Making Slotted Aloha Efficient And Fair Using Reinforcement Learning

Reinforcement learning (RL) has been proposed as a technique that allows nodes to learn to coordinate their transmissions in order to attain much higher channel utilization. Several RL-based approaches have been proposed to improve the performance of slotted ALOHA; however, all these schemes have assumed that immediate feedback is available at the transmitters regarding the outcome of their transmissions. This paper introduces ALOHA-dQT, which is the first channel-access protocol based on the use of RL in the context of slotted ALOHA that takes into account the use of explicit acknowledgments from receivers to senders. As such, ALOHA-dQT is the first RL-based approach for channel access that is suitable for wireless networks that do not rely on centralized repeaters or base stations. ALOHA-dQT achieves high utilization by having nodes broadcast short summaries of the channel history as known to them along with their packets. Simulation results show that ALOHA-dQT leads to network utilization above 75%, with fair bandwidth allocation among nodes.