A channel sounding and communications experiment was performed in the Oslofjord, using eight bottom-mounted instrument units deployed in a network configuration. Five units were equipped with a software-defined 4–8-kHz acoustic modem, programmed to transmit probe signals and communication packets in a round-robin fashion. All transmitted waveforms were recorded by all units over 35 horizontal links and 5 vertical links. The channels reveal a reverberant environment with long and dense multipath arrival patterns. Measured power-delay profiles and delay-Doppler spread functions are used to predict receiver output signal-to-noise ratio (SNR) over a signaling period of 26 h. To this end, the channel quantities are first calibrated for the propagation loss. The prediction examines the effect of ambient noise, reverberation of previously transmitted packets, the packet’s own reverberation, and Doppler spread. The delay profile can be used under calm conditions, and results in a mean prediction error (averaged over all links) of about 3 dB, even hours after the measurement of the profiles. The mean error on individual links (averaged over time) is reduced to 1–2 dB by using up-to-date channel information. The relevance of predicting output SNR is finally illustrated by establishing a relationship between output SNR and the probabilities of bit and packet error.

NILUS MK 2 is the newest generation of FFI's demonstrator system for easily deployable underwater sensor networks. The NILUS MK 2 sensor nodes are battery-operated and equipped with embedded processors running Linux. GStreamer is an open source modular multimedia framework, which we have chosen for implementing acoustic signal processing algorithms in such sensor nodes. We investigate the implementation of beamforming and bearing estimation algorithms in GStreamer for NILUS MK 2. Using data collected during the first NILUS MK 2 trial, the resulting bearings compare well to GPS-based "ground truth", and online (embedded) and post-trial offline (PC) processing results are in good agreement. Processing requirements are found to be modest, and the additional power consumption due to the acoustic signal processing is marginal. We find GStreamer well suited to acoustic signal processing on deployable sensor nodes.

The SONIC (Suppression Of underwater Noise Induced by Cavitation) and AQUO (Achieve QUieter Oceans by shipping noise footprint reduction) projects were awarded within the European Seventh Framework Program to develop tools to investigate and mitigate the effects of underwater sound generated by shipping activities on marine life. Model generated sound maps were identified by the European Commission as a monitoring tool to complement measurements.Sound mapping tools are being developed to provide a representation of shipping sound that is both meaningful to policy makers without requiring a background in acoustics and representative of the phenomena relevant to environmental impact. The accuracy of the end result depends on the quality of the source description as well as that of the propagation model used to compute the sources' combined contributions. In this paper we concentrate on the propagation models used for this purpose. A shipping sound map is usually expected to cover a large geographical area, including a large number of sources radiating sound over a broadband spectrum of frequencies. These requirements place restrictions on the choice of a propagation model.The challenges specific to computation of sound maps and the variety of possible approaches available to address them make it worthwhile to compare the output of different approaches, as well as with reference model solutions when available. For this purpose, five test cases were defined by scientists of the AQUO ("Achieve QUieter Oceans by shipping noise footprint reduction") and SONIC ("Suppression Of underwater Noise Induced by Cavitation") consortia. These test cases, defined with increased complexity are specified in this paper and presented with example solutions computed with a variety of models. The purpose of these test cases and associated results is to facilitate evaluation of a given sound map computation system.

This paper is dedicated to the work of Pieter Schippers and gives an overview of his achievement in sonar performance modelling over his career. This publication is the last of a long list, many of which published at UDT [1-5]. A historical review is presented of the sonar performance modeling work at TNO in Underwater Acoustics since the mid-seventies. At that time, sonar performance was a hot topic for the upcoming receiver technique using passive low frequencies towed arrays. The aim was entirely focused on detection of hostile submarines, being practically undetectable with other systems. The main challenge was the modeling of propagation, first for deep water, and later also for shallow water. The developed propagation modeling is based on eigenrays. This enabled fast semi-analytical solutions, which was a strong requirement at that time. In the late eighties active sonar regained interest [1,2], and the active and passive sonar performance model ALMOST of TNO got under development. Sonar modeling got included and source and receiver vertical directivity patterns are applied. Furthermore, ambient noise and reverberation were modelled, based on noise sources at the sea surface, and scatterers at sea surface, volume and bottom, respectively. For active and passive detection performance, detection probability is computed from the modelled SNR for different types of signal processing in the receiver. After the Cold War the sonar bandwidth increased to enable shallow water operations [3]. Targets could no longer be considered as point targets and extended target modeling was started [4]. More attention was paid to input parameters, to properly model the complex environments is shallow water. Recently, graphical user interfacing got more attention [5]. This allows even unskilled operators to get their sonar performance prediction in due time and presented in an intuitive way. The future of sonar performance modeling is both in improving models and inputs, at the same time evaluating errors and uncertainty in the modeling. Pieter will watch this development from this lazy chair.

