Publications

We address the challenge of detecting an arbitrary- shaped underwater acoustic signal. Instead of setting a detection threshold, which due to noise transients may result in a high false alarm rate, our method classifies each measured sample as either ’noise’ or ’signal’. Utilizing a-priori knowledge of only the minimal duration of the signal, the decision is made using loopy belief propagation over a factor graph. Numerical simulations and sea experimental results show that our scheme achieves a favorable trade-off between the recall and false alarm rates, and noise robustness which far exceeds that of benchmark schemes.

With the massive development of underwater small robotic vehicles and matching acoustic modems, applications for Internet of Underwater Things (IoUT) are emerging. IoUT involves communication between non-synchronized network nodes organized in a mesh. A limiting factor of such communication is the so-called near-far effect, where transmissions from a node (near) close to a common receiver blocks the transmissions of a farther node (far). Due to the high-power attenuation in the underwater acoustic channel, near-far is common in underwater acoustic communication networks, and the phenomena occurs even for a distance ratio of 80% between the near and far nodes to the receiver, and the large number of nodes in IoUT compounds the effect of this phenomena. While current approaches only consider the jamming effect to the far signal, in this paper, we consider cancelling the interference from both sources by estimating and equalizing the channels on parallel, thereby significantly improving the decoding of both signals. As a result, IoUT performance improves. To limit mutual interference, we propose an automatic switching mechanism that controls the cancellation operation both in channel estimation and channel equalization. Simulation results show that our approach obtains significant improvement for communication from both near and far nodes. Results from a designated sea trial demonstrate that when both nodes are affected by their mutual transmissions, our proposed method improves the output signal-to-noise ratio (SNR) significantly.

Abstract—A limiting factor of communication networks is the so-called near-far effect, where transmissions from a node (near) close to a common receiver blocks transmissions of a farther node (far). Due to the high power attenuation in the underwater acoustic channel, near-far is common in underwater acoustic communication networks (UWANs), and the phenomena occurs even for a rate of 1/2 between the distances of the near and far nodes to the receiver. In this paper, we offer a novel interference cancellation (IC) approach for handling such interference by decoding the ”near” signal and the ”far” one in parallel, thereby improving detection for both signals. Simulation results show that, compared to the traditional decision feedback equalizer, our approach gains 12 dB in the output SNR of the jammed source and 4 dB in the output SNR of the jamming source.
Index Terms—Underwater acoustic communications, near-far, interference cancellation, compresive sensing

Abstract—We consider the challenge of tracking and estimating the size of a single submerged target in a high reverberant un- derwater environment using a single active acoustic transceiver. This problem is common for a multitude of applications, ranging from the security and safety needs of tracking submerged vehicles and scuba divers, to environmental research and management implications such as the monitoring of pelagic fauna. Considering that the target can be either slow (e.g., a scuba diver) or fast- moving (e.g., a shark), we avoid continuous signalling, and rely on the emission of wideband pulses whose reflection pattern are evaluated and reshaped in a time-distance matrix. As opposed to common approaches that track targets through template matching or by using tracking filters, we avoid making difficult assumptions about the target’s reflection patterns or motion type, and instead perform probabilistic tracking using a constraint Viterbi algorithm, whereby detection is determined based on maximum likelihood criterion. In this process, we use the expectation-maximization (EM) approach to manage stationary reflections through distribution analysis, which otherwise may be misidentified as targets. Based on the tracked path, we then evaluate the target’s size. To test our approach, we performed extensive simulations as well as eight sea experiments in different environmental settings to track both a scuba diver and a sandbar shark (Carcharhinus plumbeus). The simulation results show a tracking performance that is close to the Crame ́r-Rao lower bound, and the experiment results show a good trade-off between detection rate and false alarm rate for a low signal-to-clutter ratio of 5 [dB], and average tracking error of 1.5 [m] and 6.5 [m] in the detections of a scuba diver and sandbar shark, respectively. For reproducibility, we share our sea experiment data.

