Anomaly detection for the security of cargo shipments

Abstract

Detecting anomalies in the maritime domain is nowadays essential as the number of goods transported by maritime containers keeps increasing. An anomaly can be described in several ways depending on the application domain. For cargo shipments, an anomaly can be defined as an unexpected relationship between ports. This chapter describes a new approach for detecting anomalies in the sequential data used to describe cargo shipments. The technique is divided in two steps. First, we find the normal itineraries with a regular expression technique. Then, we compare a given itinerary with a normal itinerary using a distance-based method in order to classify the given itinerary as normal or suspicious. The first results of this method are very promising, and it can be further improved when integrated with time-based information. This chapter presents both the methodology and some results obtained using real-world data representing container movements.