R.K. Lennartsson

Underwater target tracking by means of acoustic and electromagnetic data fusion

An interesting possibility for improved surveillance capabilities in littoral waters is the integration of multisensor systems by means of data fusion. Here, we describe how data fusion can be used for localisation and tracking of targets by means of passive underwater acoustic and electric field sensors. The data was fused using a Kalman filter. The filter was applied on bearing estimates from the acoustic data and estimates of the target position from the electrode array. The results show the positions of the targets are gained by fusion between the acoustic bearing and the electrode sensors, within the limits of their sensitivity

Quadratic phase coupling analysis of passive sonar data using biphase techniques

In this work we report on the application of a higher order statistics based method that can be used to improve passive signature estimation over the conventional bispectrum methods. The method employs the biphase for extraction of frequency triplets with possible quadratic phase coupling. We demonstrate the method using data recorded by bottom mounted hydrophones during a sea trial in the Baltic Sea with a small fiberglass motor boat as a target. Several frequency pairs are shown to exist for which, according to the biphase test, there is significant quadratic phase coupling. This information can be useful to form complementary features for passive signature classification.

Fused Classification of Surface Ships Based on Hydroacoustic and Electromagnetic Signatures

This paper reports a classification study using data-fusion on real-world, underwater signatures from surface ships. In this study we design and evaluate classifiers optimized to discriminate between small and big ships based on features extracted from extremely low frequency electric (ELFE) and hydroacoustic signatures. Classification is performed separately on each signature, and finally the individual decisions are fused into a common decision. The data set analysed here was recorded at a sea trial conducted in the Baltic Sea off the east coast of Sweden. The data set contains 23 passages of surface ships of various size, divided into small (up to 21484 tons) and big (above 21484 tons)ships based on their displacement.

Analysis of underwater acoustic and electric noise in the port of Gothenburg

In this paper we study some of the underwater background noise properties in the port of Gothenburg. The analysis includes both electric and acoustic noise. We investigate the dynamic range of the absolute noise power as well as its temporal variations. Furthermore, we study the spectral variations of the low-frequency noise power. Stationarity and normality tests are applied in order to quantify the amount of transients and temporal variations in the data. We find that the properties of the acoustic and electric fields are very different. The temporal variability is larger for the acoustic data. The local shipping noise is the most important acoustic source, while the wind is less important. Shipping also give rise to the highest recorded levels in the electric field noise, but due to the short detection ranges, compared with acoustics, only ships passing nearby the sensors will contribute to the background.

Improved passive acoustic detection of divers in harbor environments using pre-whitening

Underwater surveillance in a harbor is typically performed using active sonar systems. The performance of an active sonar can drastically change due to rapid variations in the sound propagation. This paper presents a method for detecting divers with open circuit breathing systems using passive acoustics. The authors have previously reported on a method for passive acoustic diver detection that employs two hydrophones. The proposed method uses a single hydrophone, resulting in a simpler and cheaper system. The method works by whitening the background noise and mitigating the influence of short-duration transients and rapid variations in the background noise. Results presented here show that the proposed algorithm performs better than the previous dual hydrophone method. Compared to the dual hydrophone method applied to the same frequency band, the proposed method achieves greater detection ranges at significantly lower false alarm rates. Pre-whitening permits us to use a wider analysis frequency band, resulting in further improvements to the detection range.

Fused passive acoustic and electric detection of divers

Underwater surveillance against small and low signature targets is challenging, especially in disturbed and shallow water areas such as harbors. Today the most frequently used system for surveillance in harbors is active sonar. However, the detection range of an active system can rapidly degrade due to changes in the sound propagation conditions and in the ambient noise. We focus our research on how passive underwater sensors can be used as a complement to active acoustic systems. We have previously reported on diver detectors for passive acoustic and electric field data, with promising results. The use of the two sensor systems together is mainly motivated by the fact that the acoustic and electric background noise often are uncorrelated. In this paper we test different data fusion methods in order to combine the decisions from the two detectors into one unified decision. The performance of the detectors and the data fusion methods is evaluated using data from a sea trial conducted in the port of Gothenburg in 2009. We present an approach to fusion that will result in a robust system solution for harbor security.

Stationarity Analysis of Ambient Noise in the Baltic Sea

Most signal processing techniques are only valid if the assumption of stationarity is true. This is the basis for making reliable and consistent estimates. Stochastic processes can be categorised by their stationarity properties ranging from stationary to non-stationary processes. The degree of stationary has implications on a number of factors in signal processing, but mainly on the level of reliability of any estimate. Estimates from highly non-stationary data can at times be so bad that the variance of the estimate is by far greater than the estimate itself. In this paper, the degree of stationarity is addressed from a stationarity length point of view and sonar data is tested using the Kolmogorov-Smirnov two sample test

Improved Classification Ability in Littoral Environments by Decision Fusion

In this paper we investigate if the classification ability in littoral environments can be improved by decision fusion. We design and evaluate classifiers for discrimination of two ship types. The data was recorded with an underwater multisensor system at a sea trial conducted in the Baltic Sea off the east coast of Sweden. The sensor system consisted of both electrode sensors and hydrophones. The electrode sensors recorded extremely low frequency electric (ELFE) signatures and the hydrophones recorded underwater acoustic signatures of passing surface ships. Classification is performed separately on each signature and finally the separate decisions are fused into one unified decision. In addition, the possible benefits of temporal decision fusion are investigated.

Sonar data conditioning by means of stepwise outlier rejection

Target signature estimation is important in passive sonar signal processing. Higher order spectral methods such as bispectral estimation are often used as a complementary tool to conventional analysis. However, these methods require larger data sets to achieve consistent estimates. The sensitivity for non-stationarities are also more pronounced for bispectral methods. Here, we use a stepwise outlier rejection algorithm as both a stationarity test and a data conditioner for experimental sonar data. We demonstrate the importance of testing for significant quadratic phase coupling and using outlier rejection as a first step in bicoherence analysis of sonar data.

Performance evaluation of underwater target tracking using data fusion on acoustic and electromagnetic data

Underwater passive acoustic target tracking is challenging in littoral environments. One way to mitigate the difficulties is to add non-acoustic sensors and use data fusion. The topic of this paper is how to evaluate, in an objective way, the performance of data fusion in this application. Different performance measures are discussed. The performance measures are applied on data from a trial where one acoustic and one electric source were towed by a ship, simulating an underwater target. The trial was performed in a very shallow water littoral environment. Data were recorded using a passive acoustic horizontal line-array and an array of underwater electric sensors. The results show that there is a risk of degraded performance after fusion, if the data from the sensors have very different quality. In order to reduce this risk, it is crucial to reliably evaluate the quality of each estimate at all times.