Frequency Difference Matched Field Processing Research Articles

High-frequency source localization based on steered frequency-wavenumber analysis using sparse array

In this study, we address the problem of estimating the position of a nearby target emitting a high-frequency signal using sparse array. One of the representative localization techniques, Matched Field Processing (MFP), is known to demonstrate reliable performance below 1 kHz. However, at frequencies higher than the design frequency of the array sensor, the localization becomes challenging due to spatial aliasing. Nevertheless, Frequency-Difference Matched Field Processing (FDMFP) is a technique that successfully estimates the position of the target even in the presence of spatial aliasing. It performs signal processing in the low-frequency band using low-frequency components obtained from two high-frequency components. To ensure reliability, we compared and validated the results of this study with the results of FDMFP. In this research, we completed the algorithm by controlling the phenomenon of striation shifting caused by spatial aliasing in the conventional frequency-wavenumber(f-k) analysis through beam steering. We steered the received signal to the location of signal reception and identified the target's position using the f-k spectrum of the steered signal. The data used in the study is the sound of snapping shrimp from SAVEX15, with a frequency range of 5–24 kHz wideband impulse signals.

Reformulation of frequency-difference matched-field processor for high-frequency known-source localization.

Frequency-difference matched-field processing is a high-frequency source localization technique formulated by matching the frequency-difference autoproduct of the measured field and replicas at the difference-frequency. Although it successfully localizes sound sources by sparse vertical array in shallow or deep ocean with an environmental mismatch, there is still some ambiguity in replica modeling and signal processing. Here, the existing conventional processor is modified to match the bandwidth-averaged autoproduct of the measured field with replicas of the bandwidth-averaged autoproduct, or approximately its self-term for the expected source locations. The proposed processor is consistent with the perspective of matched-field processing and can naturally relieve some drawbacks of the existing one, such as low peak or low dynamic range on the ambiguity surface. Numerical tests are carried out in several shallow ocean environments and the source localization using experimental data are performed to confirm the properties of the proposed processor. It is found that the high-frequency diffracted field always leaves traces on its bandwidth-averaged autoproduct field. These high-frequency marks cause a bias in source localization in the presence of a sound speed mismatch even in low difference-frequencies.

Source Range Estimation with Horizontal Arrays Using Frequency Difference Matched Field Processing in Shallow Water

Due to the modal interference, the ambiguity function of conventional matched field processing (CMFP) exists periodic solutions. To solve this problem, a frequency difference matched field processing (FDMFP) method with horizontal linear array (HLA) for estimating the source range is proposed. The method calculates the ambiguity function of two-neighbor-frequency between the measurement and the replica of each hydrophone, and then computes the high-order cross-spectrum. The range corresponding to the maximum of the ambiguity function is the optimal estimation. Because of the larger modal interference distances, the FDMFP can efficiently provide one-only optimal range instead of some of CMFP.

Incomplete acoustic time reversal, frequency shifting, and matched field processing

In a reciprocal time-invariant medium, sound can travel in both directions along acoustic paths. This bi-directionality enables acoustic retrofocusing via time reversal (TR) and acoustic source localization via matched field processing (MFP). Given this common basis, the simplest frequency-domain formulations of acoustic TR and MFP are essentially identical; both involve a quadratic product of forward- and backward-propagating complex fields. For this presentation, all scenarios begin with the acoustic field from a remote source that propagates forward through a potentially-complicated acoustic environment to a transducer array where it is recorded. The difference between TR and MFP then arises because the second backward-propagating (time-reversed) acoustic field may be (i) a genuine field in the actual environment as in acoustic TR, or (ii) a predicted field developed from a model of the environment as in MFP. This presentation describes a third possibility, frequency-difference MFP or incomplete acoustic TR, that combines (i), (ii), and a multiplicative property of harmonic plane waves to produce frequency-downshifting. The resulting unconventional formulation involves a cubic product of complex fields and allows robust reduced-resolution source localization with sparse arrays in complicated uncertain environments. Example results from simulated, laboratory, and ocean propagation data are shown. [Sponsored by ONR, NAVSEA, NSF].

