Solving Dynamic Multi-Criteria Resource-Target Allocation Problem Under Uncertainty: A Comparison of Decomposition and Myopic Approaches

Abstract

This paper is concerned with multi-criteria and dynamic resource allocation problem in a naval engagement context. The scenario under investigation considers air threats directed towards a ship that has to plan its engagement by efficiently allocating the available weapons against the threats to maximize its survivability. This dynamic and multi-criteria decision-making problem is modeled using a multi-criteria decision tree and solved with two approaches: the multi-criteria decomposition approach and the multi-criteria myopic approach. We propose a novel metric for comparing two strategies within a multi-criteria decision tree and have developed a testbed in order to simulate the engagements. The results show that, when sufficient decomposition conditions are verified, the decomposition approach produces superior decision-making strategies compared to the myopic approach. Conversely, when the multi-criteria decision aid (MCDA) method does not satisfy the decomposition conditions (e.g., TOPSIS), there is no guarantee that decomposition will provide the best compromise strategies. From a military perspective, this work will help develop tactics, procedures and training packages for such a highly complex and dynamic decision-making problem. The plans generated by the approach presented here can also serve as a reference for assessment of the quality of the engagement plans yielded by real-time planning algorithms.