Sonic Fourier Transform Imaging Using GHz Ultrasonic Transducer Array

The Fourier transform is one of the most widely used mathematical tools for numerical solutions and understanding signals and images. In the digital domain, the 2D fast Fourier transform (FFT), which has a computational complexity of O(N2logN2), is often used. The compute time can be a bottleneck for many real-time edge applications that require Fourier computation at high frame rates. A solution is to use an ultrasonic hardware accelerator that can compute Fourier transform and convolution at a drastically higher speed and lower power, achieving O(N) complexity [1] – [5]. Presented in this work is the first experimental demonstration of the Sonic Fourier Transform (SonicFT) measurement using a piezoelectric MEMS framework. AlN transducers were used as transmitters to simulate different images being input into an ultrasonic hardware accelerator. The transmitted ultrasonic wavefronts were imaged electronically, using a GHz ultrasonic imager operating at 1.85 GHz, consisting of a 130nm CMOS-integrated 128 by 128 array of 50x50μm AlN transducers. The images obtained are consistent with the expected Fourier transforms, demonstrating an ultracompact Fourier transform capability.