Surface Ship Wake Detection using Active SONAR

Kishore, Gurram and Jana, Soumya (2018) Surface Ship Wake Detection using Active SONAR. Masters thesis, Indian Institute of Technology Hyderabad.

Text
Thesis_Mtech_EE_4192.pdf - Submitted Version
Restricted to Repository staff only until 2023.
Download (2MB) | Request a copy

Abstract

Detection of enemy ship in the oceanographic regions of any country is crucial for its defence. Existing techniques to detect the enemy ship have certain drawbacks like low signal to clutter ratio or developed counter techniques. An improved algorithm for active SONAR based wake detection is proposed in this thesis. SONAR became popular in oceanographic applications since world war II. SONAR sensor transmits sound signal and receives the echo signal re ected from the target. A moving ship in ocean generates bubbles behind it due to ship's hull and propeller cavitation etc. The bubbles mixed with turbulent sea water is called wake. This wake persists several minutes and carries signature of the ship. Therefore, ship wakes are quite helpful in detection of ships. The proposed technique detects ship based on features of echo signal re ected from the wake. An experimental environment was set up consisting of an upward looking SONAR placed 30m below the sea surface, transmits acoustic signal and receives the corresponding echo signal re ected from the target. The transmitted signal incidents on the wake if ship is present otherwise it hits sea surface. The received echo signal has signature of the target it hits i.e. wake or sea surface implicitly. The implicit signature of target is identi�ed by extracting features of the echo signal. The features include Physics based features and signal processing features. In this thesis six features are considered. The extracted features are classiffied using support vector machine (SVM) classification algorithm. SVM constructs a hyperplane by maximizing the margin between the classes with labelled training data. It assigns label to the test data based on which side of hyperplane it falls. Here only two classes are present, where 1) Wake is present and 2) lacks any wake (ambient). The proposed technique is robust in detection of the wake because it is able to detect even weaker wake generated by ship moving with low speeds (as low as 6 Knots) also.

[error in script]

IITH Creators: