Remote ship detection using relative multiscale weighted link entropy in marine environment

Remote passive detection of vessels within oceanic settings is paramount in bolstering port security and safeguarding coastal and offshore activities. In this context, our study introduces a sophisticated, complex network-based detector tailored for ship detection in maritime environments. Initially, we formulate a weighted transfer network, facilitating the seamless mapping of time series data onto this network structure. To enhance the efficacy of our approach across diverse temporal scales, we introduce the Relative Multiscale Weighted Link Entropy method. Furthermore, our analysis delves into the distributional attributes of the Relative Multiscale Weighted Link Entropy values, contrasting ambient noise data with and without ship-generated noise in the context of the South China Sea. Then, this paper presents a Neyman-Pearson criterion-based relative multiscale weighted link entropy detector for ship signal detection in the marine environment. The results show that the relative multiscale weighted link entropy method, compared with the narrowband energy detection method, has a 3.3 dB SNR gain under the same conditions. It can detect ships 20km away in a marine environment. In summary, juxtaposed with conventional methods such as narrowband energy detection, our proposed methodology negates the reliance on pre-established target data and demonstrates superior detection performance at diminished SNRs.