Distributed Joint Power, Association and Flight Control for Massive-MIMO Self-Organizing Flying Drones

This article studies distributed algorithms to control self-organizing flying drones with massive MIMO networking capabilities - a network scenario referred to as mDroneNet. We attempt to answer the following fundamental question: what is the optimal way to provide spectrally-efficient wireless access to a multitude of ground nodes with mobile hotspots mounted on drones and endowed with a large number of antennas; when we can control the position of the drone hotspots, the association between the ground users and the drone hotspots, as well as the pilot sequence assignment and transmit power for the ground users? To the best of our knowledge, this is the first time that massive MIMO capabilities are considered in self-organizing flying drone networks. We first derive a mathematical formulation of the problem of joint power, association and movement control in mDroneNet, with the objective of maximizing the aggregate spectral efficiency of the ground users. It is shown that the resulting network control problem is a mixed integer nonlinear nonconvex programming (MINLP) problem. Then, a distributed solution algorithm with polynomial time complexity is designed by solving three closely-coupled subproblems: access association, joint pilot sequence assignment and power control, and drone movement control. As a performance benchmark, a globally-optimal but centralized solution algorithm is also designed based on a combination of the branch and bound framework and convex relaxation techniques. Results indicate that the distributed solution algorithm converges fast (within tens of iterations) and achieves a network spectral efficiency very close to the global optimum obtained by the centralized solution algorithm (over 90% in average)