Tip:
Highlight text to annotate it
X
AgentC is a prototype of agent-based system for securing the maritime transit.
At the core of the system is an (1) agent-based simulation of maritime traffic integrated with (2) a set of multi-agent route planning and scheduling algorithms.
The simulation is based on a number of sources of real-world data on traffic and maritime activity.
First, vessel routing information is obtained from Automated Identification System (AIS) position broadcasts
Second, piracy incident data is automatically extracted from data provided by the International Maritime Bureau piracy reporting centre.
Notice the shift in the piracy activity over the previous six years (from 2004 to 2010).
Finally, geographical data describing shoreline, specific zones, ports and pirate hubs is included.
The system can simulate thousands of vessels and focuses on the area off the Horn of Affrica, i.e. the Gulf of Aden and Western Indian Ocean.
Three vessel classes are simulated in detail by the system: ,
Long-haul cargo vessels (green) representing the bulk of maritime traffic and the main targets of pirate attacks,
pirate vessels (red) of different types and strategies...
and naval vessels (blue) deployed in the area to deter pirates and assist cargo vessels in case of pirate attack
The platform simulates vessel interactions in detail -- let us see this in a case of a simulated pirate attack.
The pirate vessel P-AIS0 approaches the international coorridor and launches attack on the Amber K cargo vessel.
The cargo vessel detects the approach and alerts the naval authority of the imminent attack.
The authority alerts the nearest warship which sends its on-deck helicopter to quickly suppress the attack and block the pirate vessel.
The warship arrives later to disarm the pirate vessel.
All events in the simulation are recorded for post-processing and further analysis.
Vessel behavior is modeled using finite state machines (FSM).
Each vessel agent controls one or more vessels with real-world operational characteristics.
A range of pirate strategies are modelled, ranging from pure reactive ones with limited line of sight to fully informed and adaptive ones.
Naval forces are modeled in a hierarchical manner reflecting the real-world chain-of-command.
The AgentC system focus on the secure maritime transit problem.
The aim is to optimally coordinate the routes of cargo vessels and naval forces so that probability of a successful pirate attack is minimized.
The overall problem can be decomposed into several subproblems whose solution each contributes to the security of the transit.
The first subproblem we address is improving the existing Gulf of Aden Groupt Transit scheme
To evaluate our solutions, we have developed a detailed simulation of maritime activity in the Greater Gulf of Aden.
The naval forces patrol the Greater Gulf of Aden area using the 4W grid. Each task-force is assigned a set of cells to patrol.
The long-haul vessels follow the International Recommended Transit Corridor (IRTC)
At present, transiting cargo vessels are asked to follow a Group Transit scheme, where vessels are grouped according to their speed and transit the gulf together.
As a first way to improve the secuirty of transit, we have sought possible improvements to the existing group trasnit scheme.
We proposed changes to existing transit schedules and proposed methods which can form transit groups dynamically,
Our improvements increase the speed and consequently also the security of transit.
A disadvantage of the International Recommended Transit Corridor (IRTC) is the predictability of vessel positions which make them more easily targeted by pirates.
In a more advanced approach to improve transit security, we have therefore employed game theory
to transit ships through the Gulf of Aden in an optimally randomized manner.
The situation is formalized as a Transit Game between the transiting vessels and the pirates; pure strategies correponds to routes through/in the gulf.
Iterative oracle methods and an external linear programming library is used to find a solution in the form of a Nash equilibrium of the game.
The solution of the transit game is then deployed in the simulation and evaluated against simulated pirates.
The evaluation shows a significant drop of the attack rate compared to the existing predictable IRTC routing.
Piracy threat can never by fully suppressed without an effective deployment of law-enforcing forces.
As a third counter-piracy technique, and to complement randomized transit routing, we have therefore developed a new patrol scheduling mechanism.
We seek the optimum policy for naval vessels, given routes of cargo vessels as a Stackelberg equilibrium of a Patrolling game.
The policy can be deployed into the simulation and its effectiveness in protecting the transit evaluated.
Using the proposed Group Transit Scheme, randomized transit routing from the Transit Game and patrolling policies from Patrolling Game...
...AgentC system is able to improve the security of maritime transit through piracy-affected areas.