There is an increasing concern that anthropogenic underwater noise may have a negative impact on marine life. Governmental authorities are introducing regulations to address this problem. The European Commission, for instance, has adopted the Marine Strategy Framework Directive requiring EU Member States to achieve or maintain Good Environmental Status, regarding underwater noise amongst other forms of pollution. A task group (the technical sub-group on underwater noise) has formulated indicators for underwater noise pollution, resulting in an advice to monitor low frequency sound in particular frequency bands (sound pressure level in the third-octave bands centred at 63 Hz and 125 Hz). Model generated sound maps were identified as a monitoring tool to complement measurements. Recently, the SONIC (Suppression Of underwater Noise Induced by Cavitation) project was awarded within the European Seventh Framework Program to develop tools to investigate and mitigate the effects of underwater noise generated by shipping activities. In this paper, we will present the SONIC approach to generate shipping sound maps. The sound map generation tool uses Automatic Identification System (AIS) as well as biological distribution data to generate maps representative of the sound exposure that marine mammals and fish would experience. This tool uses a fast acoustic model developed specifically for this purpose that was compared to normal modes and parabolic equation models. Results of the models compared with reference models are presented in this paper.

During naval operations, sonar performance estimates often need to be computed in-situ with limited environmental information. This calls for the use of fast acoustic propagation models. Many naval operations are carried out in challenging and dynamic environments. This makes acoustic propagation and sonar performance behavior particularly complex and variable, and complicates prediction. Using data from a field experiment, we have investigated the accuracy with which acoustic propagation loss (PL) can be predicted, using only limited modeling capabilities. Environmental input parameters came from various sources that may be available in a typical naval operation. The outer continental shelf shallow-water experimental area featured internal tides, packets of nonlinear internal waves, and a meandering water mass front. For a moored source/receiver pair separated by 19.6 km, the acoustic propagation loss for 800 Hz pulses was computed using the peak amplitude. The variations in sound speed translated into considerable PL variability of order 15 dB. Acoustic loss modeling was carried out using a data-driven regional ocean model as well as measured sound speed profile data for comparison. The acoustic model used a two-dimensional parabolic approximation (vertical and radial outward wavenumbers only). The variance of modeled propagation loss was less than that measured. The effect of the internal tides and sub-tidal features was reasonably well modeled; these made use of measured sound speed data. The effects of nonlinear waves were not well modeled, consistent with their known three-dimensional effects but also with the lack of measurements to initialize and constrain them.

Accurate environmental information is required for obtaining confident sonar performance predictions. This environmental information is, however, often unreliable or unavailable. To support antisubmarine warfare (ASW) operations, a through-the-sensor approach has been developed in which relevant acoustic seabed properties are derived from reverberation data, and a demonstrator system has been installed on a Royal Norwegian Navy frigate. It determines relevant acoustic seabed parameters from the reverberation data near real time. This demonstrator system has been validated in several sea trials conducted off the coast of Bergen, Norway. The acoustic seabed parameters derived in these trials have a good correspondence with the available prior information. Furthermore, the results show that acoustic seabed parameters derived from reverberation data in previous trials can be used to improve reverberation prediction for subsequent trials, even when environmental conditions, i.e., sound-speed profiles, are different. Because the demonstrator makes information on acoustic seabed properties directly available for in situ sonar performance prediction, it can be used as a tactical decision aid.

Sea-surface scattering by wind-generated waves and bubbles is regarded to be the main nonplatform related cause of the time variability of shallow acoustic communication channels. Simulations for predicting the quality of acoustic communication links in such channels thus require adequate modeling of these dynamic sea-surface effects. For frequencies in the range of 1-4 kHz , there is an important effect of bubbles on sea-surface reflection loss due to refraction, which can be modeled with a modified sound-speed profile (SSP) accounting for the bubble void fraction in the surface layer. The bubble cloud then acts as an acoustic lens, enhancing the rough-surface scattering by the resulting upward refraction. It is shown here that, for frequencies in the considered range of 4-8 kHz, bubble extinction, including both the effects of bubble scattering and absorption, provides a significant additional contribution to the surface loss. Model-based channel simulations are performed by applying a ray tracer, together with a toolbox for generation of rough sea-surface evolutions. This practical simulation framework is demonstrated to provide realistic results for both stationary and mobile communication nodes by capturing specific features observed in experiments, such as time variability, fading reverberation tails, and wind-speed dependence of the Doppler power spectrum.