Deep learning-based high-frequency source depth estimation using a single sensor.

The sensitivity of underwater propagation models to acoustic and environmental variability increases with the signal frequency; therefore, realizing accurate acoustic propagation predictions is difficult. Owing to this mismatch between the model and actual scenarios, achieving high-frequency source localization using model-based methods is generally difficult. To address this issue, we propose a deep learning approach trained on real data. In this study, we focused on depth estimation. Several 18-layer residual neural networks were trained on a normalized log-scaled spectrogram that was measured using a single hydrophone. The algorithm was evaluated using measured data transmitted from the linear frequency modulation chirp probe (11-31 kHz) in the shallow-water acoustic variability experiment 2015. The signal was received through two vertical line arrays (VLAs). The proposed method was applied to all 16 sensors of the VLA to determine the estimation performance with respect to the receiver depth. Furthermore, frequency-difference matched field processing was applied to the experimental data for comparison. The results indicate that ResNet can determine complicated features of high-frequency signals and predict depths, regardless of the receiver depth, while exhibiting robust environmental and positional variability.

High frequency source depth localization using deep neural network

Localization of high-frequency sources in underwater is a difficult problem because the propagation model is increasingly sensitive to environmental parameters as the frequency increases. A deep learning approach is proposed to estimate the depth of a high-frequency underwater source using a single hydrophone trained on real data. A residual neural network is trained by a spectrogram of measured signal and estimates the depth of the source as a regression problem. The method is applied to data collected during the shallow water acoustic variability experiment 2015 in the northeastern East China Sea and compared with the results of the frequency difference matched field processing which uses 16 sensors. It was also validated that the trained model can display the generalized results for the signals measured at other times or at similar settings but different source-receiver location. As a result, the neural network is able to more accurately estimate the depth of high-frequency source and shows the features found by the network from real data are effective for localization, while the model often fails to generate accurate replicas.

Explorations of in-situ source localization in the deep ocean using frequency difference matched field processing and matched autoproduct processing

Matched field processing (MFP) is a well-known technique for source localization in challenging environments. MFP involves correlating array-recorded fields with calculated replica fields developed from knowledge of the acoustic environment. Incomplete knowledge of the environment causes mismatch between measured and calculated fields, and this mismatch can cause MFP to fail at relevant ranges and frequencies in the deep ocean. Given that the severity of this mismatch increases with frequency, a proposed remedy to the mismatch problem is to analyze the frequency-difference autoproduct of the recorded field, rather than the field itself. The phase of the frequency-difference autoproduct is expected to mimic that of an out-of-band field at a selectable below-band frequency, thereby mitigating the severity of the mismatch at in-band frequencies. This autoproduct is then matched either to a replica field at the below-band frequency, or to an appropriate calculated autoproduct field. This presentation explores these methods, and higher-order corrections, to localize moored sources using data from the North Pacific Acoustic Library. Localization statistics are presented based on the closest moored source, 130 km from the receiving array with a signal bandwidth from 200 to 300 Hz. [Sponsored by ONR.]

Explorations of in-situ passive source localization in the Philippine Sea using frequency-difference matched field processing

Matched Field Processing (MFP) is a well-known technique for source localization in challenging acoustic environments. Accurate source localization using this method requires a correlation between array-recorded and replica acoustic fields calculated using knowledge of the acoustic environment. For a well-understood acoustic environment, this method is successful. However, recorded signals may be sensitive to details of the environment which are unknown and not included in the calculation of the replica field. The severity of this mismatch increases with frequency, as phase coherence is lost more easily at higher frequencies. In this case, a mismatch between the recorded and calculated fields will occur which can cause MFP to fail at relevant frequencies and ranges in the deep ocean. A proposed remedy to this problem may be analyzing the frequency-difference autoproduct of the measured acoustic field instead of analyzing the field itself. This presentation explores using this method to localize moored sou...

