Computational Analysis of Merchant Marine GPS Data

摘要

A series of quantitative and structural analyses are applied to geospatial data regarding the movement of Merchant Marine vessels in the English Channel. CASOS has been tasked with developing new computational techniques for analyzing data about Merchant Marine behavior under a Social Network Analysis framework. In this paper we describe an experiment doing so for geospatial data from AIS transponders in 1700+ Merchant Marine vessels during exercise in the English Channel. Our analysis has three phases: Spatial clustering algorithms are used to detect places of interest and relationships between entities the data. Extracted relational information is analyzed in network form. A suite of network analytic measures are applied to find patterns on the network and individual node levels. We apply an intervention analysis which models an intervention (surveying ships at ports) and suggests a strategy for allocating surveillance. The analysis framework is unusual in taking a relational perspective to spatial data, and novel in its principled treatment of the relationship between spatial, two mode, and one mode network representations of data, and in its approach to proposing intervention strategy.