Multi-Armed-Bandit-Based Framed Slotted Aloha for Throughput Optimization

In this letter, a Multi-Armed-Bandit (MAB) framework is proposed for maximizing the network throughput of Framed Slotted Aloha (FSA). The throughput analysis shows that even with the simplest collision receiver, the network throughput of the proposed MAB-based FSA can reach 1 as nodes independently learn to transmit in turn. Yet a tradeoff between throughput and fairness should be considered when the frame size is smaller than the network size.