Cross-term analysis in frequency-difference-based source localization methods

In previous work, it has been shown that a quadratic product of frequency-domain acoustic fields at different frequencies but the same spatial location leads to an auxiliary field which may contain field information at frequencies below the original signal’s bandwidth (Worthmann, Song and Dowling, 2015, JASA 138, 3549-3562). This quadratic product, termed the frequency-difference autoproduct, has been shown to be valuable for beamforming and source localization in the presence of environmental mismatch and/or array sparseness. However, in a multipath environment, this quadratic product leads to undesired cross-terms. Bandwidth averaging procedures have been found to suppress some of their detrimental influences in some cases, but not all. Additionally, the poor dynamic range observed in frequency-difference beamforming and frequency-difference matched field processing are associated with the imperfect mitigation of these cross-terms. In this presentation, the nature of these cross-terms is analyzed, and signal processing tools are developed which attempt to robustly mitigate the detrimental effects of these cross terms. These signal processing tools can be used to potentially improve localization performance when using frequency-difference autoproduct-based source localization schemes. [Sponsored by ONR and NSF]

Simulation of passive source localization in near-Arctic conditions using frequency-difference matched field processing

Matched Field Processing (MFP) is a well-known technique for passive source localization in complex acoustic environments. It involves correlating array-recorded acoustic fields with replica fields calculated using a model of the acoustic environment, and works well is known environments. However, acoustic signals often depend on unknown environmental details which are not included in the replica-field calculations. The severity of this mismatch increases with frequency and source-array range, and it can cause MFP to fail in the signal band at relevant ranges in the arctic ocean. A proposed remedy to this problem is to utilize the frequency-difference autoproduct of the measured acoustic field instead of the acoustic field itself, and to perform the replica calculations at the difference frequency. The simulated performance of frequency difference MFP is shown for a generic near-Arctic environment at signal frequencies of 200 to 300 Hz, and ranges of 30 to 300 km. Here, mismatch between in the acoustic and replica field is modeled by imposing a range-dependent altimetry and surface sound speed profile to simulate surface ice, and by random time delays applied to each ray-path between the source and each receiver to simulate refractive index fluctuations in the ocean. [Sponsored by ONR.]

The effects of diffraction on the frequency difference and frequency sum autoproducts

Previously, a remote sensing technique termed frequency difference matched field processing was developed for source localization in the shallow ocean (Worthmann et al., 2015, 138, 3549-3562). In this technique, field measurements at the in-band frequency are shifted down (or up) in frequency through the use of the bandwidth-averaged frequency-difference (or frequency-sum) autoproduct, a nonlinear construction made from field amplitudes and averaged over the available signal bandwidth. These bandwidth-averaged autoproducts may have phase structure similar to genuine acoustic fields at out-of-band frequencies when the original acoustic field is well-described by a sum of ray-path contributions. While ray theory may be a useful field description in many situations, it does not include diffraction. In this presentation, the effects of acoustic diffraction on the autoproduct are analyzed in an environment where diffraction varies in importance depending on the spatial location. Specifically, the behavior of t...

Simulations of source localization in the deep ocean using frequency-difference matched field processing

Matched field processing (MFP) is an established technique for source localization in multi-path acoustic environments that relies on correlating array-recorded and simulated sound fields. However, when the recorded sound field’s frequency is high enough, this field may be sensitive to details of the acoustic environment that are uncertain or unknown, and therefore not included in the simulated sound field. Thus, the actual and modeled acoustic fields are mismatched, the severity of this mismatch increases with frequency, and may cause MFP to fail at relevant frequencies and ranges in the deep ocean. One remedy to this problem may be analyzing the frequency-difference autoproduct of the acoustic field instead of analyzing the field itself. Thus, this presentation covers the simulated performance, and likely limitations, of frequency difference MFP [Worthmann and Dowling (2015). J. Acoust. Soc. Am. 138, 3549-3562] in a generic deep-ocean channel using a ray code (Bellhop) for the source-broadcast field in the signal bandwidth and a mode code (Kraken) for the replica calculations in the difference-frequency bandwidth. Here, refractive index fluctuations in the ocean are modeled by random time delays for each ray-path between the source and each receiver. [Sponsored by ONR.]