We consider the challenge of tracking and estimating the size of a single submerged target in a high reverberant underwater environment using a single active acoustic transceiver. This problem is common for a multitude of applications, ranging from the security and safety needs of tracking submerged vehicles and scuba divers, to environmental research and management implications such as the monitoring of pelagic fauna. Considering that the target can be either slow (e.g., a scuba diver) or fast moving (e.g., a shark), we avoid continuous signaling, and rely on the emission of wideband pulses whose reflection pattern are evaluated and reshaped in a time-distance matrix. As opposed to common approaches that track targets through template matching or by using tracking filters, we avoid making difficult assumptions about the target's reflection patterns or motion type, and instead perform probabilistic tracking using a constraint Viterbi algorithm, whereby detection is determined based on maximum likelihood criterion. In this process, we use the expectation-maximization approach to manage stationary reflections through distribution analysis, which otherwise may be misidentified as targets. Based on the tracked path, we then evaluate the target's size. To test our approach, we performed extensive simulations as well as eight sea experiments in different environmental settings to track both a scuba diver and a sandbar shark (Carcharhinus plumbeus). The simulation results show a tracking performance that is close to the Cramér-Rao lower bound, and the experiment results show a good tradeoff between detection rate and false alarm rate for a low signal-to-clutter ratio of 5 dB, and average tracking error of 1.5 and 6.5 m in the detections of a scuba diver and sandbar shark, respectively. For reproducibility, we share our sea experiment data

We consider the problem of estimating the trajectory of a submerged source emitting acoustic signals without using any anchor nodes or receiving array. This approach is required for several applications, including the localization of acoustic sources such as marine mammals or underwater vehicles, for which the cost of covering a broad area with multiple receivers would be excessively high. Since multi-lateration is impossible in this scenario, we perform localization by incorporating bathymetry information. Specifically, we assume that the receiver retains knowledge of the environmental parameters that affect the signal propagation, and that the bathymetry of the area is sufficiently diverse to induce distinguishable channel impulse responses for different source locations. Our method compares the channel estimates obtained from the received acoustic signals against a database of channel responses, pre-computed through an acoustic ray tracing model. The set of possible node locations that result are then organized in trellis form to obtain a final estimate of the source’s trajectory via a path tracking method similar to the Viterbi algorithm. Our results show that the proposed approach can estimate node locations and paths with very small error, provided that the receiver has sufficiently accurate and up-to- date environmental information.

We report the details of ASUNA, a freely shared dataset for underwater network emulation (ASUNA). ASUNA tackles the time-consuming and costly logistics of multiple underwater networking sea trials by providing a benchmark database of time-varying network topologies recorded across multiple sea experiments, thus facilitating experiment replay and network emulation. The ASUNA database currently includes 20 diverse, time-varying topology structures, multimodal communication technologies, and different link quality measurements. With the aim of becoming a standard benchmark, ASUNA is open to extensions as new data becomes available from the underwater communications community. We provide the details of ASUNA structure, the list of recorded topologies, as well as examples of how to use the database as part of an emulation system to test the performance of two scheduling protocols. We freely share the database and the emulation code both through a web server and via the Code Ocean repository.

Monitoring of the marine environment requires large amounts of data, simply due to its vast size. Therefore, underwater autonomous vehicles and drones are increas- ingly deployed to acquire numerous photographs. How- ever, ecological conclusions from them are lagging as the data requires expert annotation and thus realistically can- not be manually processed. This calls for developing au- tomatic classification algorithms dedicated for this type of data. Current out-of-the-box solutions struggle to provide optimal results in these scenarios as the marine data is very different from everyday data. Images taken under water dis- play low contrast levels and reduced visibility range thus making objects harder to localize and classify. Scale varies dramatically because of the complex 3 dimensionality of the scenes. In addition, the scarcity of labeled marine data pre- vents training these dedicated networks from scratch. In this work, we demonstrate how transfer learning can be uti- lized to achieve high quality results for both detection and classification in the marine environment. We also demon- strate tracking in videos that enables counting and measur- ing the organisms. We demonstrate the suggested method on two very different marine datasets, an aerial dataset and an underwater one.

