Maritime Domain Awareness By Anomaly Detection Leveraging Track Information

Abstract

Techniques for anomaly detection in the maritime domain are developed in this thesis using an area metric that measures the degree of similarity, or distinction, between ships tracks using the area between ships tracks. A modified k-means algorithm is applied to the areas calculated between ships tracks to extract sea lanes and classify these ships tracks into sea lanes. Bayes theorem of conditional probabilities is used to calculate the probability of ship tracks being in different sea lanes. Two types of anomaly detection are examined in this thesis. Ships not sailing in standard sea lanes are detected using Bayes theorem, and ships following one another are detected using user-defined thresholds on the minimum area between sea tracks. The development of the metric that compares ships track data using area is the most significant result of this thesis.

Open PDF

Document Details

Document Type
Technical Report
Publication Date
Sep 01, 2018
Accession Number
AD1065072

Entities

People

  • Tauseef Ashraf

Organizations

  • Naval Postgraduate School

Tags

Communities of Interest

  • Ground and Sea Platforms

DTIC Thesaurus Topics

  • Algorithms
  • Anomaly Detection
  • Artificial Intelligence Software
  • Automatic Identification Systems
  • Bayes Theorem
  • Bayesian Networks
  • Change Detection
  • Coast Guard
  • Data Set
  • Detection
  • Detectors
  • Digital Data
  • Dimensionality Reduction
  • Identification Systems
  • Information Science
  • Machine Learning
  • Maritime Domain Awareness
  • Maritime Security
  • Neural Networks
  • Probability
  • Random Variables
  • Security
  • Standards
  • Supervised Machine Learning
  • Unsupervised Machine Learning

Fields of Study

  • Environmental science

Readers

  • Marine Hydrodynamics
  • Military/Explosive Ordnance Disposal (EOD) Technology
  • Statistical inference.