I have been working on single-agent reinforcement learning algorithms to learn the environment. But I am wondering how to use Q-Table for multi-agent reinforcement learning? Specifically when the two opposing agents shared the environment statistics. For example, the anti-jamming case when two agents, a user, and a jammer, take their actions to oppose each other. The objective of legitimate user is to avoid the jammer and the objective of the jammer is to dismantle communication by generating high power noise signals in the frequency band of legitimate user.

Should I consider both agents separately and formulate the Q table separately?

or

Should the Q table be shared by two agents? How to do so?