Research on Radar Jamming Decision Making Based on Multi-Agent Reinforcement Learning

With the increasing diversity of weapons and equipment, jamming devices are gradually transitioning from single units to multi-unit joint jamming. Although there may be a unified command and control system for jamming devices, there needs to be adequate information exchange between them, with an apparent human subjective factor and insufficient intelligence and automation levels. In essence, they are still fighting independently and even affecting each other. To solve these problems, this chapter proposes a multi-agent reinforcement learning algorithm that combines prior information, using a centralized training and distributed execution approach, as well as the ability to handle continuous action spaces so that multiple jamming devices can sense environmental information and the actions of other jamming devices, providing a more comprehensive understanding of the task situation and enabling them to develop more effective strategies. This cooperation can significantly shorten the time required to complete the task and improve efficiency.