Adaptive frequency-difference matched field processing for high frequency source localization in a noisy shallow ocean.

Remote source localization in the shallow ocean at frequencies significantly above 1 kHz is virtually impossible for conventional array signal processing techniques due to environmental mismatch. A recently proposed technique called frequency-difference matched field processing (Δf-MFP) [Worthmann, Song, and Dowling (2015). J. Acoust. Soc. Am. 138(6), 3549-3562] overcomes imperfect environmental knowledge by shifting the signal processing to frequencies below the signal's band through the use of a quadratic product of frequency-domain signal amplitudes called the autoproduct. This paper extends these prior Δf-MFP results to various adaptive MFP processors found in the literature, with particular emphasis on minimum variance distortionless response, multiple constraint method, multiple signal classification, and matched mode processing at signal-to-noise ratios (SNRs) from -20 to +20 dB. Using measurements from the 2011 Kauai Acoustic Communications Multiple University Research Initiative experiment, the localization performance of these techniques is analyzed and compared to Bartlett Δf-MFP. The results show that a source broadcasting a frequency sweep from 11.2 to 26.2 kHz through a 106 -m-deep sound channel over a distance of 3 km and recorded on a 16 element sparse vertical array can be localized using Δf-MFP techniques within average range and depth errors of 200 and 10 m, respectively, at SNRs down to 0 dB.

High frequency source localization in a shallow ocean sound channel using frequency difference matched field processing.

Matched field processing (MFP) is an established technique for source localization in known multipath acoustic environments. Unfortunately, in many situations, particularly those involving high frequency signals, imperfect knowledge of the actual propagation environment prevents accurate propagation modeling and source localization via MFP fails. For beamforming applications, this actual-to-model mismatch problem was mitigated through a frequency downshift, made possible by a nonlinear array-signal-processing technique called frequency difference beamforming [Abadi, Song, and Dowling (2012). J. Acoust. Soc. Am. 132, 3018-3029]. Here, this technique is extended to conventional (Bartlett) MFP using simulations and measurements from the 2011 Kauai Acoustic Communications MURI experiment (KAM11) to produce ambiguity surfaces at frequencies well below the signal bandwidth where the detrimental effects of mismatch are reduced. Both the simulation and experimental results suggest that frequency difference MFP can be more robust against environmental mismatch than conventional MFP. In particular, signals of frequency 11.2 kHz-32.8 kHz were broadcast 3 km through a 106-m-deep shallow ocean sound channel to a sparse 16-element vertical receiving array. Frequency difference MFP unambiguously localized the source in several experimental data sets with average peak-to-side-lobe ratio of 0.9 dB, average absolute-value range error of 170 m, and average absolute-value depth error of 10 m.

Interpretations of the frequency difference autoproduct in multipath environments

When locating remote acoustic sources in a shallow ocean sound channel, the established array signal processing technique known as matched field processing (MFP) has shown much success. However, MFP is sensitive to mismatch between the modeled and actual environments, and may fail to localize acoustic sources in the presence of such mismatch, particularly at high frequencies. A recent nonlinear array signal processing technique, frequency difference MFP (Abadi et. al. 2012, Worthmann, et. al., under review), has shown some success in localizing high frequency sources by moving the replica calculations to a lower, out-of-band, difference frequency where the detrimental effects of environmental mismatch are less severe. To extract the requisite out-of-band difference frequency information from the measured signals, a quadratic product, termed the autoproduct, is formed from complex signal amplitudes separated by the difference frequency but still lying within the signal bandwidth. Through the use of simple multipath propagation environments, the nature of this autoproduct is explored, and the reasons that it provides out-of-band field information are presented. More complex propagation environments are simulated as well to demonstrate some of the expected and unexpected behaviors of the autoproduct. [Sponsored by the Office of Naval Research and the National Science Foundation.]

