TDMA Frame Length Minimization by Deep Learning for Swarm Communication

In this study, we consider a scenario where Unmanned Aerial Vehicles are deployed in an area of interest in order to monitor this area, and the data gathering process following the monitoring is applied towards a central UAV (sink) for analysis. Our objective is to find a minimum-length TDMA schedule by jointly determining the scheduled concurrent links at each slot, along with their used power levels and the established routes toward the sink, under SINR requirements. Since the problem is known to be NP-hard, we aim to provide solutions to real-world situations using efficient heuristic approach. Our research goal is to use deep neural network to approximate the solution of finding the minimum frame length for the TDMA frame. We propose the TDMA frame length minimization by the deep neural network approach (named MTFLet). We have performed extensive experimental evaluations of MTFLet and simulations showing that MTFLet outperforms other state-of-art methods.