In this paper, we provide a maritime counter-piracy framework to represent the strategies put in place and the interaction between a set of actors (patrollers and attackers) in a wide maritime scenario. Specifically, we model the interaction between patrollers and attackers in terms of a Stackelberg leader-follower game. With respect to the previous literature, we provide several innovations, in that we consider moving targets. Moreover, we let the attackers and patrollers ground their decisions not only on the current situation, but also on the expected evolution in the short or medium term. This is done by introducing in the payoff functions some projection terms, i.e., elements able to consider the future evolution of the scenario, up to a specified time window horizon. We conclude this paper with a simulation campaign considering a realistic test bed settled in the Gulf of Aden. Our experiments show that such a prediction is quite effective in generating better decisions allowing the patrollers to dynamically change targets. However, if the time horizon considered in the projection term is too large, we observe that the patrollers tend to patrol the area rather than focusing on specific vessels, and the resulting protection strategy has reduced effectiveness.
A Stackelberg game-theoretical approach to maritime counter-piracy
Oliva G;Setola R;
2019-01-01
Abstract
In this paper, we provide a maritime counter-piracy framework to represent the strategies put in place and the interaction between a set of actors (patrollers and attackers) in a wide maritime scenario. Specifically, we model the interaction between patrollers and attackers in terms of a Stackelberg leader-follower game. With respect to the previous literature, we provide several innovations, in that we consider moving targets. Moreover, we let the attackers and patrollers ground their decisions not only on the current situation, but also on the expected evolution in the short or medium term. This is done by introducing in the payoff functions some projection terms, i.e., elements able to consider the future evolution of the scenario, up to a specified time window horizon. We conclude this paper with a simulation campaign considering a realistic test bed settled in the Gulf of Aden. Our experiments show that such a prediction is quite effective in generating better decisions allowing the patrollers to dynamically change targets. However, if the time horizon considered in the projection term is too large, we observe that the patrollers tend to patrol the area rather than focusing on specific vessels, and the resulting protection strategy has reduced effectiveness.I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.