Recursive filters for subband decomposition algorithms in ultrasonic detection applications

In this study, we present a low-cost, low-power, and compact hardware platform for real-time ultrasonic detection applications. In order to cope with the computational complexity of the inherent signal processing algorithms, recursive IIR filter architectures are proposed for realizing the frequency diverse transforms; discrete Fourier transform (DFT) and discrete cosine transform (DCT), which provide robust detection performance. Both Goertzel (for DFT implementation) and folding kernel for DCT offer significant hardware reductions compared to standard transform implementations. In order to sustain the high throughput required for real-time signal processing, a sparse transform method is proposed where only the required transform coefficients are computed. FPGA implementation results confirm savings in hardware usage and viable performance for a portable ultrasonic imaging device.