An Evolutionary Vulnerability Detection Method for HFSWR Ship Tracking Algorithm

摘要

A high-frequency surface-wave radar (HFSWR) ship tracking algorithm's performance is significantly affected by the dynamics of ships, in which track fragmentation can be frequently observed. However, it is still unclear about in which scenarios the dynamics of ships sabotages the tracking performance. In this paper, an evolutionaxy-based vulnerability detection method is proposed to automatically collect scenarios of different ship motion dynamics that can cause quantitative failures in a HFSWR ship tracking algorithm. Firstly, a grammar-based scenario model which can describe multiple types of temporal relationships and generate autonomous motion of any number of ships with comparatively low-dimension data is proposed. Secondly, an encoding scheme of scenario is proposed and corresponding grammar-guided genetic programming (GGGP) algorithm is designed to evolve scenarios that can sabotages the tracking performance. Results show the effectiveness of this method in evolving and collecting scenarios that can cause more serious track fragmentation in the tracking results, with insights into the vulnerability of ship tracking algorithm provided.