This paper considers the problem of estimating the trajectory of an autonomous underwater vehicle (AUV) via a single passive receiver, without any anchor nodes or receiving arrays, and with the only help of a sequence of known acoustic signals emitted by the AUV. This scenario is of interest in case multilateration-based alternatives would require the deployment of many receivers and imply exceedingly high costs, e.g., for the coverage of wide areas. The proposed method exploits the knowledge of environmental parameters such as the sound speed profile, bathymetry and bottom sediments in order to estimate the location of the AUV, taking advantage of the spatial dependency of channel impulse responses that arises from the diverse bathymetry around the receiver. This dependency is captured by comparing channel estimates against a database of channel responses, pre-computed through an acoustic propagation model. This yields multiple likely AUV locations, which are filtered via a path tracking method similar to the Viterbi algorithm, in order to estimate the trajectory of the AUV. Results obtained both from simulations and from a sea experiment show that the proposed method can estimate node locations and paths with a small error, especially considering the use of a single receiver.

Accurate detection and tracking of moving targets in underwater environments pose significant challenges, because noise in acoustic measurements (e.g., SONAR) makes the signal highly stochastic. In continuous marine monitoring a further challenge is related to the computational complexity of the signal processing pipeline—due to energy constraints, in off-shore monitoring platforms algorithms should operate in real time with limited power consumption. In this paper, we present an innovative method that allows to accurately detect and track underwater moving targets from the reflections of an active acoustic emitter. Our system is based on a computationally- and energy-efficient pre-processing stage carried out using a deep convolutional denoising autoencoder (CDA), whose output is then fed to a probabilistic tracking method based on the Viterbi algorithm. The CDA is trained on a large database of more than 20,000 reflection patterns collected during 50 designated sea experiments. System performance is then evaluated on a controlled dataset, for which ground truth information is known, as well as on recordings collected during different sea experiments. Results show that, compared to the benchmark, our method achieves a favorable trade-off between detection and false alarm rate, as well as improved tracking accuracy.

We introduce the communication operations of THEMO: the new Texas A % M - University of Haifa - Eastern Mediterranean Marine Observatory. THEMO includes moored sensors, surface sensors, and secondary moorings. In all cases, the data from the sensors is transferred in near real-time to a shore station where it is freely shared with the community. Variety of communication aids are used. These include underwater acoustic communications to connect the sensors onboard the secondary moorings with THEMO's surface platform, underwater inductive communication to connect the moored submerged sensors to the buoy's controller, and radio communication connecting the mooring to the shore station. After deployment of almost a year, our results show that the communication performance are reliable, and the data flows in all weather conditions. In this paper, we describe the mooring's communication applications, and share the details of the different communication components.

In this paper, we present the mechanical structure of SYMBIOSIS: an opto-acoustic system that provides a mature, cost effective autonomous capability for the characterization, classification, and biomass evaluation of six target pelagic fish that are important to the fishery industry and that reflect on the health of the environment. SYMBIOSIS is devised as a blend of acoustic and optical components. The acoustic unit relies on an active underwater acoustic array that transmit wideband signals and analyze reflections to find moving targets. Once found, the optical system takes command and identifies the fish's species. The system is completely autonomous and made to withstand a few months of deployment between recharging. We describe the basic components of SYMBIOSIS, as well as operational procedures for the system.

The accurate detection and quantification of sub- merged targets has been recognized as a key challenge in marine exploration, one that traditional census approaches cannot handle efficiently. Here we present a deep learning approach to detect the pattern of a moving fish from the reflections of an active acoustic emitter. To allow for real-time detection, we use a convolutional neural network, which provides the simultaneous labeling of a large buffer of signal samples into clutter and moving targets. This allows to capture the structure of the reflecting signal from the target and to separate it from clutter reflections. We evaluate system performance both on synthetic (simulated) data, as well as on real data recorded over 50 sea experiments in a variety of sea conditions. When tested on real signals, the deep network trained on simulated patterns showed non-trivial detection capabilities, suggesting that transfer learning can be a viable approach in such scenarios, where tagged data is often lacking. However, training the network directly on the real reflections allowed to reach a more favorable precision-recall trade-off, approaching an ideal detection bound. We also evaluate an alternative model based on recurrent neural networks which, despite exhibiting slightly inferior performance, could be applied in scenarios requiring on-line processing of the reflection sequence.