High frequency source localization in a shallow ocean sound channel using frequency-difference matched field processing

Matched field processing (MFP) is an established technique for locating remote acoustic sources in known environments. Unfortunately, environment-to-propagation model mismatch prevents successful application of MFP in many circumstances, especially those involving high frequency signals. For beamforming applications, this problem was found to be mitigated through the use of a nonlinear array-signal-processing technique called frequency difference beamforming (Abadi et. al. 2012). Building on that work, this nonlinear technique was extended to Bartlett MFP, where ambiguity surfaces were calculated at frequencies two orders of magnitude lower than the propagated signal, where the detrimental effects of environmental mismatch are much reduced. Previous work determined that this technique has the ability to localize high-frequency broadband sources in a shallow ocean environment with a sparse vertical array, using both simulated and experimental propagation data. Using simulations, the performance of this technique with horizontal arrays and adaptive signal processing techniques was investigated. Results for signals with frequencies from 10 kHz to 30 kHz that propagated in a 100-m-deep shallow ocean sound channel with a downward refracting sound speed profile will be shown for source array ranges of one to several kilometers. [Sponsored by the Office of Naval Research.]

Localization of a high frequency source in a shallow ocean sound channel using frequency-difference matched field processing

Matched field processing (MFP) is an established technique for locating remote acoustic sources in known environments. Unfortunately, environment-to-propagation model mismatch prevents successful application of MFP in many circumstances, especially those involving high frequency signals. For beamforming applications, this problem was found to be mitigated through the use of a nonlinear array-signal-processing technique called frequency difference beamforming (Abadi et al. 2012). Building on that work, this nonlinear technique was extended to MFP, where Bartlett ambiguity surfaces were calculated at frequencies two orders of magnitude lower than the propagated signal, where the detrimental effects of environmental mismatch are much reduced. In the Kauai Acomms MURI 2011 (KAM11) experiment, underwater signals of frequency 11.2 kHz to 32.8 kHz were broadcast 3 km through a 106-m-deep shallow-ocean sound channel and were recorded by a sparse 16-element vertical array. Using the ray-tracing code Bellhop as the propagation model, frequency difference MFP was performed, and some degree of success was found in localizing the high frequency source. In this presentation, the frequency difference MFP technique is explained, and comparisons of this nonlinear MFP technique with conventional Bartlett MFP using both simulations and KAM11 experimental data are provided. [Sponsored by the Office of Naval Research.]

Frequency-difference matched field processing in the presence of random scatterers

Matched field processing (MFP) is an established technique for locating remote acoustic sources in known environments. Unfortunately, unknown random scattering and environment-to-propagation model mismatch prevents successful application of MFP in many circumstances, especially those involving high frequency signals. Recently a novel nonlinear array-signal-processing technique, frequency difference beamforming, was found to be successful in combating the detrimental effects of random scattering for 10 kHz to 20 kHz underwater signals that propagated 2.2 km in a shallow ocean sound channel and were recorded by a 16-element vertical array. This presentation covers the extension of the frequency-difference concept to MFP using sound propagation simulations in a nominally range-independent shallow ocean sound channel that includes point scatterers. Here again, 10 kHz to 20 kHz signals are broadcast to a vertical 16-element array, but the frequency difference approach allows Bartlett and adaptive MFP ambiguity surfaces to be calculated at frequencies that are an order of magnitude (or more) below the signal bandwidth where the detrimental effects of environmental mismatch and random scattering are much reduced. Comparison of these results with equivalent simulations of conventional Bartlett and adaptive MFP for different of source-array ranges are provided. [Sponsored by the Office of Naval Research.]