical computing can promote the development of edge computing and Internet of Things (IoT).