The recent boost in undersea operations has led to the development of high-resolution sonar systems mounted on autonomous vehicles. These vehicles are used to scan the seafloor in search of different objects such as sunken ships, archaeological sites, and submerged mines. An important part of the detection operation is the segmentation of sonar images, where the object's highlight and shadow are distinguished from the seabed background. In this paper, we focus on the automatic segmentation of sonar images. We present our enhanced fuzzy-based with Kernel metric (EnFK) algorithm for the segmentation of sonar images which, in an attempt to improve segmentation accuracy, introduces two new fuzzy terms of local spatial and statistical information. Our algorithm includes a preliminary de-noising algorithm which, together with the original image, feeds into the segmentation procedure to avoid trapping to local minima and to improve convergence. The result is a segmentation procedure that specifically suits the intensity inhomogeneity and the complex seabed texture of sonar images. We tested our approach using simulated images, real sonar images, and sonar images that were created in two different sea experiments, using multibeam sonar and synthetic aperture sonar. The results show accurate segmentation performance that is far beyond the state-of-the-art results.

We study sharks biomass in open-sea using non-invasive active acoustics. The importance of continuous long-term monitoring of top-predator biomass is vital in understanding the healthiness of the ecosystem. Instead of the traditional fishery data, catch-and-release methods and visual inspections, which are problematic to supply reliable statistics, we rely on acoustic tools for the quantitative estimation of the number of sharks in a given area, their size, and the evaluation of their motion patterns and behavior. We take a blind classification approach and identify sharks' related reflections from sea boundary reflections based on a track-before-detect approach. Specifically, by emitting a series of wideband acoustic signals, we create a time-delay image whose rows correspond to the received reflection response. We rely on the observation that sea boundary reflections are characterized by a random clutter-like pattern, while shark's related reflections are continuous and steady. Thus, we detect a shark in a clutter by identifying in the time-delay image continuous but curved lines whose structure meet certain limitations, namely, the shark's maximal speed and its expected carangiformmotion pattern. In this paper, we will describe our method in details and show results from a sea experiment that included a verified detected shark.

Multiple-input–multiple-output (MIMO) system offers a promising way for high data rate communication over bandwidth-limited underwater acoustic channels. However, MIMO communication not only suffers from intersymbol interference, but also introduces the additional co-channel interference, which brings challenge for underwater acoustic MIMO channel estimation and for channel equalization. In this article, we propose novel interference cancellation (IC) methods for handling this co-channel interference problem in the design of both channel estimation and channel equalization. Our method for channel estimation utilizes the spatial joint sparsity and the temporal joint sparsity in the multipath structure to estimate sparse channels with common delays under distributed compressed sensing framework. In this way, we enhance channel estimates with common delays, thus, suppress co-channel interference. Meanwhile, to address the case of multipath arrivals with different delays, which are estimated as noise under simultaneous orthogonal matching pursuit (SOMP) algorithm, we introduce forward–reverse strategy to SOMP algorithm, which is referred to as the FRSOMP algorithm. Our proposed FRSOMP algorithm performs the SOMP algorithm to achieve the initial channel estimates, performs the forward-add process, which attempts to add promising candidates into support sets, and performs the reverse-fetch process to check if the candidates in the support set are retained or removed. The purpose of channel estimation is to directly calculate the filter coefficients for channel-estimation-based decision feedback equalization (CE-DFE). In this article, we also propose a novel CE-DFE receiver with IC component. We design IC filters based on the traditional CE-DFE, and we derive the coefficients of the feedforward filters, feedback filters, and IC filters based on the channel estimate metric obtained by the FRSOMP algorithm, so the co-channel interference will be suppressed both in channel estimation and channel equalization. We demonstrate the performance of our approach by numerical simulation, lake experiment, and sea experiment. Results are provided to demonstrate the effectiveness of the proposed methods, which show that the proposed methods obtain higher output signal-to-noise ratio, lower bit error rate, and more separated constellations compared with the traditional compressed sensing channel estimation method and the traditional CE-DFE method.

