Skip to Main Content
The detection of sea-bottom targets has encountered the problems using acoustic arrays with lower resolution on board. Moreover, the uncertainty of moving receivers and sources, the extremely low signal-to-noise ratio due to weak signals from the buried targets, and the computational efficiency requirement of real-time processing present challenges to detection of the targets. For low signal-to-noise ratio, the traditional scenario, detection then tracking is not suitable for this kind of problem. The detection can't be declared unless the signal-to-noise ratio is greater than a reasonable threshold while more information is gathered. Enormous array signal processing methods could enhance the signal-to-noise ratio while the information is gathered spatially. However, the positions of the sensors are not known precisely due to the limitation of underwater navigation technique. Without knowing the spatial distribution of receivers, the results of detection by means of array signal processing cannot be accurate. In contrast, the solution to this problem becomes reversed to traditional detection procedure. A Track-Before-Detection (TBD) algorithm is introduced. Unlike the traditional approaches, the TBD detections are not declared at each ping. Instead, a number of pings of data are processed. This technique integrates the measurements along the possible AUV trajectories. Using the slowly changing environment information or the fixed but unknown target fields, the states of AUV are tracked first. Then, the weak signal detection is declared after confidence of the trajectories estimation is established. However, enormous possible trajectories of AUV needed to be searched while the TBD algorithm is applied, which makes direct online implementation of this technique impossible. A dynamic programming (DP) algorithm is introduced to solve the highly interconnected stochastic network which TBD creates in a much more efficient way. Therefore, together with the DP algorithm, t he TBD algorithm is feasible to implement online. This algorithm has been applied in GOATS (Generic Oceanographic Array Technology Sonar) project. Both monostatic and bistatic detection results are demonstrated.