This paper considers the application of signal processing methods to passive indoor positioning with acoustics microphones. The key aspect of this problem is time-delay estimation (TDE) that is used to get the time difference of arrival of the source’s signal between the pair of distributed microphones. This paper studies the approach based on generalized phase spectrum (GPS) TDE methods. These methods use frequency-domain information about the received signals that make them different from widely applied generalized cross-correlation (GCC) methods. Despite the more challenging implementation, GPS TDE methods can be less demanding on computational resources and memory than conventional GCC ones. We propose an algorithmic implementation of a GPS estimator and study the various frequency weighting options in applications to TDE in a small room acoustic environment. The study shows that the GPS method is a reliable option for small acoustically dead rooms and could be effectively applied in presence of moderate in-band noises. However, GPS estimators are far less efficient in less acoustically dead environments, where other TDE options should be considered. The distinguishing feature of the proposed solution is the ability to get the time delay using a limited number of the adjusted bins. The solution could be useful for passively locating moving emitters of narrow-band continual noises using computationally simple frequency detection algorithms.
1 August 2020 in Journal of Physics: Conference Series
The paper discusses the implementation of correlation algorithm for time delay estimation on a Raspberry Pi single-board computer. The implemented correlation algorithm is based on Fourier transform. In the course of the study, we applied two alternative solutions for the software implementation of discrete Fourier transform. The first solution stands on FFTW library and uses general-purpose quad-core ARM Cortex A53 processing unit. The alternative method uses VideoCore IV graphic processing unit and is implemented via firmware GPU_FFT library. We have performed a computational experiment on a Raspberry Pi 3B to determine which solution is more preferable for the implementation of correlator. After a comparative study we figured out that estimated processing time is highly dependent on computations parameters and input signals. For small FFT window sizes CPU is proved to be a preferable option. However, for large FFT windows GPU allows significantly accelerating the computations. At some point, you can achieve even better performance by using batching and GPU for direct FFT and CPU for inverse FFT. According with the results, we have concluded that both alternatives have their own potential advantages and particular drawback. We also establish, that Raspberry Pi 3 B computer with HiFiberry extension can be used as a real-time correlator for audio signals.
1 December 2021 in Journal of Physics: Conference Series
The article discusses the application of wavelet analysis for the time-frequency time-delay estimation. The proposed algorithm is wavelet transform-based cross-correlation time delay estimation that applies discrete time wavelet transform to filter the input signal prior to computation of cross-correlation function. The distinguishing feature of the algorithm that it uses the variation of continuous wavelet transform to process the discrete signals instead of dyadic wavelet transform that is normally applied to the case. Another feature that the implication of convolution theorem is used to compute coefficients of the wavelet transform. This makes possible to omit redundant discrete Fourier transforms and significantly reduce the computational complexity. The principal applicability of the proposed method is shown in the course of a computational experiments with artificial and real-world signal. So the method demonstrated expected selectivity for the signals localized in the different frequency bands. The application of the method to practical case of pipeline leak detection was also successful. However, the study concluded that this method provides no specific advantages in comparison with the conventional one. In the future, alternative applications in biological signal processing will be considered.
The paper considers use of parallel processing on multi-core central processing unit for optimization of the coherence function evaluation arising in digital signal processing. Coherence function along with other methods of spectral analysis is commonly used for vibration diagnosis of rotating machinery and its particular nodes. An algorithm is given for the function evaluation for signals represented with digital samples. The algorithm is analyzed for its software implementation and computational problems. Optimization measures are described, including algorithmic, architecture and compiler optimization, their results are assessed for multi-core processors from different manufacturers. Thus, speeding-up of the parallel execution with respect to sequential execution was studied and results are presented for Intel Core i7-4720HQ и AMD FX-9590 processors. The results show comparatively high efficiency of the optimization measures taken. In particular, acceleration indicators and average CPU utilization have been significantly improved, showing high degree of parallelism of the constructed calculating functions. The developed software underwent state registration and will be used as a part of a software and hardware solution for rotating machinery fault diagnosis and pipeline leak location with acoustic correlation method.
28 August 2014 in IOP Conference Series: Earth and Environmental Science