The recent boost in undersea operations has led to the development of high-resolution sonar systems mounted on autonomous vehicles, and aimed to scan the sea floor and detect objects. An important part of sonar detection is the image denoising, where the background is smoothed and noise components are removed while preserving the object's borders. Sonar image denoising is a challenging task, mostly due to the heavy intensity inhomogeneity of the background and the heavy spatial varying background. In this paper, we propose an algorithm for sonar image denoising that is based on the adaptation of the nonlocal means-based filter. The noise in the highlight and background regions is modeled by the exponential distribution, while the noise in the shadow region is modeled by the Gaussian distribution. We estimate the label of each pixel through image segmentation to estimate the parameters of each distribution. Then, the minimum entropy criteria is used to decide which statistics model in the denoising filter gives the best results. Results for synthetic sonar images and over real sonar images demonstrate that the proposed method successfully removes the noise components while preserving the object's edges.

We focus on mitigating strong interferences that can jam underwater acoustics emissions aimed for detection or communications. We consider two types of interferences: narrowband like continuous waveform (CW) and wideband like chirp. Both types are assumed to be strong, such that, without interference cancellation, performance is poor. We offer two interference canceling algorithms, each corresponding to a different interference type. The two algorithms are designed to mitigate strong interference, while maintaining the desired signal intact. These algorithms can be executed sequentially to manage both types of interference simultaneously. Our solution takes advantage of the sparsity of the underwater acoustic channel, as well as the assumed correlation of the interference signals. Numerical simulations, as well as results from a sea experiment, show that our algorithms significantly reduce the effect of strong interferences for the fast time-varying and long-delay spread underwater acoustic channel.

We focus on mitigating strong interferences that can jam underwater acoustics emissions aimed for detection or communications. We consider two types of interferences: narrowband like continuous waveform and wideband like chirp. Both the types are assumed to be strong, such that, without interference cancellation, performance is poor. We offer two interference canceling algorithms, each corresponding to a different interference type. The two algorithms are designed to mitigate strong interference, while maintaining the desired signal intact. These algorithms can be executed sequentially to manage both types of interference simultaneously. Our solution takes advantage of the sparsity of the underwater acoustic channel, as well as the assumed correlation of the interference signals. Numerical simulations, as well as results from a sea experiment, show that our algorithms significantly reduce the effect of strong interferences for the fast time-varying and long-delay spread underwater acoustic channel.

Presentation of results

We propose underwater delay-tolerant routing via probabilistic spraying (UDTN-Prob), a routing protocol for underwater delay-tolerant networks based on the store-and-forward paradigm. Our protocol exploits limited statistical knowledge of the time between two subsequent contacts between pairs of network nodes in order to filter the packets injected into the network, so that only those with a sufficiently high chance of being delivered to their intended destination within a given deadline are actually transmitted. In addition, the foreseen duration of a contact is estimated via a preliminary packet exchange, so that the nodes get a fair share of the contact time to exchange their own data. The transmission is protected against channel-induced packet losses via an automatic repeat query scheme modified to adapt itself to typical underwater transmission times and to the variation of round-trip times induced by node mobility. We simulate the protocol using the DESERT Underwater libraries, that make it possible to accurately reproduce the nodes' behavior and mobility patterns. Our results show that the proposed protocol achieves significantly better performance than spray-and-wait, which is currently the most typical choice among store-and-forward protocols. Moreover, we show that a two-hop statistical knowledge of the node contact process yields marginally higher utility with respect to a simpler one-hop knowledge, which is also much easier to collect or estimate.

We present a scheme to estimate the direction of arrival of acoustic signals reflected by underwater targets using wideband hydrophone arrays of opportunity. Such arrays may be obtained by arranging together multiple smaller sub-arrays that were originally designed to work independently. The array of opportunity that results may be subject to practical mounting limitations, hence the typical constraint that closest array elements should not be spaced more than one half-wavelength may not be upheld. In these conditions, the array is affected by spatial ambiguity. Our proposed scheme solves this issue by fusing direction-of-arrival information with side information on the estimated target location (obtained via multilateration). This makes it possible to eliminate most of the ambiguity, and yields accurate direction-of-arrival estimates. Our simulation results show that our scheme achieves satisfactory direction of arrival estimation and localization results. Moreover, even by relying on arrays of opportunity, we can outperform classical direction-of-arrival algorithms applied to larger arrays with half-wavelength spacing design.

