Dynamic Resource Allocation With Decentralized Multi-Task Assignment Approach for Perimeter Defense Problem

Abstract

In this article, a dynamic resource allocation with decentralized multi-task assignment (DREAM) approach using a dynamic sized team of defenders is presented to protect a perimeter of a high-security area from heterogeneous intruders. In the DREAM approach, the spatio-temporal problem of neutralizing the heterogeneous intruders is converted into a decentralized time-ordered multi-task assignment problem. A single-step dynamic resource allocation algorithm based on the computed assignments determines an optimal number of defenders required. The DREAM approach adds more defenders to the team from the reserve stations or removes the excess defenders from the team such that it always utilizes minimal resources for protecting convex territory. A trajectory computation algorithm converts the optimal multi-task assignments to trajectories for the defenders. The working of the DREAM approach for a typical perimeter defense problem is illustrated using a simulated convex territory protection problem. Based on the results from the ablation study, it is seen that one needs to deploy a higher number of defenders in the team for a better success rate to handle highly maneuvering intruders. Furthermore, a higher defender-to-intruder speed ratio helps in better resource utilization. These results clearly indicate that the DREAM approach can protect any convex territory efficiently with minimal resources.