Sidescan high-frequency (HF) sonar (i.e., with frequencies higher than 100 kHz) is ideally suited for providing high-resolution images of the seafloor. However, since sound does not penetrate into the sediment at these frequencies, such systems cannot be used for the detection of buried objects, such as naval mines, improvised explosive devices (IEDs), and unexploded ordinance (UXO). Sidescan low-frequency (LF) sonar is a promising technology for the detection of objects buried in soft seafloor sediment. Acoustic energy is attenuated less by the sediment at lower frequencies and can therefore penetrate deeper, facilitating the detection of buried objects. Furthermore, a sidelooking configuration yields a much higher area coverage rate compared to downward-looking systems (e.g., the BOSS system [1]), thus enabling efficient surveys. In practice there are two fundamental issues with sidescan LF sonar. The resolution of conventional sidescan sonar is poor at low frequencies due to the lower directivity of the beams. Moreover, in addition to the targets of interest, many clutter contacts are also observed, including other buried objects (e.g., boulders) and geological features below the mud (e.g., sand ripples). Thus, a means of classification is necessary to distinguish between targets and clutter and to suppress the false alarms. Synthetic aperture sonar (SAS) processing is essential for attaining adequate

Accurate knowledge on sonar performance is critical for design and deployment of any type of sonar. The assessment of sonar performance, however, is a complex task since it depends on system parameters and settings, target properties and tactics and on the environmental conditions. TNO has accumulated more than 30 years of sonar performance modelling experience in the ALMOST performance model. The model, based on eigenrays, allows the computation of sonar performance for active, passive and intercept sonars. In this paper, the new user interface developed around the ALMOST computation kernel is presented. It is referred to as APPROXA. This interface is designed for both regular and expert users. For regular users, appropriate default settings are selected. Expert users have the possibility to modify parameters according to their knowledge. APPROXA allows a one-click selection of environmental input data, measured or historical, on a geographical display. The environmental data can be extracted from historical databases and/or gridded data stored as Additional Military Layers (AML). The computations are performed in an N*2D fashion and performance results are presented in comprehensive tactical picture. APPROXA is therefore an accurate and easy to use tool for sonar systems design and deployment.

The European Defence Agency project RUMBLE-2 (ref.[1]) offered the opportunity for investigating experimentally acoustic reverberation at about 1.kHz, under grazing incidence (less than about 20°), from very porous clayey seafloors of a continental shelf (mean grain size spanning from about 7 to 10 in  units); the instrument was an operational, carefully calibrated LF towed triplet-array sonar. Assuming Lambert law, observed values for the normal backscattering strength μ0 spanned between –20 dB to –12 dB, i.e. 10-15 dB higher than usually recommended values for soft seabeds. The objective of this article is to understand such high values of inverted μ0’s during the RUMBLE2 experiments, which appear at first glance considerably high, particularly when considering the nature of the sedimentary cover. As displayed by the map on the left of Figure 1, the inverted μ0’s are ranging from about –20 to –12 dB, for a seabed covered by very thin clays with Mean Grain Sizes of 8 to 10  units (porosities more than 85%), as shown by the chart on the right of Figure 1. Usually recommended values for μ0 are 10 to 15 dB lower, or even more; we have to explain such a disturbing discrepancy.

Naval mines, underwater improvised explosive devices (UW-IEDs), and underwater unexploded ordnance (UW-UXO) are concerns for harbour security. In conditions without burial, existing commercial systems, such as the REMUS unmanned underwater vehicle equipped with a (very) high frequency side scan sonar (900/1800 kHz), may be deployed for detection and classification. However, it is common that bottom conditions in harbours are such that objects may become covered by mud and sometimes by sand. In the case of burial, a high-frequency sonar system cannot detect objects due to its limited penetration depth. To counter the threat of buried objects in mud, a low frequency (LF: 4-26 kHz) side looking synthetic aperture sonar (SAS) demonstrator system has been developed. This type of system has a bottom penetrating detection potential, combined with a considerable swath. Even with a limited speed of advance (e.g. 2-4 kts) an operationally relevant area coverage rate can be obtained. The system performance has been evaluated in trials in the Haringvliet, an estuary in the Netherlands, conducted in close collaboration with the Royal Netherlands Navy (RNLN). In these trials, it is confirmed that the LF SAS system is capable of detecting buried objects that are not detected by REMUS. Furthermore, buried objects are detected at ranges up to four times the water depth, indicating that a relevant area coverage rate can be achieved