Skip to Main Content
A modified torpedo detection algorithm (MTDA) that improves upon the range estimates of an earlier torpedo detection algorithm (TDA) is presented in this work. The original TDA used the fast orthogonal search (FOS) algorithm to estimate the frequency of direct-path and a surface-reflected-path signal emitted by a torpedo. This technique assumed that the torpedo was traveling directly at the receiver and modeled the signals using sinusoidal functions. The range to the torpedo was estimated using the frequency estimates of the direct and reflected path and a trigonometric description of the approaching signals. The TDA was successful at detecting approaching torpedoes but had an average range error of 680 m. The modified torpedo detection addresses these deficiencies by: 1) estimating the angle between the approaching torpedo and the receiver and 2) using chirp signals as the candidates for FOS that model the theoretical direct- and reflected-path signals. In addition, the FOS algorithm is modified to fit the direct and reflected paths in pairs to eliminate the need to iterate the FOS algorithm. Signals of approaching torpedoes were simulated for several initial angles of approach using the generic sonar model (GSM) and waveform transmission through a channel (WATTCH) programs. The average error of the range estimates for sea state (SS0) 0 and 1 for the MTDA were 153 m as compared to 680 m for the TDA. In addition, the MTDA was shown to be resistant to false alarms.