An experimental study on speed measurement using sensitivity compensated signal and linear prediction processing

Ultrasonic pulse-echo method is widely employed for acoustical measurement in ocean or remote sensing of automobiles and robots. For target ranging, the time-of-flight (TOF), while for speed measurement, the Doppler frequency shift, are usually employed, respectively. However, owing to the resolution of the frequency, lower speed is difficult to be measured. In order to acquire TOF with higher accuracy, pulse compression method using frequency-modulated (FM) signal is usually employed. However, owing to the sensitivities of the ultrasonic transducers, the spectrum of the received signal will be uneven and narrow-banded, that lessens the effectiveness of the pulse compression. In order to acquire the received echo signal with broader and flatter spectrum, we have proposed sensitivity compensated (SC) transmitting signal and linear prediction (LP) processing for expanding effective spectrum. In this paper, the efficiency of these methods for measuring lower speed using TOF of dual-pulses is studied experimentally. As comparison to the traditional linear FM transmitting signal (Chirp wave), two SC signals are discussed. The amplitude modulated SC (AMSC) signal is calculated from the quotient of spectra of the Chirp wave and its reference received signal by inverse filtered processing. Because the amplitude characteristic of the inversed spectrum reflects directly to the time domain, the AMSC signal has a same waveform as an AM Chirp wave. On the other hand, the FMSC signal is calculated by equalizing the energy of the SC spectrum to the duration time of the corresponding frequency part in the time domain, and it becomes a non-linear FM signal with identical amplitude. A matched filtered processing is employed for pulse compression. Furthermore, in order to acquire shorter pulse width, an LP processing expanding the effective spectrum of the pre-compressed pulse is introduced. Therefore, five kinds of measurement result, a) using Chirp wave, b) using AMSC signal, c) us- ng FMSC signal, d) using AMSC signal with LP, and e) using FMSC signal with LP, are compared, in total. The results show that the pulse width of a single compressed pulse derived by d) is shortened to be about 1/5 of that by b), but that derived by e) shows no obvious improvement. While the results of speed measurement of 1.0 m/s ~2.0 m/s, the accuracy of b) ~d) are similar (with error less than 10%) with clear improvement from that of a), but that of e) is the worst. The tendency of these results can be described by the corresponding spectra, that FMSC does not compensate the unevenness of spectrum brought forth by the sensitivities sufficiently.