Maritime Piracy: Solving the Optimized Transit Path Problem

Abstract

Models have been developed that accurately predict the probability of pirate activity at locations throughout the Arabian Sea. With these piracy prediction models, mariners transiting this region can ensure that their course avoids the highest threat regions and that ample anti-piracy precautions are in place elsewhere. However, they are on their own to determine their "best" transit path.

Using unique piracy success predictors and transit cost calculators, along with existing pirate activity predictions, this research develops a method for determining the Optimized Transit Path through the Arabian Sea. This method simultaneously optimizes two different attributes, piracy avoidance and cost minimization, based on a mariner's priorities.

The Optimized Transit Path (OTP) algorithm calculates the minimum cumulative path through a two-dimensional, geographic matrix. The OTP algorithm finds the shortest path through the network from a starting line on one side of the matrix to a finish line on the other side. Using a computer code of the algorithm, experimental tests quantified the OTP algorithm's computation speed and required number of calculations to reach a solution. Further, the performance of the OTP algorithm at solving the piracy matrix was compared to the speed of other shortest path algorithms. Based on this study, the OTP algorithm's speed at solving the piracy matrix was comparable to that of the fastest shortest path algorithm in use today, Dijkstra's Algorithm implementing a Min-Priority queue with a Fibonacci Heap, and significantly faster than all others.

Because it can use the piracy prediction matrix directly as an input, the OTP algorithm is especially well suited for solving the piracy avoidance problem. More importantly, its calculation of Optimized Slack quantifies the additional cost of diverting from the shortest path, information not calculated by other shortest path methods. However, use of the OTP algorithm is fairly limited, as it is only well suited for matrices that represent a flat plane of interconnected geographic areas, with movement from a node limited to the eight adjacent nodes surrounding it.

Another promising application of the methods in this paper is within the field of underwater search.