A Fast and Collision Avoidance Distributed TDMA Schedule Based on the Multi-Arms Bandit

Abstract

Abstract In this paper, we propose a novel distributed TDMA protocol based on the Multi-Arms Bandit model for the dynamic UAVs environment. Because of the frequent topology changes of UAVs, we consider a distributed communication protocol, which do not require the maintenance of accurate network topology information. Through the adaptive interaction between nodes, perceiving the surrounding topology environment and learning from historical experience, each node in the network can transmit information in a fast collision-free way. Also, the MAB model is utilized in our protocol, so that the time slot can obtain sufficient multiplexing rate through multiple rounds of node selection. Experiments show that the protocol can achieve better throughput and fast enough convergence speed, even in the case of high network density, and scales well with the size of the network.