A Self-organizing Dual-Layer Defense Algorithm for Multi-agent Systems

Abstract

With the evolution of unmanned multi-agent systems, research on multi-agent confrontation has been more important. And how to deal with multi-agent system attacks for effective self-organized defense becomes an urgent problem. The previous studies of multi-agent systems confrontation have focused more on multi-agent objective assignment methods and the study of Target-Attacker-Defender games for the single agent. When the number of agents increases, the multi-agent Target-Attacker-Defender game, which combines objective assignment and the single agent Target-Attacker-Defender games, suffers from centralized distribution and computing complexity problems. Moreover, swarm self-organization technology is boomingly being used to generate novel solutions for multi-agent defense situations. In this paper, we intend to propose a self-organizing dual-layer defense model to fight multi-agent system attacks of different attack scales and different attack strengths, which is based on an improved artificial potential field model and social force model, as well as defines rules for defenders and attackers. Finally, using simulation findings, the efficiency of the dual-layer defense model in dealing with multi-agent system attacks is illustrated and analyzed at both the swarm and individual levels.