Efficient algorithm and hardware/software co-design for chirp echo estimation in ultrasonic NDE applications

In this study, a computationally efficient method for echo detection and pulse compression based on ellipse fitting on the time-frequency representation of the signal is investigated. The algorithm successively estimates the chirplet parameters to decompose convoluted signals into a linear combination of chirp components. Analytical results show that the algorithm is efficient and successful in signal representation. For the real-time application, an embedded implementation of the ellipse fitting is implemented on Xilinx Vitex-5 FPGA platform. Based on the system constraints and the efficiency of estimations, the performances of different algorithm implementation schemes have been examined. The developed system-on-chip hardware/software co-design successfully demonstrates robustness in the chirplet decomposition of experimental signals.