Value factorisation is a useful technique for multi-agent reinforcement learning (MARL) in global reward game, however, its underlying mechanism is not yet fully understood.

Therefore, fairness in FL has attracted substantial interest in the recentpast[32,22,19,13,35].