Detecting AIS position spoofing using LEO satellites, Doppler shifts, and MCMC methods.
Translated title
Detektion af AIS position spoofing ved brug af LEO satellitter, Doppler skift og MCMC metoder.
Authors
Gregersen, Emil Skovfoged ; Nissen, Martin Mølbach
Term
4. semester
Education
Publication year
2018
Pages
199
Abstract
Dette speciale er udarbejdet i samarbejde med nanosatellitproducenten GomSpace. Det undersøger, hvordan Doppler-skift i AIS-signaler, som modtages af satellitter i lav jordbane, kan bruges til at afsløre, når et skib sender en falsk position (AIS-position spoofing). Større skibe er lovmæssigt forpligtet til at udsende AIS-signaler med deres position, men denne information kan forfalskes, for eksempel ved fiskeri i beskyttede farvande. Doppler-skift er den ændring i radiosignalets frekvens, der opstår, når sender og modtager bevæger sig i forhold til hinanden. Hele systemet – skibet, radiokanalen og satellitterne – kaldes rumbaseret AIS. Specialet kortlægger, hvilke målbare og skjulte variabler i dette system der kan hjælpe med at opdage spoofing. Der opbygges sandsynlighedsmodeller, der beskriver sammenhænge mellem variablerne, og spoofing detekteres ved statistisk inferens med Markov chain Monte Carlo (MCMC), en beregningsmetode der prøver mange mulige forklaringer og sammenligner dem med data. Metoderne afprøves i opstillede scenarier for at vurdere, hvor langt et skib kan flytte sin rapporterede position væk fra den sande, før spoofing kan afsløres. Resultaterne viser, at metoderne i nogle tilfælde kan opdage spoofing ved en forskydning på 5–10 km.
This thesis was conducted in collaboration with the nano-satellite manufacturer GomSpace. It explores how Doppler shifts in AIS signals received by low Earth orbit satellites can be used to detect when a vessel reports a false position (AIS position spoofing). Larger vessels are legally required to broadcast AIS signals with their position, but this information can be falsified, for example when fishing in protected waters. The Doppler shift is the change in a radio signal’s frequency caused by motion between the transmitter and the receiver. The combined system of the vessel, the radio channel, and the satellites is referred to as space-based AIS. The thesis identifies which observable and hidden variables in this system can help detect spoofing. It builds probabilistic models describing relationships between the variables, and detects spoofing by performing statistical inference using Markov chain Monte Carlo (MCMC), a computational method that samples many possible explanations and compares them to the data. The methods are tested in designed scenarios to assess how far a vessel must move its reported position from its true location before spoofing can be detected. Results show that, in some cases, spoofing can be detected when the reported position is offset by 5–10 km.
[This abstract was generated with the help of AI]
Keywords
Documents