The in situ detection, recognition and tracking of marine animal species is a very important step of field research in the domains of, among others, biology and ecology. Still, the direct observation of marine wildlife through equipment operating in the visible light spectrum is often impaired by the challenging conditions offered by ocean waters, where light can be subject to scattering and attenuation phenomena due to the water turbidity. As the use of powerful lighting may prove ineffective and even induce behavioral changes in marine animals, the design of minimally or non-invasive observation instruments becomes particularly important. In this paper, we consider the serial Light Detection And Ranging (LiDAR) system under development at the Florida Atlantic University (FAU). This LiDAR design is based on inexpensive components and on low average power red lasers which are subject to significant attenuation in water, but are both eye-safe and invisible to marine life. Considering the challenge of detecting and evaluating the presence of marine wildlife, we present a full processing pipeline for LiDAR data, that includes water turbidity detection, non-gated backscattering compensation, contrast enhancement, and the construction of a three-dimensional model of the detected target. The pipeline is applied to a number of tank test data, under different turbidity conditions.

Underwater sensing and remote telemetry tasks necessitate the accurate geo-location of sensor data series, which often requires underwater acoustic arrays. These are ensembles of hydrophones that can be jointly operated in order to, e.g., direct acoustic energy towards a given direction, or to estimate the direction of arrival of a desired signal. When the available equipment does not provide the required level of accuracy, it may be convenient to merge multiple transceivers into a larger acoustic array, in order to achieve better processing performance. In this paper, we name such a structure an “array of opportunity” to signify the often inevitable sub-optimality of the resulting array design, e.g., a distance between nearest array elements larger than half the shortest acoustic wavelength that the array would receive. The most immediate consequence is that arrays of opportunity may be affected by spatial ambiguity, and may require additional processing to avoid large errors in wideband direction of arrival (DoA) estimation, especially as opposed to narrowband processing. We consider the design of practical algorithms to achieve accurate detections, DoA estimates, and position estimates using wideband arrays of opportunity. For this purpose, we rely jointly on DoA and rough multilateration estimates to eliminate spatial ambiguities arising from the array layout. By means of emulations that realistically reproduce underwater noise and acoustic clutter, we show that our algorithm yields accurate DoA and location estimates, and in some cases it allows arrays of opportunity to outperform properly designed arrays. For example, at a signal-to-noise ratio of –20 dB, a 15-element array of opportunity achieves lower average and median localization error (27 m and 12 m, respectively) than a 30-element array with proper λ/2 element spacing (33 m and 15 m, respectively). We confirm the good accuracy of our approach via emulation results, and through a proof-of-concept lake experiment, where our algorithm applied to a 10-element array of opportunity achieves a 90th-percentile DoA estimation error of 4° and a 90th-percentile total location error of 5 m when applied to a real 10-element array of opportunity.

In this paper, we focus on the segmentation of sonar images to achieve underwater object detection and classification. Our goal is to achieve accurate segmentation of the object’s highlight and shadow regions. We target a robust solution that can manage different seabed backgrounds. Segmentation of sonar images is a challenging task. Speckle noise and intensity inhomogeneity may cause false detections and complex seabed textures, such as sand ripples and seagrass, often leading to false segmentation. In this paper, we propose our local spatial mixture (LSM) method for image segmentation of sidescan deployed sonar systems of any type. This new method estimates pixel labels in sonar images by incorporating the possible spatial correlation between neighboring pixels for improved segmentation. LSM modifies the expectation–maximization algorithm by adding an intermediate step (I-step) between the expectation (E-step) and maximization (M-step) steps. To combat intensity inhomogeneity, we employ a new initialization algorithm, one whose thresholds are set automatically to achieve and maintain robustness in various underwater environments. Using multiple evaluation indexes that measure the geometrical features of the segmented objects, we tested LSM using synthetic and real sonar images, one of which is obtained from our own sea experiment. Our results show that LSM achieves improved segmentation performance over the state-of-the-art methods of four different approaches; LSM is also robust to different seabed textures and intensity inhomogeneity. We share the sonar images from our sea experiments.

Under submission